Smart wallets

"Smart money" is the most abused phrase in this industry. Almost every tool that shows it is showing you the same thing: wallets that were early on tokens that went up. That sounds like skill, and usually it isn't.

The denominator problem

Here is a wallet we found in our own data:

0x8366…70b4    early on 15 tokens that ran    ← looks excellent
               bought 443 of 1,041 launches   ← bought 43% of everything

Early on fifteen winners. Also early on four hundred and twenty-eight things that went nowhere. Buy enough lottery tickets and you will hold winners; that is arithmetic, not insight.

Every label on this page carries a denominator. We don't ask "how many winners did this wallet catch?" — we ask "out of everything it bought, what fraction worked?" and compare that to what a random wallet would have achieved over the same period.

The labels

💰 KP Wallet — Known Profitable

Our strictest gate. A wallet must clear all five criteria over a rolling 90-day window:

Criterion	Threshold
Unique tokens traded	≥ 20
Win rate on closed positions	≥ 65%
Realized profit, net	≥ $25,000
Average return per trade	≥ 150%
Entry timing	within the first 20% of the run to peak

Ninety days is deliberate. A lifetime window lets a wallet that crushed it two years ago qualify forever. A thirty-day window punishes a good trader for one bad month. Ninety days is long enough to prove consistency and short enough that wallets going cold rotate out on their own.

Profit is calculated FIFO — first tokens bought are first sold — with positions under $50 and holds under 60 seconds excluded, so dust tests and same-block arbitrage don't inflate a record.

🎯 Early Caller

Finding things before they're visible, repeatedly:

• Among the first 50 buyers of a token
• That token went ≥10× from their entry
• On ≥3 different tokens within a rolling 30 days
• Entry of at least $100 — filters wallets that dust-test everything
• Win rate ≥50% on closed positions

That last line is the denominator. Without it, "early on three winners" is a statement about how many tokens someone bought, not about judgement.

The labels that protect you

These are the ones no leaderboard shows, and on our data they're the difference between a real signal and a mirage.

Label	What it means	Why it matters
🔗 Cluster	This wallet makes identical trades to one or more others	We found 18 wallets holding exactly the same 4 tokens. That's one decision-maker, not eighteen confirmations
⚠️ Ubiquitous	Buys more than 10% of all new launches	Infrastructure or a bot. Its presence tells you nothing about this token
🚫 Rug deployer	Deployed a token that rugged	Excluded from every positive label, permanently
🚫 Bundle operator	Participated in bundled launches on 2+ tokens	Launch manipulation, not trading
🌱 Fresh wallet	Too little history to judge	Kept and flagged — never scored as good or bad
💤 Dormant	No activity in 90 days	A track record with an expiry date

Honest limits

• A track record is not a forecast. These wallets are wrong constantly. The label says what happened, not what will.
• We only see on-chain. A wallet that also trades on an exchange has a record we can't fully compute, and we'd rather say so than guess.
• Coverage differs per chain. Where we don't hold enough trading history to compute a wallet's record, the label is absent — not "average". Absence means unmeasured, never mediocre.