How Stratex Hub scores Polymarket wallets and simulates copy trades
This page describes what the desk measures and how: the copy-trading research behind every number on the site. It is written so that a careful reader can find the weak points, because they exist.
What is being copied
Polymarket publishes every wallet's fills. The desk reads the 30-day leaderboard (top 500 by profit), pulls each wallet's fills, and decides whether a follower could realistically have taken the same trades. “Realistically” means: at a price the order book actually offered, a few minutes after the source wallet, at $5–$20 a position.
“Margin on volume” is not ROI
The public API gives each wallet's net profit and traded volume. It does not give deposits, withdrawals, or capital at risk. So the figure shown as margin on volume is net P&L ÷ traded volume over 30 days. A wallet that turns $200k to make $20k shows 10%. That is a margin on turnover. It is not a return on the capital the wallet actually risked, which is unknown. Rankings use it as one input among several; nothing on this site calls it ROI.
How a fill is simulated
When a followed wallet buys, the desk fetches the order book for that outcome token and walks the ask side from the best price up to the price band the rules allow. The desk and every follower of that wallet eat from the same book, desk first. The volume-weighted price actually available is the entry; if the depth runs out, the later followers get a partial fill or nothing, and the ledger records which. Fees use Polymarket's taker formula by category. A fill that would have needed a price the book never showed is not simulated.
A sell by the source wallet is an exit signal: if the desk holds that outcome from that wallet, it sells what it holds at the bid. A sell never opens a position and is never treated as a short.
Timing
Every fill records the source timestamp and the wall-clock moment this system first saw it. The gap is the ingest lag (typically 4–8 minutes). A buy first seen more than 45 minutes after it happened is journaled as stale and never copied. Marks are hourly at the current mid; settlement uses Polymarket's own closed flag, never a price guess.
The control arm
Every signal from a followed wallet is also copied blind — no judgement, fixed $10, same price band and the same book walk. That is the benchmark. If the desk's filtering does not beat it over time, the filtering is decoration, and the site shows that rather than hiding it. Comparisons are per position (average P&L per position), because the blind arm holds many more positions than the filtered one.
Sample size and what counts
Many fills in one market are one observation, not many. Settled markets are counted, not fills. The desk refuses to call its own results anything until 20 filtered copies have settled, and a wallet is not eligible for publication until at least 30 distinct markets have settled over at least 45 forward days.
Qualification policy (qual-v1)
- ≥ 30 settled markets copied on paper, over ≥ 45 forward days
- ≥ 80% of desk fills were full (not partial or unfilled)
- no single market contributes more than 20% of gross profit
- max drawdown ≤ 25% of capital deployed on that wallet
- median ingest lag ≤ 15 minutes
- filtered arm ahead of the blind arm per position
Meeting the policy proves the process produced a clean sample. It does not prove the wallet has skill, and a strategy page never claims it does. Publication is an operator's signature on the evidence; a published strategy is paused automatically if it later breaches the policy, if the desk stops tracking the wallet, or if the wallet goes quiet for three weeks.
What “follow” means
Following a published strategy mirrors the desk's future copies from that wallet into your paper account, sized to your bankroll and tiers, from the moment you follow. Nothing before that moment is attributed to you. Unfollowing stops new positions; open ones keep marking until they settle. No money moves. There is no execution path in this product, and a test in the repository fails if anyone adds one.
Known limitations
- Capital at risk per wallet is unknown; margin on volume is the best available proxy.
- Off-platform hedges and positions the wallet holds elsewhere are invisible.
- Paper fills assume the book seen at ingest; a real order a few seconds later may see less.
- Survivorship: the universe is the current top 500 by profit. Each daily scan is stored in full so this bias can be measured, but it cannot be removed.
- The track record is days old. Treat every number as provisional.
Commercial terms, briefly
Everything is included for 90 days from first sign-in, no card. After that, $188 for a year, paid once; nothing renews by itself. A lapsed account is read-only and keeps its history. Prices are in USD, before any applicable tax.
Questions people ask before signing in · See the live desk · Browse the wallet rankings