The profit on a fake smart money wallet is usually real. That's what makes it dangerous. The chain shows the wallet put in 1 BNB and took out 20, and nothing about that number is forged. What's fake is the idea that it came from skill you can follow.
This page goes through the four kinds that crowd the top of every top traders list, the on-chain test that exposes each one, and what's left when you're done. It's step 4 of tracking smart money, done properly.
A wallet is fake smart money when its edge can't transfer to you. Either it knew something before the market did, it got in at a speed no follower can match, it traded with itself, or its public record was edited by moving trades between wallets.
None of these break the PnL maths. They break the conclusion. Copying an insider means buying the bags they are about to sell, and copying a bundle means buying an entry that ended in the launch block.
Look at how the tokens got into the wallet. Open its token transfers on the block explorer and filter to the token. A real trader's tokens come from the pool in a swap. An insider's often come straight from the deployer, a presale contract or another wallet, with nothing paid on the pool.
Tokens with no cost on the pool make a wallet look like a genius. Sell 20M free tokens for 8 BNB and the profit column says +8 BNB on zero spent.
The other insider tell is the funding trail. Click the wallet's first incoming transaction. If the native coin came from the deployer, or from a wallet the deployer funded, in the hours before launch, it's the team. Our deployer check walks that trail one hop back. An insider's "win" is the dev dump from a slow rug, seen from the seller's side.
Check the block it bought in. Find the block where the pool got its liquidity, then the block of the wallet's first buy. Same block or the next one means a sniper or a bundle. No human clicking a buy button gets there.
A bundle is several wallets buying together, so the tells come in groups:
A bundle also hides its own size. Profit split over five wallets puts none of them at the top, so the wallets you see are often the smaller pieces. How much supply a bundle hides, and how it sells it, is in what is a bundled launch.
Check who paid for the wallet's exits. Every sell has a buyer on the other side, buying from the pool in the blocks around it. If those buyers are wallets funded by the trader or by the same funder, the "win" was money moving in a circle.
Wash records cluster on tokens with very few unique traders. A wallet with a perfect run on ten tokens that each had a few dozen holders and no outside volume is building a record, not trading a market. It's usually done to look good on exactly the ranking you're reading. The same trades also show up as volume on a pair chart, which counts a wash buy like any other. MoonRadar vs DexScreener covers what chart volume can and can't tell you.
Look for tokens moved in right before a win. A trader who wants a clean public wallet buys with one wallet, sends the winners to the showcase wallet, and sells from there. The showcase never bought, so every sale counts as nearly pure profit, while the losers stay in wallets nobody follows.
The second staged pattern starts once a wallet is famous. Compare its hold times before and after it started getting shared. A wallet that used to hold for hours and now sells within minutes of its own buy, into the volume of people copying it, has turned into bait. Its followers are now its exit, which is why copy trading safely starts with an exit plan, not the wallet. Callers who post a wallet as proof of their record pull the same trick, so check the call ledger, not the wallet.
Take a made-up top traders list on a BNB token, five wallets deep:
The tell people miss: the only wallet worth a second look was fifth, behind four that would have tracked straight into the team's exit. Even wallet 5 is a lead to watch, not a buy signal.
A funding trail that runs through an exchange breaks the link: withdrawals usually come from the exchange's hot wallet, so you can't tie two wallets together that way. A patient insider who uses an aged wallet with a normal history will pass most of the tests.
The checks also misfire the other way. Some honest traders run fast bots and do land early. Treat a failed test as a reason not to copy, not as proof of fraud.
Paste a token into the MoonRadar scanner to rank every wallet that traded it, then paste any of them into the wallet finder to see every token it traded with per-token PnL. That's test 5 done for you. The Top traders board ranks realised dollars, not win rate, so a wallet can sit in a top tier and still lose more trades than it wins. The board marks those with a warning sign.
The bundle and sniper case gets caught earlier on coins MoonRadar calls. The launch cohort is clustered: same-block and same-size buys are grouped and labelled as one signal, and a coin where one operator controls a large share of supply at launch is hidden rather than called.
Before you put money behind a wallet that passes, follow it with paper copy trading, which mirrors its buys and sells from live on-chain data so you see what riding it would return. Every token still goes through the safety screen, because a real trader can still buy a honeypot. Past profit is a record, not a promise, and most memecoins fail.
Rank a token's traders on MoonRadar