Partial fills and leg risk on Polymarket

Partial fill leg risk happens when one leg of an intra-market arbitrage executes and the complementary leg does not. This guide explains the mechanics on Polymarket's CLOB, common scenarios you will see as an arbitrageur, the concrete risks (resolution, slippage, fees, settlement timing), and step-by-step mitigations you can apply in live trading.

Key takeaways

• Partial fill leg risk arises from asynchronous fills on a CLOB and is a function of order type, liquidity, and tick size.
• Use order types and sizing (including FAK market orders) to minimise one-sided exposure; expect slippage and taker fees on fills.
• If a single leg fills, treat the position as a short-lived directional exposure: assess resolution risk, fees, available liquidity for the offset, and whether to split/merge outcome tokens.
• Pre-define a recovery workflow (hedge, squeeze, or unwind) and instrument dry-run checks into your bot.

Why this matters

Arbitrage opportunities on Polymarket are often time-limited and micro-sized. When you submit orders for complementary outcomes — e.g., YES and NO on a binary market, or a complete set on a multi-outcome market — you assume both fills will occur nearly simultaneously. A partial execution turns a balanced, edge-defined trade into an open exposure. Knowing the mechanics and having a fixed response plan preserves edge and limits surprise losses.

How partial fills occur on Polymarket

• Order types and matching: Polymarket's CLOB accepts limit and market orders. The platform exposes a helper that issues FAK market orders by default for immediate execution. FAK executes immediately to the available liquidity and cancels any remainder, so a large FAK can partially fill if book depth is shallow.
• Tick size & extremes: Tick size is usually $0.01 and tightens to $0.001 when prices approach extremes. Tick size step changes can cause quantisation effects that make even small remaining quantities unfillable at your requested price.
• Latency and order matching: Matching is synchronous in the CLOB, but network and relay latencies, concurrent taker flow, and order book updates can result in one order matching while the opposite side cancels or is outmatched.

Common partial-fill scenarios

1. Binary two-leg arb: You place buy orders for YES and NO expecting the sum of best asks < $1.00. YES fills fully; NO partially fills or misses because the best ask moved. You now hold YES exposure.
2. Multi-outcome set: You buy N-1 outcomes and fail to fill the final leg. You're long a subset of outcome tokens in a market where a complete set should have been neutral.
3. Market order out-of-depth: You use a FAK (market) order to hit immediacy; the available depth only fills the cheaper leg, leaving remainder cancelled.

Concrete risks to enumerate (always list these)

• Resolution risk: UMA disputes or delayed resolution can extend the life of your one-sided exposure.
• Slippage & partial execution: The unfilled leg may be unfillable at reasonable prices; chasing it increases cost and may eliminate edge.
• Fees: Taker fees apply on fills (maker fees are zero). Fee changes or category-specific fees affect realised profit.
• Smart-contract & settlement timing: CTF split/merge/redeem timing and on-chain operations (even though gas is sponsored by the Relayer) can affect when you can neutralise holdings.
• Counterparty flow: Other traders may detect your one-sided positions and move prices against you while you attempt to hedge.

Immediate triage checklist (what to do in the first 30–120 seconds)

1. Confirm fills via the Data API or CLOB read endpoints. Do not assume both orders matched.
2. Calculate worst-case P&L including taker fees and potential slippage required to complete the complement.
3. Query available liquidity at better/worse prices to determine if you can buy the missing leg within acceptable slippage.
4. If liquidity is insufficient, decide whether to hedge with correlated markets (rare on Polymarket) or to reduce exposure by selling part of the completed leg.

Mitigations and workflows

• Prefer FAK for immediacy when you accept partial fills, and design sizing to match typical depth. See /glossary/fak for a definition.
• Size orders conservatively against the visible depth on both sides of the book. If bestAsk(YES) + bestAsk(NO) < $1.00 only by a thin amount, reduce leg sizes to the available depth at those price levels.
• Stagger order placement: submit the thinner-liquidity leg first, then the other. This reduces the chance of one-sided execution in many practical book shapes.
• Use split/merge workflows: on multi-outcome partial fills, consider merging the outcomes you hold into a smaller set where available, or transferring tokens to a safe counterparty if you plan a covered salvage trade. Remember CTF mechanics: split mints complete sets from pUSD and merge returns pUSD by burning complementary tokens.
• Pre-program an automatic recovery: either aggressively chase the missing leg up to a price cap, or unwind the filled leg immediately to accept a partial loss. Have hard limits and do not escalate without human sign-off for large exposures.

Practical example: binary two-leg attempt

• You spot bestAsk(YES) = 0.48, bestAsk(NO) = 0.49. Sum = 0.97, edge = 0.03.
• Book depth: YES has 100 shares at 0.48, NO has 20 shares at 0.49.
• You post 50-share FAK buys on both legs. YES fully fills (50), NO fills only 20 and cancels remainder.
• Post-fill options: buy additional NO up to your cap (accept slippage), sell 50 YES back into the book to neutralise (realising spread & fees), or hold YES and wait — each choice has trade-offs listed above.

Fees and accounting

• Makers pay zero; taker fees vary by category up to ~1.8% per taker fill. Include expected taker fees when you compute break-evens for recovery trades.
• Record per-fill fees and timestamps. Settlement timing matters: redeemed outcome tokens pay $1.00 each after UMA resolution and any dispute window.

How this affects your trading

Design every arbitrage execution with a built-in partial-fill plan. Size trades to visible depth, prefer FAK only when you accept partial fills, and implement an automated post-fill triage that queries fills, recomputes net exposure including fees, and executes one of three outcomes: complete the complement within a price cap, unwind the filled leg, or hold to resolution only when expected value justifies the risks. Doing so turns one-off surprises into routine, codified responses and preserves your edge.

Further reading

• Full system context: /guides/polymarket-arbitrage-complete-guide
• Binary-specific tactics: /guides/intra-market-binary-arbitrage-explained
• Multi-outcome workflows: /guides/combinatorial-arbitrage-on-polymarket
• CLOB mechanics: /guides/polymarket-clob-explained

Partial fill leg risk is an execution reality on any exchange. On Polymarket, the combination of tick size, book depth, and FAK behaviour makes it predictable — which means you can build robust, repeatable responses into your bot and risk-management processes.

Frequently asked questions

What is the simplest response when one leg fills and the other does not?

The simplest immediate response is to recompute net exposure including taker fees, check available liquidity for the missing leg, and either complete the missing leg within a pre-set price cap or unwind the filled leg by selling it back. Your choice should be governed by pre-defined risk limits and maximum acceptable slippage.

Do FAK orders eliminate partial fills?

No. FAK (Fill-And-Kill) increases immediacy but can still partially fill if available depth is limited. FAK fills up to existing liquidity and cancels any remainder, so you can still be left with one-sided exposure.

How should I size orders to reduce leg risk arbitrage problems?

Size each leg against the visible depth at the price levels you plan to hit. Use smaller increments than the largest available depth, or stagger leg submission so the scarcer leg is posted first. Always include a margin for maker/taker behaviour and tick-size quantisation.

Can I rely on splitting/merging outcome tokens to fix partial fills?

Split/merge are CTF operations that let you convert pUSD into complete sets and vice versa. They don't magically complete an unfilled market leg — they help you manage multi-outcome exposures once you hold tokens. Use them as part of a recovery plan but not as a substitute for hedging the missing leg.

What are the main risks I must list when documenting trade outcomes?

Always list resolution (UMA) disputes and timing risk, slippage and partial execution risk, applicable taker fees, and smart-contract/settlement timing. These factors materially affect the realised result of any partial-fill situation.