Tick Scalping Wide Spreads on Polymarket Without Becoming Exit Liquidity
When a Polymarket spread is wide enough to scalp, how maker and taker fees change the math, and the inventory and adverse-selection rules that keep one bad fill from erasing a week of gains.

Tick scalping is the business of buying at the bid and selling at the ask for a one- or two-tick gain. On Polymarket a wide spread looks like an invitation: 44 cents bid, 50 cents ask is a 6-cent gap, 12.8% of the midpoint. It is also the price the market charges for taking the other side of someone who may know more. The spread is not the edge until you have paid for inventory risk, adverse selection and fees.
The question is not whether the spread is wide. It is whether the spread is wider than the cost of the risk you are about to hold. The exceptions are narrow: genuine two-sided interest, a distant catalyst, and a book deep enough that your exit does not move the price.
When a spread is actually wide enough
Measure the spread in cents and as a percentage of the midpoint. A 2-cent spread is 4% at 50 cents and 40% at 5 cents. The percentage is what matters for a scalper, because the position is marked in cents but the return is on the stake.
Then compare the spread with the round-trip cost of crossing it. At 50 cents in a politics market, a taker round trip on 1,000 shares pays $20, or 2 cents per share. That is the hurdle a taker faces. A 6-cent spread offers 4 cents of room only to the trader who captures it as a maker; the taker who crosses both sides pays the spread instead of earning it. In sports at 0.05 the round trip is 2.5 cents; in crypto at 0.07 it is 3.5, so a 3-cent spread is negative before the price moves.
If you provide liquidity instead of crossing, the fee is zero on entry and the gross capture is the distance between your bid and your ask. Posting at 45 and 49 captures 4 cents. If the expected adverse move is 1.5 cents and the exit fee is 1 cent, the net is 1.5 cents. A 3-cent capture is not enough. The catch is that a maker fill happens when someone chooses to trade against you, and that seller may be reacting to information.
Maker versus taker
The maker version pays no fee and may earn a rebate. Polymarket's maker rebate is 25% of the fee equivalent in politics, finance, economics, culture and weather, and 15% in sports. It is a share of the taker fee your resting order generated, calculated per market and paid daily. If the market has no taker fee, there is no rebate.
The taker version is immediate and certain. You pay the spread and the fee, but you are not waiting in a queue while the price moves away. Taker scalping only works when the gap covers both costs and the expected move. On a 6-cent spread, crossing at 50 and selling at 44 is a 6-cent loss before fees. The taker version is to buy the ask when you expect the bid to rise, not to harvest the spread.
Queue position decides whether the maker version works. If 5,000 shares sit ahead of your bid at 45, a 300-share seller will not reach you. Improving to 46 shrinks the gross spread to 4 cents and tightens the break-even. Post-only orders guarantee you do not accidentally cross, not that you fill.
The inventory problem
Scalping is two-sided in theory and one-sided in practice. You post a bid at 45 and an ask at 49. The bid fills first. Now you are long 1,000 shares at 45, marked at 47, waiting for the ask. If the market moves to 42, your unrealized loss is $30.
A concrete PnL: buy 1,000 at 45, cost $450. Sell 1,000 at 47 as a taker. Proceeds $470. The exit fee is 1,000 times 0.04 times 0.47 times 0.53 = $9.96. Net profit is $10.04 on $450, or 2.2%. That required a 2-cent move and paid a fee on exit. If you had waited for 49 and the price never got there, the same position could have been marked at 40.
Inventory limits are part of the strategy. A useful limit is three times your normal clip. If your normal clip is 500 shares, do not hold 2,000 because the book looks friendly. A second limit is time: if the other side has not filled in 30 minutes, exit at the bid or hedge. A third is event risk: flatten before scheduled news.
Adverse selection
The reason a wide spread exists is often that the people posting tight quotes have been picked off. Glosten and Milgrom showed that the spread compensates a market maker for trading against informed order flow. Kyle showed that informed traders split their orders, so the maker loses gradually and earns from noise traders. Your resting bid is more likely to fill when the seller has information, which is exactly when you do not want it to fill.
The practical measurement is markout. After every fill, record the midpoint one minute, five minutes and thirty minutes later. If your average five-minute markout is negative, your fills are adversely selected. Suppose 100 fills produce an average markout of minus 0.8 cents at five minutes. A 2-cent gross spread becomes 1.2 cents before fees. If markout is minus 1.5 cents, the strategy is dead.
Fee math and the break-even spread
The fee formula is C times rate times p times (1 - p). At 50 cents, 1,000 shares cost $10 in politics, $12.50 in sports and $17.50 in crypto. For a maker the fee is zero, so the break-even spread is the expected adverse move plus the exit cost. For a taker it is the entry fee, the exit fee and the adverse move. At 50 cents in politics that is 2 cents plus the adverse move. A 25% rebate is worth 0.25 cents on a 1-cent fee and does not rescue a negative markout.
Stopping rules
A scalping strategy needs hard stops because the failure mode is a slow accumulation of inventory and a fast loss.
- Inventory stop: no more than three clips on one side.
- Loss stop: stop for the day after losses equal twice your average gross spread per trade.
- Spread stop: stop quoting when the spread falls below twice your round-trip cost.
- News stop: no new quotes in the 15 minutes before a scheduled release.
- Time stop: exit or hedge after 30 to 60 minutes with no offsetting fill.
- Markout stop: after three consecutive fills with negative five-minute markout, stop and review. This is a model-break check, not a prediction.
The loss stop matters most. A scalper who makes 2 cents per share on 1,000 shares makes $20 per round trip. A single 20-cent adverse move on the same 1,000 shares loses $200, or ten good trades. The distribution is negatively skewed even with positive expectancy, so the stop keeps one bad day from ending the account.
How to execute
- Screen for spreads above 4 cents and two-sided depth above ten times your clip.
- Compute fair value independently. If the midpoint is 47 and your fair value is 44, the spread is wide because the bid is right.
- Post only, one tick inside the touch when the queue is short.
- Use GTD orders that expire before known catalysts. A stale quote is a free option for the other side.
- Track fill rate, average markout and inventory age.
- Exit with FAK when inventory exceeds the limit. Partial fills are better than waiting for a perfect bid.
- Reconcile rebates separately from trading PnL. A strategy that is flat before rebates is a rebate harvest with tail risk.
Risks and limits
A wide spread can be a warning: the market is thin, the event is close, or one side has information. It can also mean the rule is disputed. Scalping concentrates those risks in a short holding period.
Finally, resolution risk sits underneath every scalp. A contract that can resolve 50/50 or be voided is not a 2-cent trading instrument. Read the rule before you quote it.
A short checklist
- Is the spread wider than my round-trip cost plus expected adverse move?
- Am I maker or taker on entry, and what does the exit cost?
- How many shares are ahead of me in the queue?
- What is my maximum inventory and time limit?
- What is my markout on this market, and on this category?
- Is there a scheduled event inside my holding period?
- Do I stop at my loss limit without renegotiating it?
Tick scalping is a liquidity business. The spread is revenue, adverse selection is cost, and inventory is the risk that connects them. If the revenue does not cover the cost after a realistic markout, the wide spread is not an opportunity. It is a quote from someone who wants you to take the other side.
Sources
- 1.Fees— Polymarket Documentation
- 2.Order Lifecycle— Polymarket Documentation
- 3.Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders— Journal of Financial Economics
- 4.Continuous Auctions and Insider Trading— Econometrica
- 5.Optimal Market Making— arXiv