What is a concentration flag?
When a single event drives at least 60 percent of a wallet's net result, the record cannot be read as a clean edge, no matter how good the headline number looks. More positions do not fix it.
The answer first
The concentration flag is a demotion in Convexly's four-state skill read. It fires when the single biggest winning event's PnL divided by the wallet's net PnL is at or above 0.6: one event is at least 60 percent of the net result. A record like that is one outcome wearing a track record as a costume. Whatever the wallet's measured edge or profit rank, the figure cannot be separated from that one outcome, so the honest report is "too concentrated to read", not a skill verdict in either direction.
The ratio is deliberately not capped at 1. When the wallet's other events netted negative, the single biggest win can exceed the entire net result: observed values like 2.7 are real. A ratio above 1 means the rest of the record lost money on net and one event carried everything.
Worked example: Edge Score 95.8, unreadable record
A real row from the frozen 2026-06-09 scan of our own published top-50 cohort (full table at /research/top50-skill-scan). Wallet 0x0d83ca sat near the top of the board with an Edge Score of 95.8 and positive net PnL. Its skill read:
- Realized entry edge +1.9pp across 35 resolved positions, 95 percent interval [-7.8, +11.9].
- Concentration ratio 2.70: the single biggest winning event was 2.7 times the net result, meaning everything else netted negative.
The flag fires and overrides everything above it. A reader who saw only the board position would call this wallet elite; the resolved record says its result is one event. In the same scan, 15 of the 35 testable wallets were flagged the same way, which is a large part of why raw profit leaderboards mislead.
Why more positions do not fix it
The insufficient state (fewer than 30 resolved positions) is a sample-size problem: resolve more positions and the read becomes possible. The concentration flag is a composition problem: the record IS the one event, and adding small positions around it does not change what made the number. A flagged wallet whose dominant event ages out of the record, or whose subsequent record grows large enough that no single event is 60 percent of the net, can lose the flag, but volume alone does not clear it. Concentration can also mask gamed activity, which is another reason the flag is read as a hard demotion rather than a footnote.
What the flag does NOT mean
Flagged is not an accusation and not a verdict of no skill. A genuinely skilled trader who concentrated once, correctly, will read as flagged until the rest of the record can stand on its own. The flag states a fact about composition: this figure cannot be separated from one outcome. The frozen wording lives in the lexicon, and the threshold is documented on the methodology page.
Check a wallet's concentration
Paste any Polymarket wallet address at the analyzer to see its concentration read alongside the realized entry edge, interval, and four-state verdict. Free, no signup, public on-chain data only.
Convexly publishes new methodology research roughly every 6-8 weeks plus the /learn series on a rolling cadence. Get the next paper in your inbox when it ships:
Frequently asked
What is a concentration flag?
How is the concentration ratio computed?
Why does the flag override a good point estimate?
Does more trading fix a concentration flag?
How common is the flag among top wallets?
Related explainers
- /learn/realized-edge: the statistic the flag protects from misreading
- /learn/luck-share: how the flagged count feeds the cohort-level reading
- /learn/edge-score: why concentration is rewarded in the descriptive composite yet demoted in the skill read
Related reading
LearnBrier score
LearnCalibration
LearnFalse discovery rate
LearnNegative control