How to use WhaleGrader
Short version: most big Polymarket bets aren't smart money, so don't copy whales blindly. Use the record to see who has actually beaten the price, then check whether any of them is on your side of a market.
1. Why filter at all
Across 10,562 graded $10k+ bets, whales won 54.5% of the time. Their entry prices implied 58.2%. That gap is the edge: -3.6 pts. Copying every whale would have done worse than just trusting the market price.
Now the filter. We picked 14 wallets using only bets placed before Jul 06, 08:10 EDT, then graded their later bets: +6.8 pts ±3.2, n=177. All whales over the same later period: -2.8 pts. The sample is small, so treat it as suggestive.
2. Reading a flag
A flag is one $10k+ bet. Here's a recent one from a proven wallet that has resolved:
The wallet bought Grizzlies at 39¢. That price says the market gave it about a 39% chance. It came true, so the call counts as right. One bet proves nothing: a 39¢ bet should win about 39% of the time. Skill only shows across many bets.
3. Reading a wallet
Our most reliable wallet right now, BrotherObama, called 20 of 28 graded bets. Its entry prices implied about 14.1 wins. So its edge is +21.0 pts ±9.4.
- Called rate alone misleads: winning 90% of 95¢ bets is below par.
- Edge is wins minus what the prices implied. Positive means it beat the market.
- ± is the uncertainty. If edge minus twice that is still above zero, luck is an unlikely explanation.
4. Before you trade: who's on your side?
The largest open bet by a proven wallet this week:
If you were looking at this market, the useful question is: is a proven wallet on the side you like, against it, or absent? Open the market and the Who's on each side box shows proven money and money from consistently losing wallets on each side.
What this is not
It isn't a tip sheet. A proven wallet can be wrong, the price may have moved since it bet (see copy-lag findings), and track records fade. Use it as one input. Information, not advice.