Is that volume squeaky clean?
Toss in a token address. We tumble its newest ~450 trades and sniff out fake volume: wallets flipping the same size in and out, traders who churn without keeping a bag, pairs swapping opposite sides, and one wallet bankrolling a crowd. Each stain links straight to the transactions.
Reading the care label
Five wash cycles add up to a 0–100 score. It is a nudge to look closer, not a ruling: market makers and arbitrage bots churn too, so open the evidence before deciding anything.
Round tripsup to 35 points
A wallet buys, then within five minutes sells roughly the same amount (within 10%), or the other way round, in separate blocks. Same-block pairs are skipped since those are usually MEV bots. Points = 70 × the share of sampled volume caught in these loops.
Net-flat churnup to 20 points
Wallets with at least 4 trades and 1%+ of sampled volume whose buys and sells cancel out within 10%: plenty of activity, no bag left. Points = 40 × their share of volume.
Mirror tradesup to 20 points
Two separate wallets take opposite sides of the same size (within 1%) no more than 2 blocks apart, at least twice. Points = 60 × the mirrored share of volume.
Crowding & bankrollersup to 10 + 15 points
Crowding counts when the top 5 wallets carry more than 30% of volume. Bankrollers: we look up the first SOL sent to each flagged or top wallet, using Jupiter's holder data or the wallet's earliest transaction (only if it has under 1,000). Points = 40 × the volume share of wallets with a common first sender. Exchanges seed lots of wallets, so check who the sender is.
Jupiter's own 24h "organic volume" figure sits next to the score for comparison. It is calculated their way, not ours.