A wallet with a 90 percent win rate looks like the best pick on the list. In practice it is usually the worst one to copy. Here is what makes a win rate that high, and what to read next to it.
The best wallet we scanned on 16 September 2026 had a 70.8 percent win rate over 30 days, 125 tokens up to 2x, 6 between 2x and 5x, none above, and 50 tokens in the loss buckets. It scored 82 of 100. That is what a healthy record looks like: wrong more than a quarter of the time, closed its losers, made its money on the middle of the distribution [1].
What win rate measures
Win rate is the share of tokens a wallet closed in profit over a period, usually 30 days. Sell a token above your average buy and it counts as a win, whether you made 2 percent or 20x. Sell it below and it counts as a loss. Tokens still held do not count at all.
That last sentence is where most of the trouble is.
Three ways to get to 90 percent
Never close a loser
Sell every token that goes up, even by 3 percent. Keep every token that goes down. The closed trades are all wins. The losses sit in the open positions where win rate does not look.
On a scan card this shows as a high win rate with a large negative unrealized result, and an outcome distribution made of small wins: many tokens in the up to 2x bucket, few or none above, and the losses in the open bags. Ocinct takes 10 points off when open losses exceed half of the realized profit, and shows both numbers on the card so you can see the shape.
Know before everyone else
Insiders and dev team wallets do not pick tokens, they are given them. A wallet that trades tokens deployed by its own funder, or bought in the launch bundle, wins because the trade was decided before the token existed. Its win rate is real and completely useless to you: by the time you see the buy, the part of the trade that was safe is over.
The funding trail is the check here. One or two hops from the trading wallet to a deployer or a relay is enough.
Be first
Snipers and sandwich bots win most of their trades because they are in the block before anyone can react. Copying them means landing after them and selling after them. Their 90 percent can become well under 50 for you.
These wallets are usually tagged on a scan, and their average hold is measured in seconds. See how to tell if a wallet is a bot.
What a healthy win rate looks like
On Solana memecoins, a wallet that takes real risk and closes its losers lands somewhere between 45 and 70 percent. That range means the wallet is wrong often, admits it, and makes its money on the size of its wins rather than their count.
This is why the Ocinct score stops rewarding win rate at 70 percent: zero points at 30, the full 25 at 70, nothing more above. Past that line, more wins have stopped meaning better trading.
What to read next to it
Win rate only makes sense with three other numbers on the same card:
- Return on money spent. Realized profit divided by what was bought. A 90 percent win rate with a 4 percent return is a wallet scalping crumbs. A 55 percent win rate with a 120 percent return is a wallet that lets winners run.
- Outcome buckets. How many tokens lost more than half, lost, gained up to 2x, 2x to 5x, above 5x. This is the win rate broken open. Two wallets at 60 percent can be one with a few 10x and many small losses, and one with sixty 1.1x and a bag of zeros.
- Open profit or loss. If it is large and negative, the win rate is being propped up by positions that were never closed.
Read win rate or PnL, which to trust for the same comparison from the profit side.
The rule
A 90 percent win rate is not a reason to copy a wallet. It is a reason to open the card and find out which of the three things above produced it. Most of the time you will close the card.
Sources
[1] Ocinct scan card, 16 September 2026: https://ocinct.com/how-it-works
[2] Ocinct score method, win rate capped at 70 percent: https://ocinct.com/how-it-works