How Many Trades Does It Take to Prove a Trading Edge?
Hundreds to thousands, depending on your edge and prices. The math of skill versus luck in prediction markets, with a table to check your own record.
By Lenium · Published · 4 min read
Short answer: far more than most people think. A trader with a real 5% edge, trading contracts around 50¢, needs about 1,100 trades before their record stops looking like it could be luck. A month of trading, a few dozen trades, tells you very little either way.
Why a good month proves little
Each yes/no contract either pays $1 or nothing, so every trade is a big, noisy bet. Bought at 50¢, a contract either doubles your money or loses all of it. That noise swamps small differences in skill.
Picture 1,000 traders with no edge at all, each making 30 trades at 50¢, where every trade is a coin flip. About 29% of them, 292 people, will win 17 or more of their 30 trades. That is an average return of at least +13% per trade, a record that looks like real skill and is produced by nothing but chance.
The math in one paragraph
Say you buy at price P and your edge is E: the average return per dollar you expect, after fees. A trade returns (1 ÷ P) − 1 if it wins and −1 if it loses. The spread of those returns is large, about ±100% per trade at 50¢, and the uncertainty in your average shrinks only with the square root of the number of trades. To be 95% confident a record isn't luck, the average has to sit about 1.645 standard errors above zero, which gives:
Trades needed ≈ (1.645 × spread per trade ÷ edge)²
Halve your edge and you need four times as many trades.
How many trades you need
Trades needed before a zero-edge trader would post a record as good as yours less than 1 time in 20, by the price you usually buy at and your edge after fees:
| Typical entry price | 2% edge | 5% edge | 10% edge |
|---|---|---|---|
| 20¢ | 27,500 | 4,500 | 1,200 |
| 35¢ | 12,700 | 2,100 | 520 |
| 50¢ | 6,800 | 1,100 | 270 |
| 65¢ | 3,600 | 560 | 130 |
| 80¢ | 1,600 | 230 | 45 |
At 35 trades a month, 270 trades takes about eight months; 1,100 trades takes more than two and a half years.
For a sense of scale, a 5% edge is a lot. In a study of 300,000+ prediction market contract prices, the average contract bought lost money after fees: about 10% for makers and 31% for takers.
What makes proof take longer
- Linked bets. Ten contracts on one election or one inflation number move together. For this math they count as closer to one trade than ten.
- Changing bet sizes. One large bet can dominate a record. The table assumes equal stakes.
- Changing strategy. Results from different strategies or market types don't add up into one record.
- Cheap contracts. At 20¢, each win pays four times your stake and each loss takes all of it. The noise is larger, so proof takes roughly 20 times as many trades as at 80¢ for the same edge.
What gets you there faster
- Trade where prices are less extreme. The same edge is proven in far fewer trades between 50¢ and 80¢ than at 20¢.
- Log every trade. Price, size, market, time and your reason. A spreadsheet you started a year ago is worth more than any single month's result.
- Compare your entry with later prices. If you consistently buy below where the market trades just before it closes, that shows an edge long before enough outcomes have settled to prove it. Sports bettors call this closing line value.
- Stick to markets you understand. A narrow, consistent strategy builds a clean record faster.
What this means for prop firm evaluations
A 30-day evaluation can't prove an edge, at Lenium or anywhere else. What it can do is screen out trading that relies on one big hit, which is why Lenium's rules use a 20¢ to 80¢ price band, caps on position size, and a consistency rule that stops any single market from supplying most of your profit. Treat a pass as a good start, not as proof, and keep building your record in the funded account.
How these numbers were worked out
The table assumes independent trades of equal size, each bought at the price shown, and uses a one-sided 95% test with a normal approximation, rounded. The counts mark where a trader with exactly that edge reaches the threshold on average; about half of such traders will still fall short at that point. The 29% figure for zero-edge traders is the exact binomial chance of winning at least 17 of 30 fair coin flips. Edge here means average return per dollar staked, after fees.
See how Lenium's evaluation works or read what a prediction market prop firm is.