As of Friday, June 12, 2026, the best public signal is that ICE detention has already fallen materially from its winter peak. Using ICE data, USAFacts reports 70,805 people in ICE detention at the end of December 2025. Vera reports the daily detention population then reached a record high of more than 73,400 in mid-January 2026. But later reporting shows a clear spring drop: CBS, citing ICE data obtained via FOIA, said ICE averaged about 72,000 detained per day in January 2026 versus about 63,000 in March 2026, with early-April data still showing decline; Reuters then reported that some 57,000 immigrants were detained as of early June 2026. So the most relevant short-run trend into August is downward from the January peak, not upward from it. (usafacts.org)

There is still meaningful upside risk because the administration is trying to expand detention capacity. Official DHS FY2026 budget materials say the administration strategy is 1,000,000 removals per year and 100,000 detention beds, and that the FY2026 budget increases detention capacity funding by $501.0 million to support 50,000 beds in the base budget. Reuters also reported in February 2026 that ICE planned a much larger expansion to 92,600 beds by the end of November 2026, including large detention centers and regional processing sites. USAFacts says ICE had 212 active detention centers as of December 26, 2025, and Vera says ICE held people in 456 facilities during February 2026. That means capacity growth is real, but the biggest named build-out appears scheduled mainly for later in 2026, after this question resolves on August 10, 2026. (dhs.gov)

There is also nontrivial downward pressure from legal and operational friction. Reuters found in mid-February 2026 that judges had ruled more than 4,400 times since October that ICE was unlawfully detaining immigrants, which suggests continuing release pressure and implementation friction even if enforcement remains aggressive. Pair that with the spring decline in ICE detention levels and the early-June Reuters estimate near 57,000, and my base case is that the underlying August snapshot is still more likely than not to be at or below 60,000. Because this market has a closed lower bound at 60,000, any such outcome resolves to exactly 60,000. (kelo.com)

I therefore modeled a latent underlying August 2026 currently-detained total as a three-scenario mixture, then applied the market floor rule. My scenarios are: 68% continued decline/plateau centered at 58,150; 24% modest rebound centered at 61,200; and 8% sharper rebound centered at 63,900, with standard deviations of 1,600, 1,200, and 1,000 respectively. This is an inference from the cited trend data: a large decline from January into June, real but mostly later-dated capacity expansion, and persistent legal friction. Under that model, the probability that the true August value is 60,000 or lower is about 63% to 64%, so the modal market resolution is the floor value of 60,000. The model leaves a meaningful minority chance of a rebound into the low 60,000s, a smaller chance of the mid-60,000s, and only about a 1% tail above 65,000. (investing.com)

I treated this as a two-layer forecasting problem: first, whether ICE will actually publish a newer FY2026 detention workbook by Monday, August 10, 2026; and second, conditional on publication, what the workbook’s ‘ICE Currently Detained’ total will be. The publication-risk piece matters because the resolution does not ask for the real-world detention total on August 10; it asks for the value in the most recent FY2026 workbook published on or before that date. The Deportation Data Project (DDP) says ICE’s detention management spreadsheet is normally released about twice a month, and that DDP tracks both the spreadsheet’s publication date (‘file_date’) and the underlying data pull date (‘pull_date’). TRAC’s April 10, 2026 detention update, which is derived from ICE’s public detention management releases, reported 60,311 people in detention as of April 4, 2026; that followed 68,289 on February 7, 2026 and the 70,766 peak on January 25, 2026. Separately, a CRS product said ICE data indicated 66,978 detained as of April 9, 2026. Because this market resolves specifically to the public workbook, I weight the workbook-derived 60,311 far more heavily than the broader April 9 ICE figure; DDP’s explicit distinction between publication date and pull date makes it plausible that an early-April workbook published around April 9 would still contain an April 4 count. (deportationdata.org)

The biggest unusual feature is that public reporting appears to have become erratic. The Guardian noted that after the late-2025 government shutdown, ICE started releasing detention data inconsistently. By spring 2026, outside reporting was still citing only early-April public detention figures, and other coverage said ICE’s once-regular public reports had gone weeks without updates. That makes a genuine ‘no newer workbook by August 10’ scenario material rather than remote. I therefore assign a substantial probability that the governing workbook on August 10, 2026 is still the early-April public release, which would make the resolving value essentially 60,311. (theguardian.com)

Conditional on ICE resuming or continuing publication, the directional evidence is mixed but leans to only a modest rebound. TRAC reported that detention fell from 68,289 on February 7, 2026 to 60,311 on April 4, 2026, while monthly book-ins fell from 39,694 in January 2026 to 32,531 in March 2026, an 18% drop. CBS similarly described a roughly 12% decline from the January peak to late March / early April, driven largely by fewer detentions of people without criminal records. That argues against simply extrapolating the January peak into August. (tracreports.org)

But there is also clear upside risk. CRS said ICE had roughly 70,000 detention beds available and was considering expansion to over 90,000 in FY2026. Reuters reported in March 2026 that ICE planned to spend $38.3 billion on detention centers as it prepared for an expected surge of arrests later in 2026 after hiring 12,000 more agents. GAO said ICE’s average daily population had risen from 39,314 on January 20, 2025 to 67,204 on April 1, 2026, and that authorized detention facilities had increased from 134 to 239 over the same period. Those are not point-in-time workbook totals, but they do show that capacity and institutional momentum were still expanding, which creates a real upper tail if publication resumes and summer enforcement intensifies. (everycrsreport.com)

My final distribution is therefore a mixture of three scenarios. Scenario 1 (37%): no newer public FY2026 workbook appears by August 10, 2026, so the resolving number is a point mass at 60,311. Scenario 2 (43%): ICE does publish newer workbook(s), but the detained total remains in the low 60,000s; I model that with a normal distribution centered at 62,050 with an 850 standard deviation. Scenario 3 (20%): renewed ramp-up from detention expansion and enforcement growth pushes the published total materially higher by late July / early August; I model that with a normal distribution centered at 64,550 with a 1,500 standard deviation. This yields a forecast mean of about 61.9k, a median in the low 61k range, and about a 92% chance that the resolving value is at or below 65,000. The large spike near 60,311 reflects publication risk; the right tail above 65,000 reflects capacity-expansion risk. (tracreports.org)

I could not directly fetch the ICE detention-management page or the XLSX from this tool because ICE returned 403 errors, so I relied on multiple secondary sources that quote or scrape the official ICE detention workbook values directly. That matters here because the question resolves off a very specific spreadsheet cell. ()

The best official-series anchor I found is TRAC’s preserved table of ICE detention snapshots. It shows 68,990 detained on January 7, 2026; 70,766 on January 25, 2026; 68,289 on February 7, 2026; and 60,311 on April 4, 2026. The same preserved series also shows that the year-ago August 10, 2025 value was 59,380. Separately, Brookings’ May 2026 appendix explicitly says the April 9, 2026 FY2026 workbook was the latest release they had, and it uses the “Detention FY26” sheet from that release. (tracreports.org)

Reporting cadence is a major uncertainty. The Guardian notes ICE historically published detention statistics roughly every two weeks, but after the late-2025 shutdown the releases became inconsistent. Brookings still treated the April 9, 2026 workbook as the latest available in May, and a recent TRAC crawl still surfaced April 4, 2026 as the latest public “currently detained” date. That creates a real possibility that the August 10, 2026 resolving workbook is still a stale public release rather than a fresh near-real-time one. (theguardian.com)

On the underlying population, the evidence points to a sharp early-2026 surge followed by a meaningful pullback into a lower but still historically elevated regime. CBS/Reuters reported that the number in ICE custody fell 12% from the January record high to the end of March, driven largely by fewer detentions of people without criminal records, and that ICE arrests averaged about 1,040 per day between mid-February and early March. Reuters/USA TODAY also reported that court-ordered releases rose from 46 at the start of FY2026 to 1,005 by March, reinforcing the idea that litigation and judicial pushback helped pull the population down. AP’s late-May reporting still described the detained population as having spiked “to 60,000” during Trump’s second term, which is broadly consistent with the April 4 official snapshot remaining a decent description of the current public-data regime. (cbsnews.com)

A rough flow-stock check also argues against assuming either a collapse or an immediate return to the January peak. Brookings calculates average monthly ICE-arrest book-ins of 31,462 for January-March 2026. Combining that with the April 4 detained stock of 60,311 implies about 57.5 detention-days per detainee if those ICE-arrest book-ins were the only inflow; because CBP transfers also feed detention, the true implied stay is somewhat lower. That is a plausible equilibrium for a population near 60k, so a low-60k outcome by August does not require any dramatic policy shift. Meanwhile, the year-ago August 10 level was 59,380, which also supports using the low 60s as the base-rate neighborhood rather than assuming a mechanically persistent 70k-plus regime. (brookings.edu)

The main upside risk is capacity expansion. NPR/GBH reported that private prison firms had already announced more than 6,000 additional beds across multiple states and described ICE detention as exceeding capacity, while also noting that nearly half of current detainees have no criminal record. Reuters reported Delaney Hall alone is a 1,000-bed facility and had already held more than 850 detainees by late May. Most importantly, ICE’s February 12, 2026 Detention Reengineering Initiative document says the agency plans to reach 92,600 beds, hire 12,000 new law-enforcement officers, and activate all DRI facilities by November 30, 2026. That is too late for full August realization, but it makes a modest summer rebound quite plausible. (wgbh.org)

Putting that together, I modeled four practical scenarios. First, about a 25% chance that the public workbook is still effectively frozen and the resolving value is exactly 60,311 from the April 9 release. Second, about a 5% chance that a newer release exists but the resolved value is at or below the contract floor and therefore clamps to 60,000. Third, a 40% “stable low-60s” regime centered near 62,050, reflecting continued legal pressure plus roughly equilibrium inflow/outflow. Fourth and fifth, 20% and 10% rebound regimes centered near 63,600 and 66,750, reflecting partial realization of the added bed capacity and continued enforcement. My resulting mean is about 62,293, and the distribution has a noticeable discrete spike at 60,311 because the stale-public-data possibility is genuinely material.

So my bottom line is: the most likely broad outcome range is about 61,500 to 64,000, but I assign substantial probability to an unchanged public workbook that resolves at 60,311. I also leave a meaningful right tail above 65,000 because ICE is actively building toward much larger detention capacity by late 2026, even if I do not think that full expansion is likely to be visible in the public FY2026 workbook by August 10. (tracreports.org)

I anchor on the FY2026 detention-statistics series itself. Independent dashboards that explicitly mirror ICE’s detention-management spreadsheets describe the source series as biweekly or roughly biweekly updates based on ICE releases. The visible FY2026 path shows a January spike followed by a sharp spring pullback: TRAC reported 70,766 currently detained on January 25, 2026, and 68,289 on February 7, 2026; CBS, using ICE-released data, reported average daily detention of about 72,000 in January versus about 63,000 in March; TRAC quick facts then showed 60,311 in detention on April 4, 2026; and NPR/KERA described the latest ICE data in late April as nearly 60,000. (detentionstats.com)

I take that decline seriously rather than treating January’s 70,000-plus level as the default. CBS reported that from January to March, detainees without criminal records fell 21%, collateral arrests dropped from more than one-quarter of arrests to fewer than one-fifth, and the first week of April suggested the detention-population decline was continuing. But CBS also said it was too early to know whether that less aggressive pattern would last. Late-May and early-June AP coverage still described the detained population as around 60,000 or more than 60,000 as of early April, which makes a spring plateau in the low 60,000s more plausible than an immediate snapback. (cbsnews.com)

The main reason I keep a meaningful upper tail is capacity and policy. DetentionStats, citing ICE and DHS materials, says the July 2025 OBBBA allocated $45 billion over four years for ICE detention expansion. GAO wrote that ICE’s average daily population rose from 39,314 on January 20, 2025 to 67,204 on April 1, 2026, while authorized detention facilities rose from 134 to 239 over the same span. Vera reported that detention exceeded 73,400 on a single day in mid-January 2026 and that 152 new detention facilities opened during the second Trump administration, including 13 in 2026. So the administration still has the money, legal posture, and physical capacity to push numbers back up if enforcement intensity re-accelerates. (detentionstats.com)

My quantitative anchor is a stock-flow check. TRAC reports 32,531 March 2026 book-ins, which is about 1,049 per day, and 60,311 currently detained on April 4. Dividing stock by daily inflow gives about 57.5 detention-days of stock per daily book-in; that is a system naturally sitting in the low 60,000s if March-April conditions persist. Reaching 65,000 by early August would not require a dramatic regime change—only somewhat higher inflow, somewhat longer effective detention, or both—so I do not set that tail near zero. But the observed spring data still point more toward stabilization than toward a return to January’s record highs. (tracreports.org)

Putting those pieces together, I center the forecast a bit above the April level, with most probability mass between roughly 60,000 and the mid-64,000s, meaningful clipped mass at the 60,000 floor because of the resolution rule, and a smaller but real right tail above 65,000 because capacity expansion is genuine. The code implements that view as a three-component normal mixture on a latent detention total; the first CDF element is fixed at 0.0 because the contract’s lower bound makes outcomes below 60,000 resolve to the floor rather than below it. (tracreports.org)

I anchored primarily on official or near-official measures of the detention system’s current level. A CRS report published on April 24, 2026 said ICE data showed 66,978 people detained as of April 9, 2026, while a GAO report published on June 9, 2026 said ICE’s average daily detained population had risen from 39,314 on January 20, 2025 to 67,204 on April 1, 2026, and that authorized ICE detention facilities had expanded from 134 to 239 over the same span. DHS’s FY2026 budget justification also says the administration’s strategy is aiming at much larger detention capacity, with the base budget supporting 50,000 beds and reconciliation funding intended to enable 100,000 detention beds. (everycrsreport.com)

At the same time, the short-run trend was not one-way upward. TRAC’s case-level detention data show 70,766 detained on January 25, 2026, 68,289 on February 7, and 60,311 on April 4; TRAC also reported March 2026 book-ins of 32,531 versus 39,694 in January. CBS and Anadolu similarly reported that ICE’s March average daily detention population was about 63,000, down from about 72,000 in January. (tracreports.org)

There is meaningful measurement uncertainty here: TRAC’s April 4 count of 60,311 is far below the 66,978 April 9 figure cited by CRS from ICE data. Because this market resolves to the official ICE workbook, I overweight the official and agency-linked numbers for level and use TRAC mostly as evidence about volatility and downside risk rather than as the central level. I also note that the CRS report said ICE had roughly 70,000 detention beds and was considering expansion to more than 90,000 in FY2026, which limits how bearish I want to be over only about eight weeks to resolution. (tracreports.org)

I therefore use a three-scenario mixture. I assign 48% to a plateau scenario centered near 67,500, basically holding around the early-April official level; 40% to a renewed expansion scenario centered near 70,800, reflecting continued facility and capacity growth plus the administration’s explicit push for much larger detention space; and 12% to a downside scenario centered near 62,800, reflecting the spring slowdown in book-ins and the possibility that releases, removals, litigation, or operational bottlenecks temporarily pull the official count down. The weighted mean of those scenarios is 68,256. (files.gao.gov)

That setup implies the official workbook figure is still more likely than not to be above 65,000 on the most recent release published on or before August 10, 2026. In this model, the probability of 65,000 or lower is about 16.2%, with most of the mass in the high-60,000s and low-70,000s. Because the contract has a closed lower bound at 60,000, the first CDF entry is fixed at 0.0 and any true sub-60,000 downside is absorbed into mass at the lower bound in subsequent entries. (files.gao.gov)

What will the “ICE Currently Detained” total be in the most recent FY2026 ICE Detention Statistics spreadsheet published on or before August 10, 2026?

• My most likely prediction for the ICE Currently Detained total is 60,000.00, with a 50% chance of falling between 60,000.00 and 61,425.00.
• Current reports as of June 9, 2026, indicate the detention population has fallen to approximately 57,000 individuals, down from a January peak of over 70,000.
• The resolution criteria specifically floor any value at or below 60,000 to the 60,000.00 mark, making this the most probable outcome given recent trends.
• While the $38 billion Secure America Act signed in June 2026 provides massive resources for expansion, operational lags and reporting latency mean these effects are unlikely to be fully reflected in reports published by the August 10 deadline.
• Major private detention contractors (GEO Group, CoreCivic) expect sequential growth in late 2026, but emphasize that activating idle facilities requires several months for staffing and training.
• Border encounter data shows a cooling trend, which reduces the immediate pressure on detention bed space compared to earlier in the fiscal year.

Key figures

Historical context

• Winter Peak: In January 2026, ICE detention reached a record high of approximately 70,805 to 72,000 people following the administration’s initial mass deportation push.
• Funding Shutdown: A 75-day partial government shutdown (mid-February to May 2026) led to a significant drop in detention levels as DHS recalibrated resources.
• Biden Baseline: During the Biden administration (2024), detention levels typically fluctuated between 25,000 and 38,000, showing that 2026 levels remain historically elevated despite recent declines.
• Reporting Patterns: ICE historically updates its detention management XLSX workbooks bi-weekly, though this was interrupted by the 2026 shutdown and funding impasse.

Tailwinds

• Secure America Act: The June 10, 2026, legislation provides $38 billion in immediate funding specifically for ICE detention and removal.
• Contractor Guidance: CoreCivic CFO assumes sequential population growth in Q3 and Q4 2026, aiming for a return to the 70,000 level.
• Mandatory Detention: The Laken Riley Act and other 2025/2026 legal shifts require mandatory detention for broader categories of non-citizens.
• Facility Activations: CoreCivic activated five facilities in early 2026 to meet demand, indicating infrastructure is ready if staffing is secured.

Headwinds

• Reporting Latency: ICE spreadsheets have averaged a 27-day update delay in 2026, meaning the August 10 report will likely capture mid-July data before funding impacts are fully realized.
• Operational Lag: Facility operators like CoreCivic indicate a 4-6 month lead time for activating idle beds, limiting the speed of capacity expansion.
• Lower Inflows: Nationwide encounters have flattened at approximately 31,000/month, significantly lower than the levels that supported the 70,000+ peaks in early 2026.
• Policy Recalibration: Recent DHS leadership shifts have signaled that the 100,000-bed capacity target is no longer considered an immediate necessity.

Detailed reasoning

My forecast is primarily driven by the interaction between current downward detention trends and the specific resolution mechanics of this question. As of June 9, 2026, credible reports from Reuters/Yahoo indicate that ICE is detaining approximately 57,000 people. This represents a significant decline from the January 2026 peak of 70,805. Because the resolution criteria state that any value of 60,000 or lower must resolve at the closed lower bound of 60,000, there is a high statistical probability that the final outcome will cluster at this exact floor.

The decline observed in the first half of 2026 was largely attributed to a 75-day partial government shutdown and shifting DHS leadership priorities. While the “Secure America Act” signed on June 10, 2026, provides a massive $38 billion infusion for ICE enforcement, the operational “ramp-up” period for detention is historically slow. Private contractors like CoreCivic and The GEO Group have noted that activating idle facilities typically requires 4 to 6 months for staffing, training, and utility preparations. Given that the governing spreadsheet must be published by August 10, 2026, and that these reports carry an average 27-day latency, the data will likely reflect the detention environment of early-to-mid July. This leaves only a 4-to-5-week window for the newly authorized funding to manifest in higher detention totals.

Furthermore, Macrobond data shows that nationwide border encounters have stabilized at a lower baseline (approximately 31,000 per month), which reduces the immediate “inflow” pressure compared to the surges seen in late 2024. Although CoreCivic management expressed optimism for a return to 70,000 detainees in the second half of 2026, this sequential growth is more likely to be reflected in late FY2026 or early FY2027 reports rather than the mid-summer snapshot required for this forecast. I have weighted the possibility of a modest recovery above 60,000 as a secondary scenario, but the most probable outcome remains the floor-adjusted value of 60,000.00.

Key uncertainties

• Publication Regularity: If ICE fails to publish a new workbook before August 10, the April 4 figure (60,311) might be the governing value.
• Speed of Funding Deployment: If ICE bypasses standard contracting for “warehousing” or temporary facilities, the population could spike faster than historical lead times suggest.
• Interior Enforcement Intensity: A sudden “surge” of interior arrests (flooding the zone) could rapidly consume the ~13,000 idle beds currently available in the private sector.
• Court-Ordered Releases: A rise in judicial releases (which grew to 1,005 in March 2026) could offset enforcement efforts.

Conclusion

• My most likely prediction for the ICE Currently Detained total is 60,000.00, with a 50% chance of falling between 60,000.00 and 61,425.00.
• The forecast is heavily influenced by the resolution criteria, which mandate that any official figure of 60,000 or lower resolves at exactly 60,000.
• While aggressive funding ($38 billion) and capacity targets (up to 100,000 beds) suggest a long-term increase, the short timeframe and typical reporting latency of 27 days make a substantial recovery beyond the 60,000 threshold by early August unlikely.

Question: What will ICE’s “currently detained” total be in the ICE FY2026 workbook that governs resolution on Aug. 10, 2026? The teams agree the strongest anchor is the Apr. 9, 2026 FY2026 workbook value of 60,311 (mirrored by TRAC and other public sources), and that the question’s 60,000 lower-bound rule makes small downward moves materially important. Public data show a sharp fall from a winter peak (~70k) to ~60.3k by early April, plus unusually low border inflows and reported slowdowns in some interior arrests—factors that make a near-floor or sub-60k resolution plausible if a newer workbook or verified June data (an uncorroborated 59,600 figure) confirm continued decline. Countervailing forces—large FY2025–FY2029 detention funding, announced plans to expand capacity toward ~92.6k beds, and the system’s historical ability to shift by thousands quickly—create a meaningful right tail for a summer rebound, but most reported capacity increases appear targeted after the Aug. 10 cutoff and face implementation/friction risks. Numerical modeling echoed these qualitative conclusions, centering mass slightly above 60k (median ~61.7k in one model) with heavy mass at the 60k floor and modest probability above 65k; teams differ mainly on how much weight to give an unverified June data point and the practical speed of capacity increases. Key monitoring priorities before finalizing: check ICE’s workbook releases (or mirrors like TRAC) after Apr. 9, verify any June values, and track interior arrests, book-ins, and actual facility openings/occupancy.

Forecast and Reasoning

Forecast: Probability distribution:

• 1.00% chance of value below 60000.0
• 5.00% chance of value below 60000.1
• 10.00% chance of value below 60000.2
• 20.00% chance of value below 60000.6
• 30.00% chance of value below 60010.0
• 40.00% chance of value below 60650.0
• 50.00% chance of value below 61900.0
• 60.00% chance of value below 63100.0
• 70.00% chance of value below 64300.0
• 80.00% chance of value below 65110.0
• 90.00% chance of value below 65820.0
• 95.00% chance of value below 66070.0
• 99.00% chance of value below 66235.0

Forecast Reasoning

All three analysts are using the same key factual anchors, and those anchors look directionally consistent: ICE detention was very high in late January/early February 2026, then fell sharply by spring, with the strongest shared point estimate being about 60,311 currently detained on April 4, 2026, after being roughly 63,482 on March 12 and about 68,000 to 71,000 in late January/early February. The core forecasting problem is therefore not the recent level but the direction from April to the late-July/early-August snapshot likely to govern resolution. Forecasters 2 and 3 are strongest on the mechanics of the lower bound: because any true value at or below 60,000 resolves to exactly 60,000, and because April was already only about 311 above that threshold, there is real point-mass at the floor. Forecaster 1 is strongest on the structural upside case: policy direction, funding, capacity expansion, and continued enforcement pressure make a simple linear continuation of the spring decline too pessimistic. I think Forecaster 1 likely overweights how quickly new detention capacity translates into occupied beds by early August, while Forecasters 2 and 3 likely overweight the January-to-April decline as if it will persist smoothly for another four months despite evidence of policy pressure in the opposite direction. My synthesis is: meaningful probability on the 60,000 floor, a median still only modestly above it, and a material but not dominant open-upper tail above 65,000. I therefore center the distribution in the low 62,000s, keep roughly 20% to 30% of mass effectively at or near the floor, and retain roughly 20% to 25% mass above 65,000 for a summer rebound scenario. Some analyst percentile tables appear clipped at the upper bound despite written reasoning that clearly implies above-65,000 outcomes; I follow the underlying reasoning rather than the clipped formatting.

A naive forecast would likely do one of two things: anchor almost entirely on the latest reported 60,311 figure and ignore rebound risk, or anchor on the earlier 68,000 to 71,000 peak and ignore both the spring drop and the special 60,000 floor rule. My forecast improves on that by combining the recent hard anchor, the bound-induced floor mass, and the credible policy/capacity upside tail. Confidence in the improvement is moderate, not high, because the official public data cited by the analysts appear somewhat stale relative to the August target window.

Forecast rationale (numeric):

— Iteration 1 — Across the forecasts, the core expectation is that the ICE Currently Detained total will land in the low 60,000s, likely around 62,000–63,000.

Main reasoning patterns

• Resolution floor matters a lot: All forecasts emphasize that values below 60,000 effectively collapse to the same reported outcome because of the spreadsheet’s resolution rules. That creates a noticeable probability mass at the floor.
• Near-term change should be limited: Because the forecast window is only about two months, the total is expected to move only modestly unless there is a major policy shift, surge in enforcement, or unexpected border-related pressure.
• Upward drift is more likely than decline: The central expectation is for modest growth driven by ongoing enforcement intensity and possible detention-capacity expansion.
• Upper-tail uncertainty remains meaningful: While the most likely outcome is in the low 60,000s, there is explicit allowance for a scenario where facility expansion, surge operations, or broader enforcement changes push the total into the mid-to-high 60,000s or even low 70,000s.

Areas of consensus

• The figure is unlikely to be far below 60,000 due to the floor.
• A central estimate around 62.3k–63.0k is broadly consistent across the forecasts.
• The distribution should be right-skewed, with more upside uncertainty than downside uncertainty.

Main differences in emphasis

• Some forecasts place more probability mass right at the 60,000 floor, reflecting the mechanical effect of the reporting resolution.
• Others stress a broader right tail, reflecting the possibility of rapid growth from expansion or policy-driven surges.

Bottom line

The collective view is that the most recent FY2026 ICE detention spreadsheet is expected to show a detained total in the low 60,000s, with the main uncertainty coming from whether enforcement and capacity growth produce a modest increase or a larger-than-expected right-tail jump.

— Iteration 2 — Across the forecasts, the reasoning is highly consistent:

• Primary anchor: All models start from the most recent confirmed figure of 60,311 (April 2026) and treat it as the baseline for projecting the August 2026 spreadsheet.
• Expected direction: The central expectation is flat to slightly higher by the August 10 publication date, with point estimates clustered in the low 61,000s.
• Main upward drivers: Continued emphasis on detention/enforcement policy and possible capacity expansion could push the total modestly upward.
• Main limiting factors: Bed capacity, operational constraints, funding limits, and end-of-fiscal-year pressures are seen as restraining any large increase.

Key shared structural assumption:

• The forecasts all respect the hard lower bound of 60,000, so a meaningful portion of probability is concentrated there as a floor-like outcome.
• The upper tail is kept well above the center estimate to allow for unexpected capacity activation, policy shifts, or reporting changes, with scenarios extending into the 70,000+ range.

Overall synthesis:
The collective view is that the most recent FY2026 ICE detention total will probably remain near the April 2026 level, with only a modest rise or plateau expected rather than a dramatic jump.

— Iteration 3 — Across the forecasts, the main anchor is the most recent reported detention level of about 60,311 in early April 2026, with an additional reference point of the December 2025 peak near 68,442. The shared interpretation is that the current level is already near the lower bound of the expected range, so the forecast is strongly shaped by whether ICE detention stays roughly flat, continues drifting down, or rebounds as capacity and enforcement intensity change.

Common reasoning patterns

• Recent data as the baseline: All rationales start from the April 2026 figure around 60.3k as the best near-term anchor.
• Low 60,000s as the central expectation: The models generally expect the count to remain in the low 60,000s by early August 2026, with central estimates around 62.5k–63k.
• Bounded downside: Because of the resolution floor at 60,000, a meaningful share of probability accumulates at or near that floor if the count continues to decline.
• Upside risk from capacity and enforcement: A rebound is possible if new detention capacity comes online, book-ins remain high, or deportation/release throughput slows, potentially pushing totals back toward the mid-60,000s or even higher.

Areas of consensus

• The forecasts broadly agree the most likely outcome is between 60,000 and 65,000.
• They also agree that the lower tail is compressed by the 60,000 floor.
• There is shared recognition that recent trends have eased from the December peak, but not enough to rule out a modest rebound.

Main differences

• One forecast is a bit more conservative, centering closer to 63,000 with most mass at or below 65,000.
• Another places slightly more weight on stability near 60,000 but still allows for a substantial rebound scenario.
• The degree of upper-tail risk varies: some see only modest chances above 65,000, while others leave room for a return toward the prior peak or even the 70,000+ range if conditions tighten.

Overall synthesis

The collective view is that the ICE detained population is most likely to remain anchored in the low 60,000s by the August 2026 publication, with a decent chance of hovering near the 60,000 floor and a smaller but meaningful possibility of rising if detention expansion outpaces removals and releases.

Summary of Rationale Reasoning

The rationales synthesize a core tension between a recent, sharp decline in ICE detention populations and massive new legislative funding intended to expand detention capacity. Forecasters focus on several key dynamics to determine the late-July snapshot expected in the early August publication.

• Historical Baseline and Downward Pressures: The rationales universally acknowledge a steep drop in the detained population from an early 2026 peak down to a baseline just above 60,000 in April. This contraction was driven by highly efficient expedited removals outpacing new intakes, historically low border encounters, funding standoffs, and the closure of major detention facilities.
• The Resolution Floor: Because the April baseline hovered so close to the question’s hard lower bound, forecasters agree that if the rapid pace of deportations continues to outstrip intake, the population will naturally drain below 60,000. In this scenario, the strict resolution criteria act as a hard floor, anchoring expectations significantly.
• Upward Pressures and New Funding: Counteracting the downward trend is the passage of the “Secure America Act” in June 2026, which injected billions of dollars into ICE with a mandate to vastly expand bed capacity. Forecasters anticipate this will trigger aggressive new interior enforcement sweeps. Additionally, recent court rulings mandating bond hearings are expected to increase the average length of stay for detainees, causing the system to back up and the population to grow.
• Operational Frictions and Timing: Despite the massive influx of funding, forecasters agree that explosive growth by late July is heavily constrained by logistical realities. Retrofitting massive warehouses into detention centers, overcoming severe medical staffing shortages, and fighting local litigation will slow the physical expansion of beds.
• Data Transparency: Finally, some rationales note a recent pause in ICE’s bi-weekly data reporting. If this blackout persists through the summer, the question may resolve based on the static, older April figures.

In summary, forecasters weigh the likelihood of an immediate administrative scale-up fueled by new funding against the ongoing friction of logistical bottlenecks, rapid deportation pipelines, and the strict technical floor of the question’s resolution criteria.

SUMMARY

Question: What will the “ICE Currently Detained” total be in the most recent FY2026 ICE Detention Statistics spreadsheet published on or before August 10, 2026? Final Prediction: Probability distribution:

• 10.00% chance of value below 60000.1
• 20.00% chance of value below 60000.2
• 40.00% chance of value below 60310.975
• 60.00% chance of value below 61155.7
• 80.00% chance of value below 65100.0
• 90.00% chance of value below 71000.0

Total Cost: extra_metadata_in_explanation is disabled Time Spent: extra_metadata_in_explanation is disabled LLMs: extra_metadata_in_explanation is disabled Bot Name: extra_metadata_in_explanation is disabled

Report 1 Summary

Forecasts

Forecaster 1: Probability distribution:

• 10.00% chance of value below 60000.1
• 20.00% chance of value below 60000.2
• 40.00% chance of value below 60311.25
• 60.00% chance of value below 62000.0
• 80.00% chance of value below 66000.0
• 90.00% chance of value below 73000.0

Forecaster 2: Probability distribution:

• 10.00% chance of value below 60000.1
• 20.00% chance of value below 60000.2
• 40.00% chance of value below 60310.7
• 60.00% chance of value below 60311.4
• 80.00% chance of value below 64200.0
• 90.00% chance of value below 69000.0

Research Summary

The research summarizes recent ICE detention dynamics through early 2026: after a sharp post-2024 enforcement ramp that pushed reported “currently detained” counts from ~40,000 (end of 2024) to an early-2026 peak in the ~70,000–73,400 range, the most recent verified ICE update (April 4, 2026) showed a decline to 60,311. That decline—roughly 8,000–10,000 people over about two months—was attributed to an 18% drop in monthly bookings (to ~32,531 in March) coupled with accelerated deportations (averaging ~1,286 per day). The brief also notes ICE paused public updates for 56 days amid a funding standoff, creating a precedent for reporting gaps.

On the supply side, the briefing highlights a large 2025 budget increase ($45 billion) and an aggressive capacity push toward 100,000 daily detainees via rapid procurement and retrofitting of commercial warehouses, including a cited 8,000-bed facility in Social Circle, GA targeted to open by August 2026. Prediction-market activity specifically for the August-2026 ICE statistic appears limited or illiquid, and the analyst flags the question’s 60,000 lower-bound resolution rule as critical given the April 4 value (60,311). The research frames two key contingencies for the August publication: (1) timely activation of new mega-facilities could cause a spike well above 60,000, whereas (2) continued deportation outflows and any facility delays could allow the figure to fall below the 60,000 threshold.

Sources used:

• ICE Detention Statistics (U.S. Immigration and Customs Enforcement): https://www.ice.gov/detention-management/detention-statistics
• Polymarket: https://polymarket.com
• Kalshi: https://kalshi.com
• (Additional news reports and brief references cited in the original research; specific article links were not provided in the brief.)

RESEARCH

Report 1 Research

Hello! I have gathered a comprehensive briefing on the most relevant news, historical data, and prediction market context to assist your superforecaster with this Metaculus question on the “ICE Currently Detained” total for August 2026.

Here is your detailed rundown:

1. Recent News & Current Trends (Immigration and Customs Enforcement)

The current landscape of ICE detention in 2026 is highly dynamic, defined by massive budget increases juxtaposed against a recent short-term decline in the detained population.

• Record Highs followed by a Recent Decline: After taking office in 2025, the second Trump administration dramatically expanded enforcement. The “currently detained” population spiked from around 40,000 at the end of 2024 to an all-time record of roughly 70,766 (with some estimates up to 73,400 depending on counting metrics) between late January and early February 2026 [4][5][7]. However, the most recent verified data from April 4, 2026, shows the population declined to 60,311 [1][7].
• Drivers of the Decline: This 12% drop (roughly 8,000 to 10,000 people over two months) was driven by an 18% decrease in new monthly bookings (dropping to 32,531 in March) combined with an acceleration in deportations, which averaged 1,286 per day [1][7][22].
• Massive Capacity Expansions Expected by August: Despite the recent drop, ICE received a historic $45 billion budget allocation in 2025 to reach a target capacity of 100,000 daily detainees [17][18]. To achieve this, ICE is bypassing standard procurement to rapidly purchase and retrofit 34 commercial warehouses. For example, a massive 8,000-bed facility in Social Circle, Georgia, is explicitly targeted to open by August 2026 [17].
• Reporting Irregularities: While ICE historically updates its statistics every two weeks, it recently paused publishing for 56 days (from February 7 to April 4, 2026) due to a congressional funding standoff [7]. This establishes a recent precedent for delayed reporting.

2. Relevant Prediction Markets

Note on Search Noise: Tracking prediction markets for this specific question requires careful disambiguation. The acronym “ICE” also stands for the “Intercontinental Exchange” (the parent company of the NYSE), which recently invested $2 billion into the prediction platform Polymarket in late 2025 to distribute its data [10][11].

• Immigration-Specific Markets: I did find precedent for this exact topic on Polymarket. A market titled “How many detained in next ICE update?” traded in January 2025. However, the volume was relatively low (totaling roughly $16,209) and the market was highly short-term, resolving within a day of opening based on ICE’s daily updates at the time [20].
• Current Liquidity: Currently, there appear to be no liquid, long-term prediction markets specifically forecasting the August 2026 ICE detention statistics on major platforms like Polymarket or Kalshi.

3. Base Rates & Reference Classes

To provide your forecaster with an “outside-view” anchor, here are the quantitative benchmarks and historical base rates:

• The 60,000 Lower Bound Constraint: The most critical factor for this forecast is the resolution criteria’s closed lower bound of 60,000. The most recent reliable data point (April 4, 2026) puts the number at 60,311 [1]. Because the number is resting precariously close to the threshold, the probability mass for a resolution exactly at the 60,000 lower bound should be heavily considered.
• Historical Peaks (Prior to 2025): During the first Trump administration, the detained population peaked at roughly 55,000 in August 2019 [15]. During the Biden administration, it hit a low of under 14,000 in early 2021 before climbing back to roughly 40,000 by late 2024 [5][15].
• Current Volatility & Rate of Change: ICE has proven it can ramp up detention aggressively—adding roughly 30,000 detainees to its daily count over the course of 2025 [5]. However, the February-to-April 2026 window showed a contraction rate of roughly 4,000 detainees per month [22].
• The “August” Seasonality / Timeline: The question targets the publication on or before August 10, 2026. This timeline aligns with the scheduled opening of new “mega-facilities” (like the Georgia warehouse) [17]. If these massive detention centers activate on schedule in late summer, it could cause a sudden, sharp spike in the detained population. Conversely, if facility openings are delayed by construction or litigation—and deportation rates continue to outpace bookings—the population could easily slip below the 60,000 threshold.

I hope this rundown gives your superforecaster a strong foundational view for their modeling! Let me know if you need any additional specific data points.

FORECASTS

R1: Forecaster 1 Reasoning

I agree with the critic’s main calibration point: my previous distribution put too much mass on a simple “trend continues below 60,000” story and too little on the publication process. The important correction is to model three separate things: (1) the true detained-population stock, (2) whether ICE publishes a newer qualifying workbook by August 10, and (3) whether the relevant metric/table maps cleanly to the public snippets and ADP figures.

I still think the center of mass belongs near the lower bound / April value, not in a 70k+ central case. The last clean currently-detained figure remains 60,311 as of April 4, 2026, and March book-ins were 32,531, after a sharp fall from the early-2026 peak. TRAC also records the April 4 total directly, which is the closest public analogue to the resolving table. (tracreports.org) WOLA’s summary strengthens the “process risk” point: ICE had just published for the first time in nearly two months, despite a nominal every-two-week public reporting obligation, and March interior arrests had dropped to the lowest level since August 2025. (wola.org)

The critic’s GAO point is useful, but I do not treat the GAO ADP number as a direct substitute for the resolving “currently detained” total. ADP and point-in-time current detention can diverge substantially, especially after a January/February spike followed by decline. Still, GAO’s report is a warning against overconfidence: ICE detention capacity is expanding, but also operationally messy, with Camp East Montana opened rapidly, underfilled at times, and suffering serious standards/oversight problems. (gao.gov)

I am increasing the high tail, but less than the critic. Axios reported plans for 92,600 capacity across new facilities, including large sites of 7,000–10,000 beds, so a rebound above 65k does not require a miracle. (axios.com) But capacity should not be treated as instantaneous occupancy: Social Circle’s up-to-10,000-person site faced a water-service cutoff and infrastructure objections, and even mega-sites can remain underfilled or delayed. (gpb.org)

My revised rough mixture:

• ~28%: latest qualifying workbook shows ≤60,000, resolving at the closed lower bound.
• ~22%: no newer qualifying workbook / stale April 4 workbook, resolving near 60,311.
• ~25%: newer update but still modestly above 60k and below 65k.
• ~25%: rebound above 65k, including a small but real 85k+ tail.

This moves meaningfully toward the critic: lower-bound mass down, stale-publication mass up, high-tail mass up. I still keep the median around the April value because both the actual stock-flow signal and publication risk concentrate probability near 60–61k.

Updated prediction:

Percentile 0.1: 60000.001
Percentile 1: 60000.01
Percentile 5: 60000.05
Percentile 10: 60000.10
Percentile 20: 60000.20
Percentile 30: 60311.05
Percentile 40: 60311.25
Percentile 50: 60311.45
Percentile 60: 62000
Percentile 70: 63800
Percentile 80: 66000
Percentile 90: 73000
Percentile 95: 82000
Percentile 99: 101000
Percentile 99.9: 122000

R1: Forecaster 2 Reasoning

I agree with the critic’s main update: publication risk should be weighted more heavily than I did before. As of June 12, the public tracking I can verify still points to 60,311 detained as of April 4, 2026, with 32,531 March book-ins, so this is not merely a hypothetical stale-data tail; it is already an active two-month publication gap. (tracreports.org) The Guardian’s methodology note also supports the idea that ICE’s formerly biweekly cadence became inconsistent after the shutdown, so “latest published workbook” is materially different from “true August population.” (theguardian.com)

I therefore reduce my mass at the 60,000 floor relative to my prior round: a below-60k true population only resolves at 60,000 if ICE publishes a newer qualifying release. Some of that probability should instead become 60,311 stale-workbook mass. I now think the combined probability of “floor or stale April 4 workbook” is around 60%, with the stale-workbook component closer to 25–30% than my previous 15–20%.

I still defend keeping meaningful upside. The system already reached the high-60k/low-70k range earlier in FY2026, so a rebound above 65,000 does not require new warehouse capacity. But I agree the mega-warehouse story is probably less August-relevant than the raw budget headlines suggest: AP reports 11 warehouses bought for more than $1 billion, a hold/review of the plan, an OIG investigation, and lawsuits or legal/environmental obstacles in multiple states. (apnews.com) So I am not increasing the high tail; I keep P(>65k) around 18%.

My revised mental buckets:

• ~31–33%: newer update shows ≤60k → resolves at 60,000.
• ~27–30%: no newer qualifying update / stale-publication mode → around 60,311.
• ~20–23%: updated value modestly above current, 60.3k–65k.
• ~17–19%: operational rebound or definition shift above 65k.

Updated prediction:
Percentile 0.1: 60000.001
Percentile 1: 60000.01
Percentile 5: 60000.05
Percentile 10: 60000.10
Percentile 20: 60000.20
Percentile 30: 60000.49
Percentile 40: 60310.70
Percentile 50: 60311.05
Percentile 60: 60311.40
Percentile 70: 61500
Percentile 80: 64200
Percentile 90: 69000
Percentile 95: 73500
Percentile 99: 85000
Percentile 99.9: 100000

Forecast — Post 460

Modal: 60000-60625 (48.0%) • partial_shift • 8m25s

Interpretation

Reading: strict

The resolution criteria are highly specific and literal: they name an exact cell location (Total row, Total column), exact sheet name (‘Detention FY26’), exact file format (XLSX), exact source URL, exact observation timestamp (2026-08-11 21:00 UTC), and explicit fallback hierarchy — leaving minimal room for interpretive latitude. The resolver must read a specific numeric value from a specific cell in a specific document.

Time bounds

• observation_window: Most recent FY2026 ICE detention statistics workbook linked from ice.gov/detain/detention-management with a posted/release date on or before 2026-08-10, as observed at 2026-08-11 21:00 UTC (with retry up to 2026-08-12 06:00 UTC if unreachable).
• resolution_date: 2026-08-11

Edge cases identified

1. If ICE publishes a new workbook between August 10 and August 11 at 21:00 UTC, it would be excluded since the cutoff is ‘posted on or before August 10, 2026’.
2. If the ‘Detention FY26’ sheet is renamed or restructured (e.g., combined FSC and Adult into different subtables), the resolver must identify the equivalent total currently-detained figure, which may require judgment.
3. If the XLSX is replaced by a dashboard or CSV, the equivalent total must be identified — the format transition introduces ambiguity about what counts as ‘equivalent’.
4. The lower bound of 60,000 functions as a floor, not an annulment — if the actual detained count is below 60,000 (e.g., due to a policy change), the question still resolves at 60,000.
5. If multiple FY2026 workbooks are linked with the same date stamp, it is unclear which one governs — the criteria say ‘latest date stamp’ but ties are not addressed.
6. The Wayback Machine fallback may capture a page state that does not include the most recently linked workbook, introducing a gap between what was ‘most recent before August 10’ and what the archive captured.

Research (3/3 variants, shared evidence pool)

Total evidence registered (shared pool): ?

Research Brief

Evidence confidence: medium

Scenario 1: Floor resolution at 60,000 — detained count at or below floor [medium evidence]

Conditions favoring

If the ICE currently-detained total in the most recent qualifying workbook is at or below 60,000. Sources 12, 13, 28 suggest the ‘currently detained’ snapshot has been hovering near or below 60,000 as of May–June 2026. Declining monthly bookings (source 14) and the downward trend from January peak support continued decline. If ICE publishes a new workbook before August 10 reflecting a sub-60,000 current count, or if no new workbook is published and the April 4 figure (60,311) represents the only reference, the floor is very close to current observed levels.

Conditions against

The April 4 authoritative figure is 60,311 (above the floor). Policy direction toward expansion (source 10, 22) and the $45B detention funding could push numbers upward before August. If ICE resumes data publishing and shows recovery above 60,000.

Scenario 2: Stable near-current: 60,000–62,000 range in the most recent qualifying workbook [high evidence]

Conditions favoring

If ICE publishes a new workbook before August 10 reflecting a currently-detained figure in the 60,000–62,000 range — consistent with the April 4 level of 60,311 (authoritative source) and the plateau suggested by multiple third-party reports through June 2026. The random-walk median without drift centers near 60,311. Declining booking rates and stable release rates could maintain this level.

Conditions against

Sources 28 and 12/13 suggest the level has drifted toward or below 60,000 since April, meaning the current level may already be at/below this range. A new upward enforcement surge (per source 10/22) could push above this range.

Scenario 3: Elevated rebound: 63,000–70,000 range [low evidence]

Conditions favoring

If a renewed enforcement surge occurs before August 2026, driven by the newly enacted $45B detention expansion law (source 10) and stated policy goal of 1 million annual removals (source 22). ICE could ramp up arrests and reopen or expand capacity, pushing currently-detained well above the April plateau. Historical precedent: ICE went from ~56,000 at FY2025 end to 73,400 by mid-January 2026 in only a few months.

Conditions against

Declining monthly booking rates from January to March 2026 (source 14); ‘recalibration’ as of May 2026 (source 22); multiple sources as of June 2026 showing ~60,000 rather than growth; operational constraints (capacity, litigation).

Scenario 4: No new workbook published; April 9 workbook (60,311) governs resolution [medium evidence]

Conditions favoring

ICE stopped publishing its biweekly detention data in early April 2026 (sources 15, 29). If ICE does not publish a new FY26 XLSX workbook before August 10, 2026, the April 9 workbook remains the most recent qualifying release, and the resolution value is 60,311 (just above the 60,000 floor).

Conditions against

ICE has a history of resuming data publication after pauses (resumed in April after the February/March gap per source 18). Two months is a long gap; political or legal pressure may force publication. The question’s fallback to Wayback Machine could capture any interim release.

Key uncertainties

• Whether ICE will publish a new FY2026 detention statistics XLSX workbook between June 12 and August 10, 2026 — the publication gap has already exceeded 10 weeks as of June 12
• The actual ‘currently detained’ snapshot value at the time of any new ICE publication, given third-party sources show divergence (some cite <60,000, others cite ~66,000 average daily which is a different metric)
• Whether the downward trend in detention counts observed from January peak through April 2026 has continued, reversed, or plateaued in May–August 2026
• Whether the $45B detention expansion (One Big Beautiful Bill Act) has produced an operational increase in detention capacity and bookings by August 2026
• Whether the ‘~66,000’ figure cited by Hearst/Houston Chronicle is average daily population (a different metric) or currently-detained snapshot — the authoritative XLSX shows both metrics separately
• The volatility of the ICE detention figure over the 60-day horizon from current data anchor (April 4) to resolution (August 10)

Forecast variants (3/3)

Variant 0: inside_view (inside_view_v1)

Inside View Causal Analysis

Current state → mechanisms → outcome distribution:

Current state: The authoritative ICE FY26 XLSX workbook (released April 9, 2026, data through April 4) shows “ICE Currently Detained” Total = 60,311 (source 4/21, [AUTHORITATIVE]). ICE stopped publishing its biweekly detention data in early April 2026 (sources 15, 29); as of June 12, 2026, no new workbook has been published for ~10 weeks.

Causal chain analysis:

1. If no new workbook is published by August 10 (scenario 4, relative_evidence_strength: medium): The April 9 workbook governs → resolution value = 60,311 → falls in 60,000–60,625 (bin_0). The gap is already 10+ weeks with no sign of resumption (source 29: ICE explicitly stopped publishing). This is plausible, but 2 months remains and ICE has historically resumed after gaps (source 18: resumed after 56-day gap in early 2026).

2. If ICE publishes a new workbook showing ~60,000–62,500 level (scenario 2, relative_evidence_strength: HIGH): The random-walk projection median without drift centers at ~60,311 (source 23). Third-party sources (ABC News June 12, Brookings May 18, Kocher May 22) all cite “around 60,000” or “under 60,000” — consistent with the April 4 plateau continuing. If the actual count is in 60,000–62,500, this maps to 60,000–60,625 (bin_0) or 60,625–61,250 (bin_1) or 61,250–61,875 (bin_2) or 61,875–62,500 (bin_3).

3. Floor at 60,000 triggered (scenario 1, relative_evidence_strength: medium): Kocher May 22 cited “under 60,000” as approximate current level (source 28). If a new workbook reports sub-60,000, resolution = 60,000 (floor) → bin_0. This adds mass to bin_0.

4. Elevated rebound to 63,000–70,000 (scenario 3, relative_evidence_strength: LOW): The $45B expansion (source 10) and 1M removal goal (source 22) provide policy pressure upward, but declining booking rates (source 14), “recalibration” as of May 2026 (source 22), and current ~60,000 third-party estimates make a near-term surge unlikely by August.

Key insight on scenario weighting:

• Scenario 2 (stable near-current, HIGH evidence strength) and Scenario 4 (no new workbook, MEDIUM) both converge to bin_0/bin_1 territory — the April 4 anchor of 60,311 is the strongest authoritative data point.
• Scenario 1 (floor, MEDIUM) adds more mass to bin_0.
• Scenario 3 (elevated rebound, LOW) spreads some mass into higher bins but is the weakest-supported scenario.

The 60,000 floor creates a truncation/compression at bin_0: any sub-60,000 outcome resolves there. The authoritative anchor is 60,311, which sits in bin_0. The random-walk projection’s median stays in bin_0/bin_1 range. The high-evidence scenario centers squarely in bin_0–bin_2.

Distribution construction:

• 60,000–60,625 (bin_0): Catches the floor trigger + the April 4 value (60,311) + slight downward drift scenarios. Very high mass. ~40%
• 60,625–61,250 (bin_1): Modest recovery from current level without new workbook or stable near-current. ~18%
• 61,250–61,875 (bin_2): Slight upward movement. ~12%
• 61,875–62,500 (bin_3): Moderate recovery. ~8%
• 62,500–63,125 (bin_4): Some recovery, consistent with average daily population being somewhat higher than snapshot. ~7%
• 63,125–63,750 (bin_5): Beginning of elevated range from low-evidence rebound scenario. ~6%
• 63,750–64,375 (bin_6): Low-evidence rebound territory. ~4%
• 64,375–65,000 (bin_7): Low-evidence tail. ~3%
• Above 65,000 (bin_8): Extreme rebound — very low probability given current trajectory. ~2%

The strong concentration in bin_0 reflects: (a) the 60,311 anchor sitting in that bin; (b) the floor mechanism compressing all sub-60,000 outcomes; (c) third-party sources suggesting current level at/below 60,000; (d) the high-evidence scenario explicitly covering 60,000–62,500 with its center at 60,311; (e) the low-evidence status of the rebound scenario.

Variant 1: outside_view (outside_view_v1)

Reference class: the ICE “ICE Currently Detained” snapshot values as published in the FY2026 biweekly detention XLSX workbooks on ice.gov/detain/detention-management. Base-rate anchor: the most recent authoritative FY26 workbook (FY26_detentionStats_04092026.xlsx) shows 60,311 as of April 4, 2026 (authoritative file; sources 4 and 21). Historically, successive ICE snapshot releases move by a few thousand rather than by tens of thousands, so absent strong contrary evidence the outside view stays tightly anchored near the last published value.

Adjustment reasoning (outside view, limited adjustments): I give heavy weight to the high-evidence scenario identified in the brief — “Stable near-current: 60,000–62,000” (relative_evidence_strength: high) — which centers mass around the April 4 value of 60,311 (sources 4/21 and the brief’s random-walk median, source 23). Medium-strength scenarios (no new workbook so the April 9 workbook governs; and the possibility the true snapshot drifts at-or-below the 60,000 floor) justify concentrating still more mass at or just above 60,000 (relative_evidence_strength: medium; sources 15, 29, 28). A low-evidence rebound scenario (rapid rise into the mid-60ks driven by enforcement/capacity expansion) receives only modest mass (relative_evidence_strength: low; sources 10, 22, historical January spike 2/3), because the brief marks that scenario as evidence-thin and precedent shows both rapid rises and declines are possible but not the default.

Concrete adjustments from the base-rate anchor: +/− a few percentage points at the bin level. The April 4 authoritative value (60,311) lies inside 60,000–60,625 (bin_0); the 60,000 floor (if a later official snapshot is ≤60,000) also resolves to bin_0. Accordingly I place the single largest probability in 55% (bin_0) and most remaining mass in the adjacent 60,625–61,250, 25% (bin_1), reflecting the high-evidence stable scenario and the medium-evidence risk that the official value either equals the April 4 figure or hits the 60,000 floor (sources 4/21, 12, 13, 28). Smaller but non-negligible probability is assigned to bins 2–3 (combined 12%) to capture ordinary snapshot volatility around the anchor (source 23 random-walk p25–p75 band). Bins 4–7 receive modest mass (combined 7%) to reflect the low-evidence elevated-rebound scenario (sources 10, 22, 2/3, and Hearst’s average-daily signals which are not the authoritative snapshot — source 19/20). I assign essentially zero but non‑zero tail mass (1%) to >65,000 (bin_8) because the brief explicitly treats such rebounds as possible but evidence-weak.

Other resolution notes used in weighting: the brief flags a publication gap — ICE stopped regular publishing after early April (sources 15, 29), so it’s plausible no newer workbook appears by August 10 and the April 9 file governs (this pushes mass to the April 4 value, bin_0; medium evidence). The resolution rules enforce a strict reading of the specified cell/sheet/file (the forecast reflects that strict anchor to the authoritative workbook if no newer file is present). I therefore concentrated mass near the last-authoritative snapshot rather than spreading widely.

Citations to brief facts used: authoritative April 9 workbook and its 60,311 total (sources 4 and 21), multiple third‑party reports citing ~60,000 or <60,000 (sources 12, 13, 28), the random-walk projection and drift scenarios (sources 23, 24), and the $45B detention expansion / policy push that supports a low-evidence rebound scenario (sources 10, 22). The scenario evidence-strength tags from the brief guided weighting (high -> most mass; medium -> significant mass; low -> small tail mass). Confidence in the evidence is medium (per the brief), so I kept a concentrated but not overconfident distribution.

Variant 2: contrarian_v2 (contrarian_v2)

Adversarial reading

The bulk of the evidence (authoritative April 9 workbook at 60,311 per sources 4/21/1/7, repeated third-party citations of ~60k or <60k in May–June per sources 12/13/28, sustained publication gap per sources 15/29) pulls the consensus toward resolution at or inside the 60,000 floor, i.e., bin_0.

Counter-evidence found (or absence)

The $45B expansion law and historical rapid rebound precedent (73,400 peak) exist but are tagged low-evidence for the August horizon; the high-evidence scenarios in the brief are the floor and stable-near-60k cases. No substantive counter-evidence contradicts the low/near-floor tilt.

Distribution implications

Because the floor is active and the high-evidence scenarios cluster at or below ~60,311, bin_0 receives materially more mass than a naive uniform spread across the 60k–65k+ range would assign. The small tail above bin_1 reflects only the low-evidence rebound scenario and the possibility of a new workbook showing modest recovery.

Consolidate

Strategy: logit_mean • Drafts: 3/3

Reviewers (3/3)

anchoring → shift_higher (medium magnitude, medium confidence) anchoring_v1

The draft places ~57.6% of mass in 60,000–60,625 (bin_0), treating the 60,311 April 4 authoritative anchor as the modal outcome AND stacking the 60,000 floor-trigger mass on top of it. This is a classic anchoring error: the forecaster anchored the p50 to the LAST KNOWN DATA POINT (April 4 = 60,311) rather than the modal expectation over a 4-month forward horizon (April→August). The brief’s own random-walk projection assigns a p5–p95 band of 48,366–72,256 around that anchor, implying a standard deviation of ~7,262 — which spans the entire question range and then some. Treating 60,311 as the modal bin-center ignores that the distribution widens materially over a 60-day horizon. Additionally, the rationale uses the phrase “the high-evidence scenario explicitly covering 60,000–62,500” but 60,000–62,500 spans bins 0–3, yet the draft allocates only ~7.4% to bins 1–3 combined (out of the 16% allocated to bins 1+), understating the within-range spread. The forward-looking distribution should be wider and more centered around 60,625–62,500, with bin_0 still elevated (due to the floor compression effect) but not holding nearly 58% of all probability mass.

Flagged concerns

• Anchor-to-last-known-value: The draft places the p50 exactly at the April 4 authoritative value of 60,311 (bin_0), treating a 10-week-old snapshot as the modal forward expectation over a further 60-day horizon. The random-walk projection (source 23) has a p5–p95 span of 48,366–72,256, implying the distribution should be materially wider. Placing 57.6% in the single bin containing the last known value overfits to the stale anchor.
• Floor-mass stacking without separation: The rationale bundles three distinct scenarios into bin_0: (a) sub-60,000 floor trigger, (b) no new workbook so April 9 governs (60,311), and (c) the stable-near-current scenario at 60,311. These are separate probability masses that happen to land in the same bin, but the aggregation of all three into ~58% is aggressive. Scenario (b) alone is medium-evidence, and the floor trigger (a) is also medium — their JOINT probability stacked into one bin creates overconfidence in a narrow range.
• Optimistic anchoring on publication gap: The draft treats ‘no new workbook = April 9 governs = 60,311’ as a near-certain outcome by noting ICE stopped publishing. But the brief also notes ICE previously resumed after a 56-day gap (source 18), the current gap is already 10+ weeks, and two more months remain. If ICE resumes publication and reports any different value — up or down — the governing snapshot changes. The draft insufficiently weights the scenario where publication resumes and shows a different level.

ceiling → shift_higher (low magnitude, medium confidence) ceiling_v1

The structural floor at 60,000 creates a hard compression mechanism that the draft acknowledges but may underweight in one critical respect: the April 4 authoritative value of 60,311 sits only 311 units above the floor, meaning any slight downward drift (which multiple sources confirm is occurring — Kocher May 22 cites “under 60,000,” ABC News June 12 cites “around 60,000”) would trigger the floor. The brief’s own downward-drift random-walk (source 24, -50/day drift) puts the August median at ~57,311, well below the floor — meaning under the conditioned downward-trend scenario, a large portion of the distribution collapses into bin_0. More critically, the draft does not explicitly account for the structural path-dependency: if ICE does NOT publish a new workbook by August 10 (medium-evidence scenario per sources 15, 29), the April 9 workbook governs and the resolution value is exactly 60,311 — a single point in bin_0. The combination of (a) floor compression of all sub-60,000 outcomes, (b) the April 4 anchor at 60,311 already in bin_0, and (c) third-party evidence the count has since fallen below 60,000, means bin_0 should dominate even more than the draft’s already-high 57.6%. The draft assigns ~8.4% to bins 4–8 (above 62,500) which seems too generous given the low-evidence status of the rebound scenario and the 10-week publication gap suppressing any upward-revised anchor.

Flagged concerns

• Floor compression underweighted: The 60,000 floor compresses ALL sub-60,000 outcomes into bin_0. The authoritative April 4 value is 60,311 (311 above floor). Third-party sources (Kocher May 22: ‘under 60,000’; ABC News June 12: ‘around 60,000’) suggest the true current snapshot is at or below the floor. If a new workbook is published showing ≤60,000, it resolves at exactly 60,000 — bin_0. If no new workbook is published, it resolves at 60,311 — also bin_0. The combined probability of these two dominant outcomes alone (no-new-workbook + floor-trigger scenarios, both medium evidence) suggests bin_0 should be ~65–70%, not 57.6%.
• Path-dependency on publication gap: Sources 15 and 29 confirm ICE stopped publishing in early April 2026, already a 10-week gap as of June 12. The resolution criteria state ‘most recent FY2026 workbook posted on or before August 10.’ If ICE does not resume publishing, the April 9 workbook (60,311, bin_0) is the only qualifying release. This structural path-dependency concentrates mass in bin_0 but the draft only partially reflects this.
• Elevated bins (4–8) over-weighted relative to evidence: The draft assigns ~8.4% cumulative to bins 4–8 (62,500–65,000+). The only scenario supporting these bins is the ‘elevated rebound’ scenario explicitly tagged relative_evidence_strength: LOW in the brief. With declining booking rates (source 14), a 10-week publication gap, and third-party estimates at ~60,000, the probability of 62,500+ by August seems overstated at 8.4%.

math → redistribute (medium magnitude, medium confidence) math_v1

The forecast’s distribution sums to 1.0029 (exceeds 1.0 by 0.0029), which is within the ±0.01 tolerance. However, there is a significant internal inconsistency: the rationale derives a random-walk median of ~60,311 (source 23, explicitly cited) and a “downward drift” median of ~57,311 (source 24), yet assigns 57.6% to 60,000–60,625 (bin_0) which covers only the 625-point range just above the floor. For a median of 60,311, this bin contains only the lower tail; coherent math would place the modal bin at bin_0 or bin_1 with the median centered in that bin, not compressed at its boundary. The stated “high-evidence scenario” explicitly covers 60,000–62,000 (source 2, brief), yet 63.8% of mass is assigned to bin_0 alone (60,000–60,625), implying the forecast doesn’t match its own stated scenario bounds.

Flagged concerns

• Probability mass placement vs. stated median: Rationale cites random-walk median ~60,311 (source 23) and downward-drift median ~57,311 (source 24), but assigns 57.6% to 60,000–60,625 (bin_0), which is only a 625-point range. For a median of 60,311 to be consistent, the highest density should be in bin_0 or bin_1, but the cumulative mass in bin_0 alone (57.6%) exceeds what a median at 60,311 would imply. This suggests the distribution is shifted lower than the stated median.
• Scenario-stated bounds vs. actual assignment: The brief’s high-evidence scenario (scenario 2, ‘Stable near-current: 60,000–62,000 range’) explicitly spans bins 0–3 (60,000–62,500). Yet the forecast assigns 57.6% to bin_0 alone. If the scenario is high-evidence and spans a 2,500-point range, the mass should be distributed more evenly across that range, not concentrated in the first 625 points of it.
• Interpretation of floor mechanism: The forecast correctly recognizes the 60,000 floor (bin_0 lower bound) and acknowledges it in the rationale (‘The 60,000 floor creates a truncation/compression at bin_0’). However, the floor should only apply if the actual detained count is ≤60,000. The April 4 authoritative value is 60,311 (above the floor), and the stable-scenario range is 60,000–62,000. The floor itself (bin_0) should capture only sub-60,000 outcomes, not dominate the entire distribution unless evidence strongly suggests the count has drifted below 60,000 since April 4.
• Low-evidence scenario weighting: The forecast correctly assigns low mass (9.9% total in bins 5–8) to the elevated-rebound scenario (63,000–70,000, relative_evidence_strength: low per the brief). This is mathematically sound and proportional to the evidence weight.

Reconcile: partial_shift

Reviews consumed: 3

Reviewers diverge on direction: the anchoring reviewer argues bin_0’s ~57.6% mass overfits to a 10-week-old stale snapshot and the distribution should widen (shift_higher); the math reviewer flags an internal inconsistency between the stated random-walk median (~60,311) and the heavy concentration in the lowest 625-point bin, recommending redistribution toward bins 1–3; the ceiling reviewer instead argues bin_0 should be even higher (~65–70%) due to floor compression plus the no-new-workbook path. I partially shifted: I accepted the anchoring and math reviewers’ point that bin_0 over-concentrates a 4-month-forward distribution into a 625-wide bin straddling the floor, and reallocated some mass into bins 1–3 where the high-evidence “stable 60,000–62,000” scenario actually lives. I partially rejected the ceiling reviewer’s call to push bin_0 even higher — the structural floor argument is real but the brief explicitly notes ICE has previously resumed publication after gaps (source 18) and the random-walk band is wide, so concentrating 67%+ in a 625-wide bin would compound the same overfitting the anchoring reviewer flagged. Net: bin_0 down from 57.6% to ~48% (still dominant due to floor + April 4 anchor + sub-60k third-party signals), bins 1–3 raised modestly to better reflect the high-evidence scenario span, and the elevated tail (bins 4–8) trimmed slightly in line with the ceiling reviewer’s valid point that the rebound scenario is low-evidence.

Rejected critiques

• [ceiling] bin_0 should be raised to ~65–70% due to floor compression plus the no-new-workbook scenario combining at the 60,311 point.

  • Why rejected: Accepting this would compound the anchoring/math reviewers’ valid concern about over-concentrating a 60-day-forward distribution into a single 625-point bin. ICE has resumed publication after prior gaps (source 18), and the random-walk p5–p95 band is ~24k wide, so a 67%+ bin_0 would understate forward uncertainty. I kept bin_0 dominant (~48%) to reflect the floor + anchor effects without over-narrowing.

• [anchoring] Distribution should be substantially wider, with bin_0 reduced toward ~37% and meaningful mass extended into bins 4–8.

  • Why rejected: Partially accepted — bin_0 reduced and bins 1–3 raised. But fully widening into bins 4–8 ignores the structural floor compression and the strong sub-60k third-party signals (Kocher, ABC, Brookings). The random-walk band is symmetric in the abstract but asymmetric in practice because the floor truncates the lower tail into bin_0.

Final Distribution

Evidence Sources (29)

1. TRAC: Immigration Detention Quick Facts — TRAC Immigration (tracreports.org) reported ICE held 60,311 people in detention as of April 4, 2026, per ICE data.
2. Vera Institute: Ten Things About ICE Detention Through March 2026 — ICE daily detention population reached a record high in mid-January 2026 of more than 73,400 people on a single day, per Vera Institute dashboard.
3. ICE Detention and Deportation by the Numbers (April 2026) — ICE detained population peaked at 70,766 on January 24, 2026 (another source says >73,400 in mid-January), reflecting a record high in the data series.
4. FY26_detentionStats_04092026.xlsx (ICE) — ICE published an FY2026 detention statistics workbook at https://www.ice.gov/doclib/detention/FY26_detentionStats_04092026.xlsx (filename with date 04/09/2026) that includes a ‘Detention FY26’ sheet and an ‘ICE Currently Detained by Processing Disposition’ table (per the file URL and directory).
5. TRAC: ICE Detention Is Down from Its Early-2026 Peak — From peak ~68,289 in early February 2026, ICE detention fell to 60,311 as of April 4, 2026 — a drop of ~8,000 (~12%) in roughly two months, per TRAC.
6. CBS News: ICE detained fewer non-criminals since Minnesota crackdown — Average daily ICE detention population was ~63,000 in March 2026 and ~72,000 in January 2026, per ICE-released data as reported by CBS News.
7. Austin Kocher Substack: ICE Detention Numbers Drop Significantly — As of April 4, 2026, ICE currently detained total was 60,311 — the lowest in six months, roughly equal to FY2025 end levels per ICE biweekly spreadsheet.
8. ICE Detention Trends - Vera Institute (updated April 2, 2026) — ICE detention population covered in Vera’s dashboard goes through March 10, 2026; in February 2026 ICE detained people in 456 facilities. ICE data is released biweekly on ice.gov/detain/detention-management.
9. ICE Detention Statistics Dashboard - DetentionStats.com — DetentionStats.com visualizes ICE biweekly XLSX detention statistics from ice.gov, covering FY24, FY25, and FY26, updated biweekly.
10. Brennan Center: Budget Bill Massively Increases Funding for Immigration Detention — The “One Big Beautiful Bill Act” (signed into law) allocates $45 billion for ICE detention expansion over ~4 years, with stated goal of expanding capacity from ~56,000 beds to potentially 100,000+.
11. Austin Kocher Substack (April 9, 2026) — ICE had not published detention data during DHS budget negotiations (“shutdown”), but resumed biweekly schedule with April 4, 2026 data release. The April 4 data showed 60,311 currently detained.
12. ABC News: Just 3% of recent ICE detainees had a violent felony conviction — As of June 12, 2026, ABC News reports ICE currently detains “around 60,000” people in federal immigration custody, consistent with the post-peak plateau.
13. Brookings: The administration has detained 400,000 immigrants — Brookings (May 18, 2026) reports around 60,000 people currently held in ICE detention, down from the January 2026 peak; notes $45 billion in new funding allocated to expand capacity further.
14. TRAC: ICE Detention Is Down from Its Early-2026 Peak — Monthly bookings into ICE detention declined from 39,694 in January 2026 to 32,531 in March 2026 — an 18% drop in new arrivals, per TRAC.
15. This Week by the {Immigration} Numbers - Austin Kocher Substack (May 22, 2026) — As of May 22, 2026, Austin Kocher’s newsletter noted ICE’s biweekly detention data was “over a month old” — suggesting ICE had not released a new detention report since approximately April 4, 2026 (over 6 weeks gap as of May 22).
16. Vera Institute: Ten Things About ICE Detention Through March 2026 (April 10, 2026) — According to the Vera Institute, ICE “has not published a new [biweekly detention] report since early February [2026] amid the recent U.S. Department of Homeland Security (DHS) shutdown” — suggesting sporadic reporting disruptions in early 2026.
17. Houston Chronicle/Statesman Texas ICE detention tracker (June 1, 2026) — As of early May-June 2026, multiple sources cite U.S. total ICE detention as “about 66,000” (Houston Chronicle, June 1 / Hearst Newspapers), citing ICE data through April 4 showing 18,000 in Texas (~27% of ~66,000 total).
18. Austin Kocher Substack: ICE Detention Numbers Drop Significantly (April 9, 2026) — ICE stopped publishing detention management data for ~56 days during a DHS budget negotiation period in early 2026, with the April 4 release being the first update since February 2026; ICE then returned to its biweekly schedule.
19. Texas ICE detention tracker - Austin Statesman (June 1, 2026) — Hearst Newspapers reported (June 1, 2026) that ICE detained “roughly 66,000” nationally, with ~18,000 in Texas (27%). This figure is more recent than the April 4 official report of 60,311 and suggests the detained population may have risen slightly after April.
20. Houston Chronicle Texas ICE detention tracker (May 6, 2026) — The “~66,000” figure cited by Houston Chronicle (May 6-June 1, 2026) refers to the average daily population from the ICE spreadsheet — a different metric than the “Currently Detained” snapshot total (60,311) used in the resolution criteria table.
21. ICE FY26 Detention Stats XLSX (direct link from ice.gov) — The most recent ICE FY26 detention XLSX file found was “FY26_detentionStats_04092026.xlsx” (released April 9, 2026) with data through 04/04/2026. No newer file found in searches as of June 12, 2026, consistent with Kocher’s May 22 note that data was “over a month old.”
22. LA Times: After major enforcement operations, Trump administration recalibrates immigration crackdown (May 1, 2026) — After major enforcement operations, the Trump administration recalibrated its immigration crackdown as of May 1, 2026: ICE arrests have declined but the agency plans to remove 1 million people annually with $170 billion in secured funding and expanding detention capacity.
23. level_forecast tool output — Level forecast from current value ~60,311 (April 4 ICE spreadsheet) over ~60 days (to August 2026 resolution): median 60,311; p5=48,366, p25=55,413, p75=65,209, p90=69,617, p95=72,256; std ~7,262. Recent trajectory shows sharp decline from 73,400 peak in January to ~60,311 by April, suggesting a downward drift possibility.
24. level_forecast tool output with drift — Level forecast with downward drift of -50/day (reflecting the observed declining trend): median 57,311 at resolution; p5=45,366, p25=52,413, p75=62,209, p90=66,617, p95=69,256. Under this scenario the lower bound (60,000 floor) would be triggered with non-trivial probability.
25. American Immigration Council: ICE Detention 59,000 event (scheduled August 27, 2026) — American Immigration Council scheduled an event for August 27, 2026 titled “ICE Detention 59,000: Record Levels, Horrible Overcrowding, Dwindling Due Process” — suggesting their expectation is the ICE detention population will be at or near 59,000 in late August 2026 (i.e., at or near the lower bound floor).
26. American Immigration Council event archive — CORRECTION: The American Immigration Council’s “ICE Detention 59,000: Record Levels” webinar was held on August 21, 2025 — when 59,000 was the record. The level subsequently rose to 73,400 by January 2026 and declined to ~60,311 by April 4, 2026.
27. Resolution criteria (provided) — The authoritative resolver-named source is https://www.ice.gov/detain/detention-management (the Detention Management page); the resolution rules require anchoring on the XLSX workbook linked from that page, not third-party aggregators (per the question’s ‘authoritative-source directive’).
28. Austin Kocher: This Week by the {Immigration} Numbers (May 22, 2026) — As of May 22, 2026, Austin Kocher’s immigration numbers newsletter cited “<60,000” as the approximate number of people currently in ICE detention — indicating the “currently detained” figure has fallen below 60,000, which triggers the question’s floor (resolves at 60,000).
29. This Week by the {Immigration} Numbers (June 6, 2026) — Austin Kocher Substack — Austin Kocher’s June 6, 2026 Substack post explicitly states: “ICE stopped publishing its regular detention figures in early April [2026], so the public sees these numbers only because the NYT was able to report on them.”

wooley_rhino • run 589f3e65 • $3.24 • 8m25s

Extracted Percentile Values: {0.1: 60001.1, 1.0: 60001.2, 5.0: 60001.6, 10.0: 60002.8, 15.0: 60005.0, 20.0: 60009.0, 25.0: 60014.0, 30.0: 60020.0, 35.0: 60035.0, 40.0: 60820.0, 45.0: 61320.0, 46.0: 61420.0, 47.0: 61530.0, 48.0: 61650.0, 49.0: 61790.0, 50.0: 61940.0, 51.0: 62100.0, 52.0: 62270.0, 53.0: 62450.0, 54.0: 62640.0, 55.0: 62840.0, 60.0: 63620.0, 65.0: 64480.0, 70.0: 65420.0, 75.0: 66460.0, 80.0: 67660.0, 85.0: 69120.0, 90.0: 71120.0, 95.0: 73820.0, 99.0: 77620.0, 99.9: 81820.0}

Model’s Answer: The best anchor is the official April 9, 2026 FY2026 workbook showing 60,311 currently detained as of April 4, cross-checked by TRAC and other analysts. I put much more weight on that exact-source signal than on generic historical notes. The next most important fact is the shape of recent movement: detention surged to the low 70,000s in January, then fell sharply back to about 60,000 by early April. That tells us both that the system can move fast and that a return to the January peak is possible but not the default.

Step one is anchoring on the exact metric. The resolution is not average daily population, not a monthly total, and not a broad detention-capacity number. It is one point-in-time “currently detained” total in the latest FY2026 workbook posted by August 10. So the April 4 reading of 60,311 is highly relevant.

Step two is deciding how much that anchor can move by late July or early August, which is probably the effective underlying observation period for the governing workbook. Four months is enough time for several-thousand-person movement, and early 2026 already showed swings larger than 10,000. But absent new hard evidence of another major surge, the safest base case is modest drift from about 60,000 rather than a straight-line return to 73,000-plus.

Step three is dealing with the floor. Because any value at or below 60,000 resolves to 60,000, the left tail is compressed. A relatively small downward move from 60,311 creates the same resolved outcome as a very large downward move. That makes the lower percentiles bunch near the floor.

The assumptions that matter most are these. First, that the April 4 level remains the right anchor for the summer path; I’d give that about a 25% chance of being meaningfully wrong, mainly because official updates appear sparse and there could be unpublished movement already underway. Second, that there is no major renewed enforcement or border-driven surge before early August that pushes detention back near the January high; I’d put about a 20% chance on that assumption failing. Third, that the relevant August workbook reflects a roughly current late-July or early-August snapshot rather than an unusually stale internal date; I’d give that about a 15% chance of being wrong.

A low outcome happens if the March-April downtrend in bookings persists, removals/releases keep pace, and the detained population slips below 60,000; because of the rule, that branch resolves at the floor. I think that broad low branch is substantial, around one-third.

A high outcome happens if interior enforcement intensifies, additional bed capacity is used aggressively, or there is a renewed inflow surge. I think the chance of finishing above 65,000 is meaningful but still clearly minority, roughly around 30%, with only a small chance of getting back into the mid-70,000s or higher by the relevant August snapshot.

Overall, I center the distribution just under 62,000, with a large chunk of probability mass effectively piled up at the floor and a longer right tail than left tail because the resolution rule truncates downside while upside remains open.

Percentile .1: 60001.1 Percentile 1: 60001.2 Percentile 5: 60001.6 Percentile 10: 60002.8 Percentile 15: 60005.0 Percentile 20: 60009.0 Percentile 25: 60014.0 Percentile 30: 60020.0 Percentile 35: 60035.0 Percentile 40: 60820.0 Percentile 45: 61320.0 Percentile 46: 61420.0 Percentile 47: 61530.0 Percentile 48: 61650.0 Percentile 49: 61790.0 Percentile 50: 61940.0 Percentile 51: 62100.0 Percentile 52: 62270.0 Percentile 53: 62450.0 Percentile 54: 62640.0 Percentile 55: 62840.0 Percentile 60: 63620.0 Percentile 65: 64480.0 Percentile 70: 65420.0 Percentile 75: 66460.0 Percentile 80: 67660.0 Percentile 85: 69120.0 Percentile 90: 71120.0 Percentile 95: 73820.0 Percentile 99: 77620.0 Percentile 99.9: 81820.0

The official FY2026 detention workbook is the binding public record and only changes when a new workbook is posted.

Publication cadence is intermittent and multi‑week gaps can freeze the reported ‘Total’ at its last posted pull regardless of live custody movements.

Recent documented pulls show a sharp early‑spring decline in the reported detained stock driven by fewer book‑ins and reduced border transfers.

Administrative funding and available bed capacity provide room for a recovery, so physical capacity is not an immutable ceiling.

If no newer workbook is posted by the cutoff, the recorded total will remain the April pull value even if the live population rises afterward.

If operations accelerate through higher book‑ins, renewed transfers, or rapid contracting activations, a later workbook could register a substantial mid‑summer rebound.

The outcome is highly sensitive to the timing of a new post and to any unannounced changes in reporting definitions or table structure.

Operational variability in book‑ins, border flows, and the pace of capacity deployment are the primary unresolved forces that determine whether the total stays near the low‑61k range or climbs above 65k.

Forecast (median): 60,770.8266

• The most recent confirmed ICE spreadsheet value (published in early April) puts the “currently detained” total at 60,311—and if ICE doesn’t publish a newer file by the cutoff date, that April figure is what the question would use.
• There’s a strong chance the result gets clipped to 60,000 if the newest published total comes in at or below that level, because the question has a closed lower bound.
• Since the count dropped sharply from a January high to the April level, forecasts are anchored near the low-60,000s, with downside limited by the 60,000 floor.
• At the same time, reported “around 60,000” in spring/early summer and historical summer patterns allow for a modest rebound if ICE publishes again in time.
• The biggest swing factor is whether ICE publishes a newer FY2026 detention spreadsheet before Aug 10—publication timing largely determines whether the result stays near 60,311 or reflects a later level.