PnL vs Edge Score
Compare two orderings of the same closed wallet cohort. Neither column below is a validated skill ranking.
Closed set of wallet addresses selected once on 2026-04-19 and never added to, not a screen over Polymarket. Membership was conditioned partly on having made money.
Selection, coverage and frozen-screen methods
This board is a closed set of wallet addresses selected once on 2026-04-19, not a screen over Polymarket. No wallet has entered since. The original screen was: Polymarket wallets with >=5 resolved positions and >=$5k realized PnL (2026-04-15 snapshot). Ranked by Edge Score V3b. Membership was conditioned partly on having made money, which puts profit on the causal path into the pool, so no count taken over these rows describes Polymarket wallets generally.
Where a frozen screen reads "not separable from chance at these sample sizes", that is a statement about statistical power at those sample sizes. It is not a finding that the wallet lacks skill.
The frozen luck-share and skill-scan readings describe the dated cohort published 2026-06-09, read under the method in force before 2026-07-15 and its API read window; the current board's ordering and composition have since changed. The row screens retain their 2026-04-25 snapshot date.
Expand a wallet for its sample and price coverage, composite components, and realized-edge interval against zero where a frozen skill-versus-luck read exists.
Ranked from Polymarket's recent-trade window, not from a complete read of each wallet's resolved record. That window omits positions already redeemed, so these figures describe the window and not the whole record. No skill read is published for a wallet until its record has been read in full.
50-wallet cohort · Generated 2026-09-08 · Expand a row for details
Top 20 by realized PnL
Realized PnL in the recent-trade window
Recent-window record only. Not a complete read, and not a skill ordering. 9 of 20 rows below have been read in full.
rank . wallet . edge . pnl . edge rank . tap to expand
Recent-window record only. Not a complete read, and not a skill ordering.
- Cohort
- 50
- Generated
- 2026-09-08
- PnL top 20 also in composite top 20
- 7/20
- Composite top 20 outside PnL top 20
- 13/20
7 of the 50 cohort wallets are missing from BOTH rankings above. They failed to re-score on the latest run, so the board publishes no current realized PnL for them and their Edge Score is whatever the last successful run computed. They are not excluded for performing badly; they are excluded because we did not re-read them, and a ranking cannot place a wallet on a number nobody measured this run. Until 2026-08-24 they were ranked anyway, six of them inside the Edge Score top twenty, on scores between 47 and 77 days old.
What this artifact is saying
Across the full 8,656-wallet Polymarket cohort in our V1 methodology paper, the Spearman rank correlation between a wallet's Brier score (calibration) and its realized PnL is +0.148 (in-sample, descriptive). The top 1 percent of wallets by absolute PnL captured 36.2 percent of all signed profit on the platform. In a market that concentrated, a PnL leaderboard is heavily shaped by a small number of large outcomes.
Convexly's Edge Score is a composite of three behavioral pillars fitted against signed log PnL: posture, conviction, and discipline. The original V3b fit's in-sample out-of-fold Spearman with signed log PnL was +0.514, but that figure shares positions with the PnL it is scored against and its dominant input is PnL-derived, so it overstates predictive skill; the constants were also refit on 2026-07-13, so that fit statistic does not describe the current column (dated method-change note at /methodology). In the composite's one forward test, a per-wallet temporal holdout with the threshold set before the analysis ran, the frozen score held an out-of-sample Spearman of +0.11 (95% CI [0.05, 0.18]) with forward PnL. That did not clear the +0.30 bar. Treat the right-hand column as a descriptive behavioral profile, not a validated skill ranking.
The divergence you see above therefore shows that the two orderings disagree, not which one is right. Wallets in the left column but not the right have a PnL rank ahead of their composite rank; wallets in the right column but not the left have a composite rank ahead of their PnL rank. The honest read of any single record is whether its realized-edge 95% interval clears zero, which neither column displays. A wallet-by-wallet realized-edge scan of the published top-50 cohort is being prepared as a research note; see the top-50 skill scan in the research index.
The V1-M extension paper published 2026-04-22 adds a 15,106-user Manifold cohort and a paired-window sweepcash sensitivity analysis. The original public bundle reported median concentration 8.9 percentage points lower in the sweepcash window on 1,647 paired users, with concentration delta defined on n=333. A 2026-05-04 recovered-cohort rerun excluding political and election markets reports -9.2pp, 95% CI [-12.8, -3.6], Wilcoxon p=0.0021 on n=515 of 1,208 paired users with defined concentration delta. The concentration shift survives the sensitivity check, but causal attribution to sweepcash alone remains limited by the catalog confound.
Score any Polymarket wallet against this cohort
Paste any 0x address into the analyzer and see its three- pillar Edge Score in 15 to 30 seconds. Your first wallet is free with no signup; more are free with an account. No wallet signature, ever.
Methodology: Edge Score (Convexly Research 2026), a frozen-coefficient behavioral composite; the dated method-change note is at /methodology. The composite is a descriptive behavioral profile, not a validated skill ranking. The cron at 06:00 UTC rescores the existing rows; it does not refresh which wallets are in the cohort. Use this as a comparison view, not a wallet recommendation list.