The Real Cost of Trading Polymarket: Fees, Slippage, and Walk-the-Book Math
Polymarket charges 2% taker. That's the easy part. The hard part is walking the book, paying the spread, and discovering your strategy needed a 65% win rate not 51% to break even.
Posted by
BenRelated reading
Polymarket vs Binance: What 30 Days of Live Arbitrage Audit Show
Logical arbs (YES+NO < $1) win 100% by construction. Endgame model signals win ~95%. Here's the 30-day public audit of every call we made — fills, fees, and realised PnL.
Polymarket BTC Up/Down 5-Min Markets: Walk-the-Book Reality Check
Most BTC up/down backtests assume mid-price fills. The real book is 11¢/89¢ for half the market's life. Here's what walk-the-book actually pays — and why your strategy probably doesn't survive contact with it.
Polymarket ETH Up/Down: How Often Does It Match Binance Spot Direction?
Resolved ETH Up/Down markets are an unbiased estimator of direction over their horizon — sometimes. Here's the actual match rate by event_type, and where Polymarket ETH books lead vs lag the spot move.
The official Polymarket taker fee is 2% of trade value. On a $0.50 fill that's 1¢ per share. So if you buy at $0.50 and the resolution pays $1, your net is $0.49 per share, right? On Polymarket, no — because almost nothing fills at $0.50. Here's the rest of the math.
Cost component 1: the taker fee
Polymarket charges 2% on trade value for taker orders. This is the explicit fee and the only number their UI displays. For a buy at $0.50 you pay ~1¢ per share in fees; for a buy at $0.85 you pay ~1.7¢. It scales with price, which matters because most of the time you're not buying near $0.50.
Cost component 2: the spread you pay
The market mid is the average of best bid and best ask. The price you actually pay is the ask. The difference — half the spread — is the spread cost. On BTC 5m markets with a typical 11¢/89¢ book, half-spread is 39¢ per share. Compared to the 1¢ fee, the spread is the actual cost.
Most amateur backtests price entries at the mid. That removes the spread cost entirely. You'd need a market-maker queue spot to actually get filled at the mid, and even then only when the queue runs out. For taker flow, plan to pay the ask.
Cost component 3: walk-the-book slippage
The best ask is only valid for the first n shares. If you want 50 shares and the best ask only has 10, the next 40 fill at deeper levels. On Polymarket's thin books this is often the dominant cost on positions > $20.
PolyQuantLab's backtest engine simulates this with a walk_book() function that walks the price ladder until the order is filled. Two backtests of the same strategy — one with fill_mode="mid" and one with fill_mode="walk_book" — will diverge by 10–40% in cumulative PnL on a typical strategy. The walk-book number is the one that predicts live performance.
Putting it together: the real break-even win rate
Take a 5-minute BTC market, 11¢/89¢ book, 10-share entry. You buy YES at $0.89, pay 1.78¢ fee, and walk-book slippage for the 10 shares is negligible because the inside ask easily covers it. Net cost: ~$0.91 per share for a $1 payoff if you're right.
Break-even win rate = 0.91 / 1 = 91%. That's the bar your strategy needs to clear, not 51%. For comparison, the mid-price simulation would have shown a 51% break-even and a beautiful equity curve.
What strategy designs survive this
- Logical arbs. By definition cost less than $1 walk-book to acquire $1 of payoff. 100% win rate by construction. Limited frequency. See the audit.
- Endgame model edges. Buy when our model says "88% chance of YES" and the market trades $0.65 = walk-book cost ~$0.69. Implied win rate needed is 69%; model says 88%; 19-point edge clears fees and slippage with margin.
- Maker quoting at the spread. If you have queue priority, you collect the spread instead of paying it. Phase Z of our backtest engine models queue position so you can see whether queue priority is plausible for your size.
What dies
- Anything that buys near the inside ask without a leading signal. The price you pay already incorporates the market's best estimate; your edge has to come from seeing what the market hasn't yet.
- High-frequency "arb" strategies that depend on the bid-ask compressing inside the last 60 seconds without you being inside the queue when it does. Without queue priority your fill rate will be much lower than backtest assumes.
- Mean-reversion in thin-book regimes. The thin two-sided quotes don't mean-revert at speed; they get cancelled and re-posted at a new level. You'll buy the bottom and then the book disappears.
The whole point of running a walk-book backtest is so the PnL number you see corresponds to what you'd actually realise. The whole point of running it with the fee model on is so you're not surprised by the difference between "gross" and "net" when you actually deploy capital.