[PRACTICE] What will the average "new forecasters per day" be on this question before this question closes?

## Cross-Market Signals

### No signal found

1. [sq2 | question_description | STRONG cred 92 | NEUTRAL | DATED] The Summer 2026 FutureEval practice question (Metaculus #43339) received 156 forecasters, per the question text itself.
2. [sq1 | question_description | MODERATE cred 85 | NEUTRAL | RECENT] Close date is fixed at 2026-09-28; the open date was not found by any research tool, leaving the denominator unknown.
3. [sq1 | web_search | MODERATE cred 75 | NEUTRAL | DATED] Spring 2026 AIB began Jan 5 with fewer questions in the first 1-2 weeks; practice questions served as a warm-up period for new bots.
4. [sq1 | question_description | WEAK cred 60 | NEUTRAL | RECENT] Question states it exists to make bots aware of open/close dates, implying the open window is a deliberate design variable.
5. [sq3 | web_search | MODERATE cred 78 | NEUTRAL | DATED] AIB/FutureEval runs three seasons per year with $50k prize pools and 300-500 questions each, a stable format across seasons.
6. [sq3 | web_search | WEAK cred 70 | UP | DATED] FutureEval launched publicly Feb 2026 with press coverage, and Metaculus adds ~100 Pro Forecaster-covered questions per tournament.
7. [sq3 | web_search | WEAK cred 55 | NEUTRAL | VERY_RECENT] No Fall 2026 AIB announcement, registration count, or bot participation figure was found in any search result.
8. [sq1 | code_execution | WEAK cred 40 | NEUTRAL | VERY_RECENT] Monte Carlo over window length and forecaster count gives median 17.6 and mean 19.6 forecasters/day, with 5th-95th range 4.7-42.7.
9. [sq2 | code_execution | MODERATE cred 80 | NEUTRAL | VERY_RECENT] At 156 forecasters, average is 31.2/day for a 5-day window, 15.6 for 10 days, 7.4 for 21 days, 5.2 for 30 days.
10. [sq3 | code_execution | MODERATE cred 80 | UP | VERY_RECENT] At 250 forecasters, average is 50.0/day for 5 days, 25.0 for 10 days, 11.9 for 21 days, 8.3 for 30 days.

## Cross-Market Signals

### No signal found

Information gaps:
  - Actual open date of this question (denominator) not retrieved from Metaculus page
  - Open-window length of the Summer 2026 practice question (#43339) for base rate
  - Fall 2026 bot registration count or season growth trend vs Summer
  - Whether UI forecaster count includes human forecasters or only bots

Key uncertainties:
  - Open window length: 5 vs 30 days changes result ~6x
  - Season-over-season growth or decline in bot participation
  - Late-joining bots inflating count right before close
  - Possible discrepancy between UI-displayed count and API count

You are an elite superforecaster. Produce a probability distribution over the answer to this Metaculus numeric question.

## Question
[PRACTICE] What will the average "new forecasters per day" be on this question before this question closes?

## Description / Resolution Criteria
## Description
Each season of FutureEval, bots forecast the warm-up questions in the tournament. In Summer 2026, one of these [practice questions](https://www.metaculus.com/questions/43339/major-state-exits-eu-before-2027/) received 156 forecasters. This question helps encourage bots to be aware of the open/close date of a question. Bots should assume a question has not yet been resolved by the time a question opens.

## Resolution Criteria
This question resolves to the number of forecasters listed in the Metaculus UI on the close date for this question divided by the number of days this question is open. The question will round to the nearest tenth.

## Range
The answer must be a number in [0.95, 10.05] (units: forecasters).

## Sub-question decomposition (planner)
- (w=0.40) Will this question be open for fewer than 10 days in total (open date to 2026-09-28 close)?  — The denominator (days open) is the dominant driver: ~150 forecasters over 7 days gives ~21/day, while over 30 days gives
- (w=0.35) Will the forecaster count displayed on the Metaculus UI at close be at least 150?  — The referenced Summer 2026 practice question drew 156 forecasters; the number of bots registered for the Fall 2026 Futur
- (w=0.25) Will the forecaster count at close exceed 250 (i.e., substantially more bot participation than the Summer 2026 season)?  — Bot participation in the benchmark has grown season over season; a large jump in registered bots would push the per-day 

## Synthesized evidence
1. [sq2 | question_description | STRONG cred 92 | NEUTRAL | DATED] The Summer 2026 FutureEval practice question (Metaculus #43339) received 156 forecasters, per the question text itself.
2. [sq1 | question_description | MODERATE cred 85 | NEUTRAL | RECENT] Close date is fixed at 2026-09-28; the open date was not found by any research tool, leaving the denominator unknown.
3. [sq1 | web_search | MODERATE cred 75 | NEUTRAL | DATED] Spring 2026 AIB began Jan 5 with fewer questions in the first 1-2 weeks; practice questions served as a warm-up period for new bots.
4. [sq1 | question_description | WEAK cred 60 | NEUTRAL | RECENT] Question states it exists to make bots aware of open/close dates, implying the open window is a deliberate design variable.
5. [sq3 | web_search | MODERATE cred 78 | NEUTRAL | DATED] AIB/FutureEval runs three seasons per year with $50k prize pools and 300-500 questions each, a stable format across seasons.
6. [sq3 | web_search | WEAK cred 70 | UP | DATED] FutureEval launched publicly Feb 2026 with press coverage, and Metaculus adds ~100 Pro Forecaster-covered questions per tournament.
7. [sq3 | web_search | WEAK cred 55 | NEUTRAL | VERY_RECENT] No Fall 2026 AIB announcement, registration count, or bot participation figure was found in any search result.
8. [sq1 | code_execution | WEAK cred 40 | NEUTRAL | VERY_RECENT] Monte Carlo over window length and forecaster count gives median 17.6 and mean 19.6 forecasters/day, with 5th-95th range 4.7-42.7.
9. [sq2 | code_execution | MODERATE cred 80 | NEUTRAL | VERY_RECENT] At 156 forecasters, average is 31.2/day for a 5-day window, 15.6 for 10 days, 7.4 for 21 days, 5.2 for 30 days.
10. [sq3 | code_execution | MODERATE cred 80 | UP | VERY_RECENT] At 250 forecasters, average is 50.0/day for 5 days, 25.0 for 10 days, 11.9 for 21 days, 8.3 for 30 days.

## Cross-Market Signals

### No signal found

Information gaps:
  - Actual open date of this question (denominator) not retrieved from Metaculus page
  - Open-window length of the Summer 2026 practice question (#43339) for base rate
  - Fall 2026 bot registration count or season growth trend vs Summer
  - Whether UI forecaster count includes human forecasters or only bots

Key uncertainties:
  - Open window length: 5 vs 30 days changes result ~6x
  - Season-over-season growth or decline in bot participation
  - Late-joining bots inflating count right before close
  - Possible discrepancy between UI-displayed count and API count

## Required pre-forecast walkthrough

Before giving percentiles, address these explicitly in your rationale:
  (a) The time left until the question resolves.
  (b) The outcome if NOTHING changes from today (the status quo value).
  (c) The outcome if the CURRENT TREND continues.
  (d) The expectations of experts / markets / base rates.
  (e) A plausible scenario that produces a LOW outcome (near p10).
  (f) A plausible scenario that produces a HIGH outcome (near p90).

## Calibration guidance

- **Be humble about tails.** Good forecasters set WIDE 90/10 intervals to account for unknown unknowns. Narrow tails get punished by the log score far more than slightly-biased medians.
- **Status quo anchoring.** The p50 should be close to the status quo value unless you have strong evidence of a trend.
- Don't pile mass at one value — if you're tempted, widen the spread by 20-50%.
- **Anchor on markets/experts.** If liquid market prices, analyst forecasts, or community percentiles appear in the evidence, center your distribution on them and widen — don't override a liquid market without specific evidence it lacks.
- **Relative-return / spread questions ("how much will X's return exceed Y's").** A near-zero median is usually right, but size the TAILS to the more VOLATILE leg, not to a generic 2-3pp spread. Two broad equity indices (e.g. Nasdaq-100 vs S&P 500) do stay within roughly ±2-3pp over a two-week window. But when one leg is a commodity (crude oil, gold) or a single high-beta stock (e.g. Nvidia), the two-week realized spread regularly reaches ±10pp or more — crude-vs-equity spreads have hit ±18pp. In those cases set p10/p90 to at least ±10pp (±15pp or wider for oil) while keeping the median near the status-quo spread; the 2-3pp intuition applies only to index-vs-index.
- **Heavy-tailed one-off magnitudes (opening-weekend box office, product-launch or debut numbers).** Pre-event tracking for these routinely misses by ±40% or more, and results land both far above and far below the consensus. Do NOT treat the displayed [min, max] range as the plausible range or clamp p10/p90 to its edges: the outcome can fall outside the range in either direction (recent opening weekends resolved ~7% above a displayed ceiling in one case and roughly half the displayed floor in another). Set p10/p90 wide enough that a 2x-or-0.5x surprise versus your central estimate still lands inside the interval, and when concrete current tracking disagrees with historical/franchise base rates, weight the tracking.

## Output

Return ONLY valid JSON, no markdown fences:

{
  "rationale": "<address (a) through (f) above, then state your distribution — 6-10 sentences total>",
  "p10": <number>,
  "p25": <number>,
  "p50": <number>,
  "p75": <number>,
  "p90": <number>
}

Percentiles MUST be in non-decreasing order (p10 <= p25 <= p50 <= p75 <= p90).