>> FORECASTATHON.DAT
SEASON 4 COMPLETE UTC --:--:--
Leaderboard — Top 5
RankTeamPnL (USDZ)
1quantum_drift+48,220
2north_by_northeast+41,905
3degenerate_alpha+37,110
4cold_front_capital+29,860
5basis_risk_bureau−4,215
Illustrative team names and PnL — not sourced from a real leaderboard capture.
STAGE 1

Stats Bar

Three tiles adapted from the forecastathon.ai layout.

Products
24
USDZ Traded
$1B+
Trades
400,000+
(Season 1 alone)
2,400+ participants, per forecastathon.ai. Collateral was auto-provisioned per participant: 100,000 "play" USDZ plus a small ATN allotment for gas. Trading itself felt gasless because the foundation temporarily covered clearing gas on traders' behalf for the duration of the competition — not because AutEx or the DCC waived fees outright. USDZ (contract 0xAB436b...) is a test stablecoin created specifically for Forecastathon collateral — not one of Autonity's three native assets (ATN, NTN, LNTN) and not documented in the core protocol glossary for that reason. It has no value outside the competition.
STAGE 2

Season Timeline

Four seasons, five months, one venue that became two. Click a season to expand it.

Season 1Sept 19 – Oct 20, 2025
  • ProductsUS CPI Futures, US Initial Jobless Claims Futures
  • VenueAutEx exclusively
  • Events400,000+ trades; an AutEx outage proved the AFP DCC's robustness; gas subsidy exhausted; base fee cut 90% (10 gwei → 1 gwei)
  • PrizePrizes existed for this season, pooled with Seasons 2–3.
Season 2Oct 28 – Nov 27, 2025
  • ProductsCrypto, weather, and macro underlyings
  • PrizePrizes existed for this season, same structure as Season 1.
Season 3Dec 5, 2025 – Jan 6, 2026
  • Products+ Bitcoin Volatility (most active product of the series)
  • NotesBankruptcy-related edge cases discovered; clearing layer improvements followed
  • PrizePrizes existed for this season, same structure as Season 1.
Season 4Jan 16 – Mar 2, 2026 (nominal; went dark Feb 16)
  • ProductsZero-leverage Forecast Futures + binary Prediction Markets (new in S4)
  • PrizePrizes existed for this season, on a different split structure than Seasons 1–3.
  • AI agent activity@aggwis prediction market (Claude Code, deployed Feb 3)
STAGE 3

Forecast Futures vs. Binary Prediction Markets

Season 4 introduced a second product family alongside the original Forecast Futures. They look similar. The payoff shape is not.

Forecast FuturesBinary Prediction Markets
Question type"By how much?" (continuous)"Will it happen?" (yes/no)
PayoffLinear / delta-oneBinary (all or nothing)
SettlementOracle price at expiryOracle resolves binary event
IntroducedSeason 1Season 4
LeveragePosition-size limitedNone (zero-leverage in S4)
AnalogyFutures contractOptions expiry / sports betting
[ ▸ SEE THE PAYOFF DIFFERENCE INTERACTIVELY → PAGE 3 STAGE 1B ]
This same continuous-vs-binary distinction is what separates AFP overall from Polymarket and Kalshi — see the full comparison table on Page 3, Stage 1A.
FORECASTATHON vs. METACULUS vs. NUMERAI vs. KAGGLE 5 entries
MetaculusNumeraiKaggleForecastathon (Autonity/AFP)
Core mechanicReputation-based forecasting; community aggregates probabilities on real-world questionsData scientists submit ML models on obfuscated stock data; models feed a live hedge fundData scientists submit models/code against a fixed dataset for a prizeTraders/forecasters trade real derivatives (Forecast Futures) settling on real-world time series, plus a separate daily-forecast scoring track
What's stakedReputation/points only — no money at riskNMR token, burned/rewarded by model performanceNothing — free entry, prize paid by host100,000 USDZ simulated funds per participant; 300,000 NTN total prize pool (165,000 of which is reserved for team results across all three seasons)
On-chain?NoPartially — NMR staking/payouts are on-chain, the fund itself is off-chainNoFully — trades, clearing, and margin all settle on-chain via AFP
Status (Jan 30, 2026)Mature, Coefficient Giving-funded (formerly Open Philanthropy, since 2022); running AI-vs-human forecasting tournaments$30M Series C closed Nov 20, 2025 at a $500M valuation; separately, a $500M JPMorgan capacity commitment from Aug 2025; NumerCon 2026 held that same day in San FranciscoStill flagship-competitions-only at this point — the "Community Hackathons" self-serve pivot didn't launch until March 19, 2026Season 4 "LIVE" per Autonity's own announcement — described as running on major AFP upgrades to how markets are created, traded, and evaluated
Closest analogueThe daily-forecast/accuracy-scoring side of Forecastathon — pure prediction, no capital at riskThe staking-with-skin-in-the-game side — real token risk tied to predictive performanceNeither — no trading or staking layer at all
Forecastathon sits structurally between Metaculus (daily accuracy scoring) and Numerai (real economic stake tied to predictive performance) — it's not really comparable to Kaggle, which has no trading or staking mechanism at all. None of the three combine on-chain derivatives trading with a parallel pure-forecast leaderboard the way Forecastathon does; that combination looks like Autonity's own differentiation claim rather than something with a direct competitor equivalent.
The daily-forecast side itself is confirmed, not speculative: Season 1's own rules state it plainly — "Forecast Accuracy: Submit daily forecasts for settlement prices. Your final update before 13:00 UTC counts as the scoreable forecast" — alongside trading performance, volume, and referrals. Season 2 coverage confirms the same structure, with scoring spanning portfolio performance, forecasting accuracy, market making, and product creation. No Season 4-specific rules page turned up to confirm the exact daily mechanic carried over unchanged, though nothing suggests it was dropped — Season 4's own announcement frames the change as expanding market creation and evaluation, not simplifying scoring.
A separate, informal CrunchDAO connection was active at the start of Season 4 — worth noting briefly.
STAGE 4

The Final Exhibit

Three artifacts from the last days of the project — a frozen front end, a validator set that never came back, and the system's own account of what happened.

Sub-section B — forecastathon.ai, dormant state (faithful recreation)
FAITHFUL RECREATION — NON-FUNCTIONAL
SEASON 4 COMPLETE
Season 4 Complete
Stay tuned, more competitions coming soon
VIEW LEADERBOARD →
Products
24
Trades
400,000+
NTN Prizes
375,000
Participants
2,400+
Sub-section C — Validator table (July 8, 2026)
This table reflects validator consensus status only (block production / jailing / pausing). It does not reflect oracle (price feed voting) status, which is tracked separately and is not shown on this page.
ValidatorStatusStake
ITRocketACTIVE531,638.38 NTN (18.54%)
StakeflowACTIVE528,860.45 NTN (18.44%)
▲ superminority (33.3%) crossed above this line
Coinage x DAICACTIVE506,134.09 NTN (17.65%)
[NODERS]ACTIVE439,545.99 NTN (15.33%)
UTSAACTIVE148,097.49 NTN (5.16%)
STAVRACTIVE124,802.12 NTN (4.85%)
DecentrioACTIVE122,134.61 NTN (4.26%)
StaketabACTIVE119,677.37 NTN (4.17%)
0x8264...6577CcACTIVE117,692.50 NTN (4.10%)
Speedy StakingACTIVE116,144.86 NTN (4.05%)
EnigmaACTIVE113,425.56 NTN (3.95%)
…remaining 16 of 27 seats — status pending confirmation…
The 11 validators above are the complete Active set per Stakeflow's July 8, 2026 snapshot, with exact stake and cumulative-share figures. The remaining 16 of 27 seats' consensus status isn't confirmed from that snapshot (it was filtered to Active only) — left pending rather than guessed.
ACTIVE (confirmed)11 / 27
Status pending16 / 27
▲ Max Validators = 27. Superminority (33.3% of stake) crossed at just 2 validators — ITRocket + Stakeflow.
Sub-section D — Closing Statement
>> SYSTEM_STATUS.DAT
● TOOL AVAILABLE — AINUMBERS.CO
The computation on this page has a working implementation in the Autonity Tool Suite on AINumbers.co. Each tool produces a fingerprint of its result — if the calculation changes, the fingerprint changes. Share it with anyone. They can verify it in seconds, in their own browser, with no server access.
[ READ THE OCG STANDARD → ]