which fomo traders are worth copying?

the fomo leaderboard sorts by realised profit and loss. it is a good number: it comes from filled trades rather than posted calls, it is public, and it is very hard to fake. it is also a ranking of things that have already finished happening, and you are trying to make a decision about things that have not.

why profit is the wrong first sort

two problems, one obvious and one not.

the obvious one is survivorship. a public ranking of realised pnl is a list of people who won. it is not a list of people who were right, because it does not contain the accounts that made the same bets, lost, and stopped. if a thousand people run a high-variance strategy, the top of the leaderboard will look like genius regardless of whether the strategy has any edge at all. the board cannot distinguish those cases and neither can you, from the board alone.

the non-obvious one is that profit does not predict price impact. we measured 29 first buys from one large fomo wallet and found the median token up 39% one minute after the fill and 43% at fifteen. then we split the sample by how much of that minute's volume the trader's own order actually was:

bigger orders, smaller moves. that is backwards for a price-impact story and exactly right for a cascade one: the trader's order is the trigger, and the volume that moves the price is their followers arriving through the notification. the full study, method and limits are here, and the sample is small enough that you should hold the caveat in your hand while you read the rest of this.

if it holds, the consequence is blunt. the variable is audience size. a trader with half a million followers and average returns moves a token harder than a trader with forty thousand followers and excellent ones.

reading the two columns together

look at the top of the board with both numbers side by side and the interesting part is where they disagree.

rank 9 has the largest audience on the page by a wide margin — roughly 35% more followers than rank 1, on a third of the profit. rank 8 has produced comparable profit to rank 7 with about a quarter of the following, which under the cascade reading is the more impressive record and the weaker copy target. rank 10 is on the board on a following of under forty thousand, which is either genuine edge or a small sample that has not been tested yet, and the board will not tell you which.

none of this makes any of them good or bad to copy. it makes them different bets, and knowing which bet you are placing is the point.

six filters that survive contact with reality

1. sample size, not headline profit

how many closed trades produced that number? forty trades and four hundred trades are different claims, and the leaderboard displays them identically. a record built on a handful of enormous wins is a record about one market condition.

2. did it survive a bad month?

anyone can look like a genius in a rising tape. the informative window is the worst stretch in the history: what happened to the equity curve, how long the drawdown ran, and whether the trader's behaviour changed while it was happening.

3. hold time compatible with your life

a trader who scalps in and out inside twenty minutes generates a stream of copies, a stream of gas, and a stream of taxable events. a trader who holds for days generates a handful. neither is better in the abstract. one of them is much worse if you are copying manually, and both are fine if you are not.

4. position sizes you can mirror without distortion

if a trader's typical entry is $50,000 and your ticket is $100, you are mirroring their decisions at 0.2% scale. that is fine for the entry. it is worse for the exit, because their partial sells will be smaller than your minimum trade and will resolve into all-or-nothing on your side.

5. first buys versus adds

the effect we measured lives in first buys — the first time a wallet touches a token. a trader whose flow is mostly adds to long-held positions is a fundamentally different copy target from one who is constantly opening new names, regardless of what their pnl says.

6. followers, weighed deliberately

last, because it is the one everyone is now going to over-apply. a large following means a bigger wave behind the entries you copy, and a bigger crowd trying to leave through the same door when it turns. it raises the ceiling and the floor together.

red flags that end the search

• a record that is one trade. open the history. if 80% of the pnl comes from a single position, you are copying a person who was once very right.
• size that grew faster than the record. a trader who went from $500 tickets to $50,000 tickets in a month is running a different strategy than the one that produced the number you are looking at.
• entries that only look good after the notification. if the tokens are consistently thin enough that the follower wave is the liquidity, you are not copying a trade, you are joining a queue and hoping you are near the front of it.
• no exits in the history. unrealised gains are a screenshot. a leaderboard that ranks realised pnl mostly protects you from this, but check anyway.

when to stop copying someone

set this rule before you start, because you will not set it fairly afterwards. two triggers are enough:

1. the behaviour changed. hold times collapse, sizes triple, the token profile shifts. the record you selected on no longer describes the trader you are copying.
2. your own drawdown limit. a number you wrote down while calm, in your currency, that stops new copies when it is hit. not a feeling about whether they have "still got it".

and the frame to keep the whole exercise honest, from the trader with the largest following on the board above:

"Never blindly follow anyone's trades - including mine." — PoorGoat

he is on our leaderboard and we did not remove him for saying it. copying is a way of executing your decision faster. it is not a way of not having made one.

once you have picked: how to actually copy them, and how to get the wallet address you will need to do it automatically.