pump.fun wallet win rate: how does your wallet or bot rank?
PnL trackers tell you whether a wallet is up or down. They do not tell you the more useful thing: where that wallet stands against everyone else trading the same tokens. This guide explains which numbers describe a pump.fun wallet properly, how to compute them yourself, and how the free BotSharper wallet report places any public Solana address, whether it is your own trading wallet or the wallet your bot trades from.
Why win rate alone misleads
Win rate is the share of closed positions that ended above their entry. It is easy to compute and easy to misread:
- A high win rate can hide a few large losses. Ten small wins and one position that goes to zero can still leave the wallet down.
- A perfect win rate is often a warning, not a skill. As one analysis of Solana leaderboards puts it, they are "full of wallets with 100% winrates over 10+ trades", which is implausible for a human and often points to bots or insider structures.
- Few trades are not comparable. BotSharper's Method page makes the point directly: "An operator with three trades is not comparable to one with three thousand."
So you need several indicators read together, and a population to compare against.
The indicators worth computing
| Indicator | What it describes |
|---|---|
| Closed positions | How many positions were opened and then fully closed. This is the sample size. |
| Share of winning positions | The win rate, over closed positions only. |
| Average multiple | What came back per unit put in, on average. Below x1 means less came back than went in. |
| Share of total losses | How often a position was lost almost entirely. This is the number win rate hides. |
How to compute them yourself
- Export the wallet's trades. Block explorers such as Solscan list a wallet's trading activity, and many PnL tools export it.
- Group by token. For each token, add up SOL spent on buys and SOL received from sells.
- Keep only closed positions. A position still open has no final result yet. Leave it out.
- Compute per position: the multiple (SOL received divided by SOL spent), whether it was a win (multiple above 1), and whether it was a near-total loss.
- Aggregate. Count positions, then compute the share of wins, the average multiple and the share of total losses.
This gives you your own numbers. It does not tell you whether they are good, because you have nothing to compare them with. That is the hard part.
Why a benchmark changes the answer
To know where a wallet stands, you have to compare it with a large population measured the same way. BotSharper's Method page describes six steps built so they "can be proven wrong", including:
- public on-chain data only, "no user-reported figures";
- a minimum number of closed positions before any ranking;
- a benchmark "computed with that operator left out", because "without this precaution, everyone beats their own average";
- three separate periods: the rank is set on one and checked on the following ones;
- a negative control "with ranks drawn at random" on every rebuild;
- a full rebuild every night, blocked by seven automatic checks if anything is inconsistent.
What the BotSharper wallet report shows
Paste a public address on the report page. The home page states the ranking is computed across 463 434 real operators and 13.0 million closed positions. According to the site, the address you type never leaves your browser. The report gives:
- a percentile among all operators, written as "You are ahead of X% of the operators measured";
- the fifth (quintile) the address belongs to, from weakest to strongest;
- closed positions, share of winning positions, average multiple and share of total losses;
- for each indicator, the benchmark of its fifth, to see where it sits inside its band.
For bot operators, the same report works on the address your bot trades from. The free Telegram bot @BotSharperBot takes an address in a message. To query reports automatically from your own tools, Pro access gives a personal key (0.1 SOL for 30 days, according to the Pro page).
The part most tools leave out
The BotSharper home page states: "No group of operators, not even the best fifth, clears break-even over the period measured." The limits page adds: "Ranking well means losing less than the others, not winning." The rank exists and holds over time, but the site is explicit that "the measured edge comes from execution, not from picking assets. Copying someone's moves does not copy their rank."
Known limits
- The rank describes the past. "It predicts nothing, and we do not claim it carries forward."
- One person may hold several addresses, and one address may be run by a program or a group.
- A missing address is not an inactive address: it may be below the minimum number of positions, or trade outside the markets covered.
The full list is on What we don't know. The same data also exists per token on the token check page.
FAQ
What is a good win rate on pump.fun?
There is no fixed threshold. A win rate only means something next to the number of closed positions, the average multiple and the share of total losses, compared with other wallets measured the same way.
Can I check my trading bot's wallet with BotSharper?
Yes. Paste the public address your bot trades from. It is ranked if it has enough closed positions in the period measured. Never paste a private key or recovery phrase: the site never asks for one.
Does a top-fifth rank mean my strategy makes money?
No. BotSharper states that no group of operators, not even the best fifth, clears break-even over the period measured. A good rank means losing less than the others.