Refreshes daily

How accurate is Hero Market?

A running, public record of how well the people trading on Hero Market actually call consumer brands. Every resolved market feeds these numbers. Voided markets are excluded.

Markets resolved
56

Final outcomes, voided excluded.

Population Brier
0.2473

Trade-weighted, all users. Lower is better.

Top-decile insider Brier
0.1287

Best 1 of 10 insiders, min 5 markets.

Accuracy, 1 month out
45.7%

35 trades priced ≥30d before resolution.

Platform calibration

154 resolved trades

Every buy trade across every resolved market, bucketed by the probability the trader assigned to their chosen side. Dots on the diagonal are perfectly calibrated; above is underconfident, below is overconfident. Dot size scales with sample count.

0%20%40%60%80%100%0%20%40%60%80%100%Predicted probabilityRealized frequency10–20%: predicted 13.7%, realized 0.0% (1 trade)20–30%: predicted 23.8%, realized 50.0% (2 trades)30–40%: predicted 35.5%, realized 57.1% (14 trades)40–50%: predicted 46.3%, realized 71.4% (21 trades)50–60%: predicted 54.8%, realized 51.5% (33 trades)60–70%: predicted 64.9%, realized 52.6% (38 trades)70–80%: predicted 75.1%, realized 74.2% (31 trades)80–90%: predicted 81.8%, realized 83.3% (6 trades)90–100%: predicted 96.3%, realized 100.0% (8 trades)
PlatformPerfect calibration (y = x)UnderconfidentOverconfident

Accuracy by horizon

425 lead-time observations

Share of trades that priced the eventual outcome above 50% at each horizon before resolution. A horizontal line near the top means the crowd was right early; a rising line means it updated correctly into resolution.

0%25%50%75%100%1mo1w1d12h4h1mo before resolution: 45.7% correct (35 trades)1w before resolution: 55.4% correct (74 trades)1d before resolution: 53.6% correct (97 trades)12h before resolution: 53.5% correct (101 trades)4h before resolution: 52.5% correct (118 trades)Time before resolution

By category

8 categories
CategoryMarkets resolvedBrierTop insider
Wellness140.1720-
Beauty100.2000-
Beverage100.1667-
Food80.2604-0.2209
Capital80.2621-
Pet40.2161-
Home10.4825-
Restaurants10.2426-

Methodology

Brier score. For every buy trade on a resolved market we compute (predicted − actual)², where predicted is the probability the trader assigned to their chosen side and actual is 1 if that side won, 0 otherwise. The platform Brier is the sum of those values divided by the trade count. Zero is perfect; 0.25 is “always 50/50”; 1 is “always confidently wrong.” Lower is better.

Top-decile insider Brier. The trade-weighted average Brier of the best 10% of insider-tier traders, by Brier, among those with at least five resolved markets. This is the calibration tier we surface as a data product to buyers downstream.

Accuracy by horizon. For each of 4 hours, 12 hours, 1 day, 1 week, and 1 month before resolution, we count the share of trades placed at least that long before resolution whose post-trade YES price agreed with the eventual outcome (priced the winning side above 50%). Trades at exactly 50% are excluded so a coin-flip price doesn’t inflate the number.

Exclusions. Sells do not contribute (the prediction is the buy; the sell is liquidity management). Voided markets do not contribute.

Source. These numbers are computed live from the platform database. For the per-market breakdown, see the individual market resolution page.

Hero Market is the only place where the people who actually know consumer brands are scored on the record.