Independent market-price ratings.
We tell you which prediction-market prices are safe to use in your work.
Convexly reviews public Polymarket and Kalshi data, then gives a plain rating: use, use with caveats, discount, or do not cite. Every rating keeps the caveat and receipt attached.
Secondary proof: check one public Polymarket wallet.
Or try a wallet evidence demo
What do you want to decide?
Start with the thing you need to check.
Convexly turns public prediction-market records into a short answer, a caveat, and a receipt you can inspect.
Is this trader actually skilled?
Check whether profit came from repeatable behavior, posture, conviction, and discipline instead of one lucky PnL spike.
Check a walletCan I cite this market price?
Use Market Trust to see whether a price is usable, caveated, discounted, or too unreliable to cite.
Audit a marketWhere do linked markets disagree?
Use Coherence Signals to find probability-axiom breaks that survived significance, cost, and resolution-risk gates.
Inspect signalsCan I trust Convexly itself?
Open the track record to see live evidence, pending gates, failed tests, and claim caveats before relying on the product.
Verify receiptsEvidence discipline
Trust the workflow, not the screenshot.
Methods freeze before promotion. Failed tests stay public. Every serious claim points back to the evidence ledger.
Why this exists
Prediction-market prices are easy to quote and hard to diligence.
Convexly is building the missing diligence layer: wallet-skill benchmarks, market-quality cards, and reproducible signal receipts. The evidence comes from public market records, source-controlled methods, and audit-chain receipts, so serious users can inspect what produced a rating instead of accepting a black box.
Coherence Signals · Daily
90 cross-market clusters passed the current version-labeled signal filter.
$67k modeled net EV at prescribed size across surviving candidates·Modeled, not realized PnL. Settlements and matched controls decide what we can claim later.·reproducible methodology
$99/mo·Researcher tier·hash-chain methodology receipts·cancel anytime
Check any Polymarket wallet in 10 seconds
Wallets in research cohort
8,656
Polymarket, 5+ resolved positions
Resolved positions analyzed
582,921
frozen-coefficient methodology
Wallet rankings
Top 50
≥20 resolved positions, refreshed daily
Methodology, with receipts
Scores and signals come with inspectable receipts.
Coefficients freeze before publication. Forward extensions get filed before analysis runs (filing index + per-test verdicts). Failed and null results stay visible.
Frozen-coefficient wallet skill ranking, calibration-weighted.
A wallet that is right when it claims confidence ranks higher than one that is right at random. Versions: V1 (2026-04-18) · V1.5 (negative-result paper: both primary tests failed at their ex-ante thresholds; published openly) · V1-M (cross-venue extension, 2026-04-22).
Structural-mispricing detector across linked Polymarket events.
Flags places where prices on related events violate coherence (mutually-exclusive events summing to more than 100%, etc.). Public product label: Coherence Signals. CME here means Coherent Markets Engine, not CME Group. Current public state: V0.1 paper + version-labeled daily signal archives. V0.2 spec is locked and under evidence collection; the next evidence lane is coverage-matched orderbook depth, matched controls, and resolved-signal reporting.
Operator notes for prediction-market traders
Wallet Skill Brief #1: Most "whales" are not skilled
Across the Top-50 Polymarket cohort, the rank-correlation between Edge Score and realized PnL is Spearman ρ = −0.32 (p = 0.026, n = 50). One Polymarket wallet has $78K realized PnL and ranks 45/50 on Edge Score. The dollar leaderboard and the skill leaderboard disagree systematically.
What the discipline inversion means if you actually trade
Operator note on the updated Convexly V1-M finding: the recovered politics-excluded rerun reports -9.2pp lower concentration (95% CI [-12.8, -3.6], p = 0.0021). Causal language and the calibration secondary claim are superseded.
Stop Optimizing Calibration. Start Optimizing Conviction. What 10,000 Polymarket Wallets Tell You To Do.
Five operational rules for Polymarket traders, translated directly from the 8,656-wallet cohort. Calibration is table stakes. Concentration and survival math drive profit.
For institutions
Evidence-linked artifacts for serious diligence.
Wallet verification, Market Trust cards, and API access package the same receipts behind the public product.
Sales-led, human-reviewed PDF audit on Convexly Research letterhead.
A scoped diligence artifact for teams reviewing a public Polymarket trader, market maker, or wallet-labeled workflow.
Talk to salesMarket-level ratings for whether odds are citable, caveated, discounted, or not citeable.
Coherence, liquidity, depth, participant quality, resolution reliability, manipulation-risk flags, and audit-chain evidence in one source-linked card. v0.2 is evidence-gated and explicitly experimental while the validation ledger matures.
View Market TrustProgrammatic access to the Edge Score and Coherence Signals endpoints.
JSON-over-HTTP for approved public and pilot surfaces. Hashes and source labels are returned where the underlying artifact is reproducible, so buyers can separate public receipts from internal archives.
Read the docsPnL is mostly luck. Skill is measured differently.
Across 8,656 Polymarket wallets, calibration (Brier) explains only +0.148 Spearman correlation with realized PnL. The top 1% by profit captured 36.2% of all signed gains. Convexly separates repeatable posture, conviction, and discipline from lucky screenshots; the three-pillar composite reaches out-of-fold Spearman +0.514(Fama-French bootstrap null p < 0.0001 on 10,000 permutations).
V1-M paper (cross-venue invariance)