Every week someone posts a screenshot: 99.3% across 32,614 trades, $313 turned into $438K. And every week the top comment says the same thing — none of them can prove their track record is real. So we went and checked.
First, a correction, because it is the useful part
The obvious way to read a wallet's history stops dead. Ask the API for activity record 5,001 and it answers:
max historical activity offset of 5000 exceeded
On 29 August we published that this made long track records
structurally unverifiable. That was wrong. Within hours,
fedoras in the Polymarket builders group pointed at the documentation:
the 5,000 limit is per query window, not per wallet. Page
inside start/end windows and each window gets its own
offset budget. Confirmed the same day — 10,000 records off a wallet that
had stopped dead at 5,000.
So this page now reads history by splitting time adaptively: any window that saturates is halved, down to the hour. The lesson was worth more than the bug. A limit you could not get past is not the same as a limit that exists. We had exhausted our own method, not every method, and published an impossibility on the strength of it.
What remains true is narrower and more interesting: the data is reachable, but it takes deliberate work, so almost nobody checks. The busiest wallets run about 2,500 records a day; reading one in full is tens of thousands of requests. This page caps each wallet at 30,000 records and says so in the table rather than pretending to a complete audit.
The measurement
| # | Wallet | Claimed PnL ($) | Records read | Net withdrawn ($) | Verdict | Open |
|---|---|---|---|---|---|---|
| 1 | swisstony | 23,628,090 | 30,000 | 6,000,400 | partial read (30,000 cap) | 2673 |
| 2 | Theo4 | 22,053,934 | 16,166 | 22,031,581 | corroborated (0.1%) | 16 |
| 3 | Fredi9999 | 16,619,507 | 23,878 | 16,583,563 | corroborated (0.2%) | 52 |
| 4 | RN1 | 13,054,646 | 30,000 | 0 | partial read (30,000 cap) | 3252 |
| 5 | kch123 | 11,386,691 | 30,000 | 1,982,185 | partial read (30,000 cap) | 337 |
| 6 | mintblade | 9,238,345 | 830 | 9,138,397 | corroborated (1.1%) | 0 |
| 7 | fishalive | 9,063,378 | 1,724 | 8,498,156 | corroborated (6.2%) | 0 |
| 8 | frostrizz | 8,928,561 | 2,089 | 8,833,468 | corroborated (1.1%) | 1 |
| 9 | Len9311238 | 8,709,973 | 5,810 | 8,705,499 | corroborated (0.1%) | 2 |
| 10 | sparklingwater123 | 8,474,966 | 1,139 | 8,350,906 | corroborated (1.5%) | 1 |
And now the part nobody expects
Of the 7 wallets read end to end, 7 match their claimed PnL within 10% — most of them within about one percent.
Withdrawals cannot be faked. A displayed PnL is a line in an interface; money that left an account had to exist in order to leave. So we take withdrawals minus deposits — net money out, on chain — and compare it to the number the leaderboard advertises. They agree closely.
The running assumption in every thread is that the top of this leaderboard is fabricated. On this evidence it is not. The gap is not honesty, it is effort: the proof is sitting on chain and essentially nobody goes and gets it.
Which turns the usual suspicion upside down. Be sceptical not of the eight-figure wallets that withdraw millions, but of the small screenshots — $20 into $180 in a week — where nothing was ever withdrawn and every gain still sits in open positions.
Two traps in reading any of these claims
Deposits are the denominator. “$313 into $438K” means nothing if the wallet also received $500K in deposits. The deposit is public; almost nobody quotes it. On our own account the same error ran the other way: we compared value against cumulative capital deployed instead of deposits, and believed we were down 25% while we were up 41%.
Realised gains and win rates are upper bounds, never results. You close winners — there is a buyer — and you keep losers, because there is no counterparty. Losses therefore pile up in open positions and are never counted. That asymmetry alone is enough to manufacture a 100% win rate on a losing account.
Check it yourself
The tool is open source and read-only — no key, no authentication, no orders. It refuses to state a total while positions are open or while events it cannot account for remain, and it says which:
python tools/verifier_portefeuille.py 0xWALLET --annonce 438000
Want this run on a specific wallet, or the same treatment on a set of markets? Rates and how to reach me.