is copy trading profitable?

the question is usually asked as if copy trading were one thing with one answer. it is closer to a formula, and the formula is short enough to write down.

your return = their return − latency − sizing error − fees − missed exits

every one of those subtractions is measurable. three of them are under your control. and their sum is regularly larger than the edge being copied, which is how someone follows a genuinely excellent trader into a losing year and concludes that copy trading does not work.

term 1: latency

you enter after they do, at a worse price. how much worse depends entirely on how fast the thing decays.

we measured this on one specific strategy: 29 first buys from a large fomo wallet. the median token was up 39% one minute after the leader's fill and 43% at fifteen minutes, then gave most of it back inside the hour. full method and caveats here.

against that curve, latency is not a minor drag. a follower arriving at minute thirty is buying near the top of a move that reverts. the trader's return and the follower's return are not similar numbers with a small gap between them — they are different signs.

on a strategy holding positions for days, the same delay is irrelevant. the entire question is the ratio of your delay to the edge's half-life.

term 2: sizing error

the largest term for most people, and the least discussed.

a trader risking 2% of a large account on a position that would be 40% of yours has handed you a completely different bet with the same ticker on it. copy proportionally and you have a rounding error; copy a flat ticket chosen while optimistic and one bad trade is your quarter.

the asymmetry is structural: their downside on a position is a bad week, yours might be the account. they do not know your size, they are not managing your risk, and they will not tell you when they change their mind.

term 3: fees, and the floor underneath them

the term people think they have accounted for and usually have not, because they price the rate rather than the minimum.

on fomo, the reported structure is around 0.50% with a minimum near $0.95. that minimum means any ticket under roughly $190 pays more than the headline rate — at $50 a ticket it is 1.9% a leg, 3.8% round trip.

now multiply by frequency, which you do not control when copying. four round trips a day at a $50 ticket is about $228 a month. against a $1,000 account that is nearly 23% a month in transaction costs alone. that strategy does not need a bad market to fail; a flat one will do it.

the fix is unglamorous and complete: bigger tickets, or a slower trader. the arithmetic in full.

term 4: missed exits

an enormous number of copy setups mirror entries and stop there, which is a defensible engineering decision and an indefensible product one.

on a move that peaks at fifteen minutes and reverts through the hour, the sell carries more of the outcome than the buy. automating the entry and leaving the exit to whether you happen to be looking at your phone converts a good trade into a coin flip. you have automated the exciting half and kept the hard half manual.

when it does work

put the four terms together and the conditions are specific rather than mysterious. copy trading tends to work when:

• your delay is small relative to the edge's half-life — either because you automated it, or because you chose a trader who holds for days
• your ticket clears the fee floor, so you pay a rate rather than a toll
• exits are copied, not watched
• size is yours, capped, and decided while calm
• the trader is a real record, not a spike — a survivorship check is free and most people skip it

and it tends to fail when a small account copies a fast trader at small tickets through a manual notification loop — which, unfortunately, is the default configuration almost everyone starts in.

what to actually expect

three things worth setting straight before you start:

the distribution is skewed. memecoin strategies produce long runs of small losses and occasional large wins. that means a month of red tells you almost nothing, and so does a month of green. judge over a trade count you set in advance, not over a feeling.

copying does not add edge, it removes friction. it fixes hesitation, missed alerts and the 4am problem. it cannot fix a trader with no edge, and it will execute their bad decisions faster and more reliably than you would have.

the leaderboard is selected on the outcome. it shows the accounts that won and structurally cannot show the ones that ran the same strategy into the ground. if a thousand people roll high-variance dice, the top of the board looks like genius either way.

the mechanical side — detection, sizing, execution, exits — is in how copy trading bots work and what breaks.