Scoreboard
A whale taking a side at 80¢ should be right about 80% of the time just from the price. Each graded flag adds its entry price to expected. Edge = called − expected, per flag. Only flags placed ≥ 1h before a market's start/close count.
Most whales lose to the price. A few don't. We find them, and test that the finding holds up on bets placed later.
Out-of-sample test: do proven wallets keep winning?
To avoid hindsight bias we pick "proven" wallets (≥20 graded flags, positive edge) using only markets that resolved before a cutoff date, then score only the bets those wallets placed after it. Shown whatever it says.
| Cutoff | Proven wallets | Their later bets | Edge (later) | All whales (later) |
|---|---|---|---|---|
| Jun 10, 20:30 EDT | 7 | 43/51 | +8.2 pts ±5.4 | -3.0 pts |
| Jul 06, 08:10 EDT | 14 | 127/177 | +6.8 pts ±3.2 | -2.8 pts |
± is one standard error. A small number of later bets means the result is noisy. Not financial advice.
By entry price
| Bucket | Graded | Called | Expected | Hit vs expected | Edge |
|---|---|---|---|---|---|
| .00-.20 | 1249 | 153 | 131.3 | +1.7 pts | |
| .20-.40 | 1191 | 256 | 364.4 | -9.1 pts | |
| .40-.60 | 3045 | 1381 | 1536.4 | -5.1 pts | |
| .60-.80 | 2245 | 1552 | 1557.7 | -0.3 pts | |
| .80-1.00 | 2832 | 2417 | 2553.7 | -4.8 pts |
By category
| Category | Graded | Called | Expected | Edge |
|---|---|---|---|---|
| Crypto | 248 | 146 | 163.1 | -6.9 pts |
| Economy | 705 | 455 | 438.7 | +2.3 pts |
| Other | 3102 | 1776 | 1914.8 | -4.5 pts |
| Politics | 667 | 437 | 429.8 | +1.1 pts |
| Sports | 5840 | 2945 | 3197.0 | -4.3 pts |
Excluded late flags (placed <1h before start/close): 12984 called of 25099 vs 13294.89 expected (-1.2 pts).
Pending: odds move 29, whale trade 448 · scoreboard rebuilt Oct 24, 01:28 EDT