🛡️ Fortress
Consensus favorites — the income floor
89.1% hit
95% CI 77.0–95.3 · N=46
- edge vs market +6.3 pts
- ROI per bet +7.1%
- ≥2 elite resolvers · 75–96¢
Every other tracker shows you whale wallets and inflated win rates. polymax publishes the number they can't: a survivorship-bias-free, out-of-sample backtest against ground-truth blockchain resolutions — losses included, confidence intervals shown.
Out-of-sample hit rates against actual resolved outcomes — trader quality measured on the first half of the window, strategies replayed only on the second. When a number is statistically fragile, we say so on the card.
We found the industry's dirty secret while building this: tape-only reconstructions can't see losses — a losing position leaves no redemption record, so it silently vanishes and hit rates inflate to 90%+. Ours did too, until we fixed it.
Why this matters: a copy-trading product is only as good as the number it's built on. Leaderboard PnL measures the whale's returns, not yours. Zero-slippage paper studies measure a fantasy. polymax measures what a copier holding to ground-truth resolution would actually have caught — and then keeps proving it forward: the paper autopilot settles every pick against real outcomes, building a live track record before a single real dollar moves.
Based on a June 2026 survey of every major Polymarket smart-money tool.
| Published backtest | Losses included | Out-of-sample + CIs | Kelly portfolio sizing | Paper-first autopilot | Price | |
|---|---|---|---|---|---|---|
| polymax | ✓ full methodology | ✓ | ✓ | ✓ | ✓ default | free |
| Premium copy bots ($299–499/mo) | ✗ gated or none | ✗ | ✗ | partial | ✗ | $299–499/mo |
| Mid-tier copy apps (~$30/mo) | partial, no window/slippage | unclear | ✗ | ✗ | ✗ | ~$30/mo |
| Free whale trackers | ✗ | ✗ | ✗ | ✗ | ✗ | free |
Polymarket runs on Polygon, so every trader's positions are public. Copy trading means mirroring what the most profitable wallets hold. The hard part isn't the data — it's selection and sizing: knowing which traders' positions predict outcomes, and how much to stake. That's what polymax automates: a gated elite pool, three backtested pattern detectors, and fractional-Kelly sizing.
Don't infer outcomes from the trade tape. Losing positions never redeem, so tape-only reconstructions silently drop losses. Score every historical entry against the market's actual resolved outcome (closed markets + outcome prices), include every loss, and split your data so strategy rules are chosen on one period and tested on another. That's the polymax methodology, published in full.
Kelly sizes each bet by its edge: the bigger your measured advantage over the market price, the more you stake — but never more than the math justifies. polymax uses fractional Kelly (a quarter of full Kelly, capped at 10% of bankroll) fed by each strategy's out-of-sample edge. In our backtest the Kelly-sized portfolio compounded ×1.77 while flat staking on identical bets managed ×1.25.
Honestly: the edge is real but smaller than the industry advertises, and it decays with slippage and entry delay. Our numbers (+6 to +13 points of edge over market-implied probability) survive out-of-sample testing, but only one strategy's interval is statistically watertight so far. That's why polymax runs paper-first — it has to prove the edge in your own ledger, against real outcomes, before real money moves.
No. polymax is a research and simulation tool. Past edge is not a guarantee of future returns, prediction markets carry real risk, and access to Polymarket trading is geo-restricted in many jurisdictions — compliance is on you.
Free for everyone while the forward track record builds. Open your paper desk in one tap — no signup, no card.
⚡ Start your free desk