FutureSim — 2026-07-08
13 forecasts · 11 active · 2 resolved today
· regenerated 2026-07-08 01:14Z
Paper PnL · cumulative
$+301.70
(5 open · $+225.95 realized)
Avg Brier (resolved today)
0.001
lower is better; 0 = perfect, 1 = totally wrong
Real-money readiness — paper-driven gates
The FutureSim paper evaluates the agent against ground truth (Brier),
not against market price. Beating the market net of friction is a
strictly harder bar — these three gates have to clear before paper
becomes real. See docs/futuresim-research.md §
Real-money readiness.
| Gate | Current |
Threshold | Status |
|---|
| Forward-resolved marketsCounted only when the agent predicted a market while it was OPEN and then watched it resolve. Hindsight forecasts on already-closed markets (the RESULTS-SINCE-LAST-SESSION history bundled into every run) don't count. | 73 | ≥ 50 | PASS ✓ |
| Net paper PnL (after 2% fee)Beating the market net of friction is a strictly harder bar than the paper's Brier-vs-truth evaluation. | $+207.95 over 15 closed | > $0 across ≥ 50 closed | NO |
| Calibration RMSEReported only once gate 1 passes — RMSE on fewer than 50 resolved predictions is noise. | 0.132 | < 0.10 | NO |
❌ DO NOT BET real money — gates not yet cleared.
Top edges — active markets
Deduped by question family so threshold ladders don't drown out other markets.
| Title | Agent | Market |
Δ | Resolves |
|---|
| US obtains Iranian enriched uranium by December 31? | 10% | 12% | -2.5pp | 2026-12-31 |
| GPT-5.6 released by July 31, 2026? | 97% | 99% | -1.9pp | 2026-07-31 |
| US obtains Iranian enriched uranium by September 30? | 6% | 4% | +1.0pp | 2026-09-30 |
| GPT-5.6 released by July 15, 2026? | 92% | 92% | -0.9pp | 2026-07-15 |
Calibration plot (forward-resolved predictions)
Belief evolution — top markets by edge
Solid = agent's YES probability. Dashed = Polymarket. One color per market.
Lines emerge after ≥2 days of data; day-1 runs show single points.
Trade-the-move (convergence) — second PnL track
Same entries as the hold track, but each position is sold when the market price converges to the agent's number (green fill), stopped on a 15pp adverse move (amber), or held to resolution (fallback). Solid = agent, dashed = market; the shaded gap is the edge being harvested. Separate from the readiness gates — a parallel backtest.
| Strategy | Realized | At risk (unrealized) | Open |
|---|
| Hold to resolution | $+225.95 | $+75.75 | 5 |
| Trade the move | $+180.20 | $+0.00 | 2 |
GPT-5.6 released by July 8, 2026? · NO · +$45.00 (converged, 1d)
GPT-5.6 released by July 7, 2026? · NO · +$35.00 (converged, 1d)
US announces new Iran agreement/ceasefire extens · NO · +$26.50 (converged, 1d)
US announces new Iran agreement/ceasefire extens · NO · +$21.00 (converged, 1d)
GPT-5.6 released by July 6, 2026? · NO · +$20.50 (converged, 1d)
US announces new Iran agreement/ceasefire extens · NO · +$17.50 (converged, 1d)
| qid | title | side | entry→exit | days | exit | P&L |
|---|
2658039 | GPT-5.6 released by July 8, 2026? | NO | 0.40→0.84 | 1 | converged | +$45.00 |
2760614 | GPT-5.6 released by July 7, 2026? | NO | 0.54→0.89 | 1 | converged | +$35.00 |
2364499 | US announces new Iran agreement/ceasefire extens | NO | 0.58→0.85 | 1 | converged | +$26.50 |
2364498 | US announces new Iran agreement/ceasefire extens | NO | 0.67→0.88 | 1 | converged | +$21.00 |
2658038 | GPT-5.6 released by July 6, 2026? | NO | 0.54→0.74 | 1 | converged | +$20.50 |
2354002 | US announces new Iran agreement/ceasefire extens | NO | 0.78→0.95 | 1 | converged | +$17.50 |
All 13 forecasts (click a column header to sort)
| qid | title | Agent |
Market | Δ pp |
Resolves | Resolved |
|---|
2006983 | US obtains Iranian enriched uranium by December 31? | 10% | 12% | -2.5pp | 2026-12-31 | |
2100070 | GPT-5.6 released by July 31, 2026? | 97% | 99% | -1.9pp | 2026-07-31 | |
2757753 | US obtains Iranian enriched uranium by September 30? | 6% | 4% | +1.0pp | 2026-09-30 | |
2658111 | GPT-5.6 released by July 15, 2026? | 92% | 92% | -0.9pp | 2026-07-15 | |
2658110 | GPT-5.6 released by July 13, 2026? | 86% | 86% | +0.5pp | 2026-07-13 | |
2648984 | GPT-5.6 released by July 17, 2026? | 92% | 92% | -0.3pp | 2026-07-17 | |
2648985 | GPT-5.6 released by July 24, 2026? | 96% | 96% | -0.2pp | 2026-07-24 | |
2658039 | GPT-5.6 released by July 8, 2026? | 4% | 4% | +0.0pp | 2026-07-08 | |
2760620 | GPT-5.6 released by July 9, 2026? | 76% | 76% | +0.0pp | 2026-07-09 | |
2648866 | GPT-5.6 released by July 10, 2026? | 84% | 84% | +0.0pp | 2026-07-10 | |
2355076 | US obtains Iranian enriched uranium by July 31? | 2% | — | — | 2026-07-31 | |
2492625 | US obtains Iranian enriched uranium by July 31? | 2% | — | — | 2026-06-30 | ✓ |
2760614 | GPT-5.6 released by July 7, 2026? | 5% | — | — | 2026-07-07 | ✓ |