A leaderboard is the easiest information to obtain on this venue and the easiest to be misled by. It carries three layers of lag: in its statistics, in its qualification, and — most damagingly — the fact that a strategy appearing on it is itself evidence that its window is closing. I copied one once, for about $210.
Working on weather, I found a wallet performing well at the time, took its approach apart, and built a line copying it. Result: about −$210, archived.
The real cause emerged in review, and it is worse than "I executed it badly":
So it was not that I copied it wrong — I copied something that had already expired. And a leaderboard will never tell you that; it displays the past.
| Layer | Source | Consequence |
|---|---|---|
| ① Statistical | the board totals what already happened | at best it says "last week was good" |
| ② Qualification | you need accumulated results to rank | by the time it ranks, some of the good period is spent |
| ③ Mechanical | a visible edge is a thinning edge | you seeing it ≈ others seeing it |
The third is fundamental. Most edges here come from a temporary imbalance — a new category nobody works, a reward rule just launched, a cohort still learning to price. Those windows close. And a strategy earning enough to rank is itself evidence that people are already doing it.
Even accepting the lag, what it shows you is partial:
| Trap | Consequence |
|---|---|
| "Profit" excludes subsidies | subsidy-driven players appear to be losing |
| Lifetime ÷ current holdings | a retired whale becomes a "small-capital genius" |
| Capped at 50 places | paging parameters are ignored entirely; the tail is invisible |
| The activity feed is row-capped | a busy wallet may be covered for only tens of hours |
The second caught me once: dividing lifetime profit by current holdings to get "capital efficiency", with a wallet in the sample at $4.55M lifetime and $5,700 currently held. Dividing made it a god, the estimate blew up, and I withdrew it after publication. All four traps are in eight API traps.
It is not useless — it is not a list of people to copy. Three legitimate uses:
| Use | How |
|---|---|
| Find specimens, not role models | use it to locate wallets running your approach, then take them apart on-chain yourself |
| Read the population, not the top | the median of same-shape wallets is your expectation |
| Read the ceiling | what the best earns tells you whether the path is worth entering |
I measured the second once, with a persuasive result: in one category 47 same-shape wallets had a median of −0.74¢/share — the top of the board looks impressive while the population is losing money. Read only the top and you form exactly the opposite impression.
| # | Question | Why |
|---|---|---|
| 1 | Is this wallet still making money? | read recent records, not the lifetime figure |
| 2 | Is it earning from trading or subsidies? | query both and add them |
| 3 | Do others running the same approach earn? | one winner alone may be luck |
| 4 | Do the conditions it relied on still exist? | rules change, competitors arrive, the window closes |
I asked none of these, at a cost of $210. It is the cheapest tuition on this site, and the rules it bought are written here.
A leaderboard shows the past, and you are investing in the future. Worse: the fact that a strategy is visible to you is itself evidence its window is closing. So use it to find specimens, read medians and estimate ceilings — not to find role models.