Prediction market liquidity is the depth of the order book at and around the current price — how many contracts you can actually buy or sell before your own order starts moving the price against you. It caps your profit directly: a 3% spread sitting on top of $200 of available depth is worth about $6, no matter how good your detection is. On Polymarket and Kalshi order books, liquidity is usually the real ceiling on an arbitrage strategy, not the size of the spread.
This matters more here than on a stock exchange, because most events trade on a single order book rather than a continuous stream of market makers. Thin books are the norm outside a handful of high-volume categories, and a system that only checks the top-of-book price without checking depth will report a profit the book cannot actually deliver.
What "depth" means in a prediction market order book
A Kalshi or Polymarket order book lists resting orders at each price level on both sides. Depth at a price is the number of contracts available at that price or better. If the best YES ask is 52 cents for 40 contracts and the next level is 53 cents for 120 more, the quoted price is 52 cents — but the depth at 52 cents is only 40 contracts. The 41st contract costs more.
For a two-leg arbitrage you need depth on both legs at the same time, large enough for the position you want to take. The binding constraint is whichever leg is thinner: a $5,000 Polymarket book paired with a $300 Kalshi book caps the trade at $300, not $5,000.
Why liquidity caps profit more than the spread does
A wide spread with no depth behind it is not a large opportunity with a small percentage attached — it's a small opportunity with a large percentage attached. Two spreads that look identical on a dashboard, both quoted at 4%, can differ by an order of magnitude in realizable profit depending on how many contracts actually sit behind that price. Ranking opportunities by spread size alone, without depth, has no way to tell the difference.
Where depth tends to be thin
Depth is not evenly distributed across a platform's markets — a handful of high-volume categories carry most of it, and everything else trades on comparatively shallow books. Four situations show up repeatedly:
- Long-tail events outside US politics, Fed decisions and major crypto price levels, where a matched pair exists on both platforms but trading volume is low on one or both sides.
- Markets far from resolution, before most capital has committed a view.
- Lopsided coverage: one platform carrying far more volume than the other for the same event — the thinner side sets the ceiling even if the other book looks deep.
- Momentary gaps right after a news event, before market makers have had time to add depth back to the book.
How to size a position around liquidity, not around your bankroll
For pure arbitrage, the Kelly criterion simplifies: since both legs are expected to fill, the probability of winning is effectively 100%, so position size isn't limited by edge or variance — it's limited by available depth, platform position limits, and leg-fill risk. In practice, size to the minimum of three numbers: the depth at your price on leg one, the depth at your price on leg two, and whatever cap your own risk rules impose. Then recompute the net edge at that size, not at the top-of-book price — filling past the first level means walking the book and averaging a worse price than the one you started with.
Slippage: what happens past the top of the book
Buying past the first level of depth means your average fill price moves against you as each successive level fills — this is slippage, and it eats directly into the spread you calculated. A 4% spread measured from top-of-book prices can shrink to 2% or less once the position is large enough to walk two or three levels deep on either leg. The fix is mechanical, not clever: check depth before sizing, not after, and treat the quoted price as valid only up to the quantity actually resting there.
How Arbitrage Agent handles this
Arbitrage Agent checks order-book depth on both legs before flagging an opportunity, not just the top-of-book price, and sizes a position at the thinner of the two legs rather than quoting at a price the book can't support at scale. It's also why the live numbers on /proof describe a median net edge around 0.7% per position — the honest number after fees and after sizing to what the book can actually fill, not the best-case spread on a chart. See the plans for how detection and alerts are packaged.
Key risks to keep in proportion
Leg risk remains the main operational risk: if one side fills and the other doesn't before the price moves, you're left holding a directional position instead of a hedged pair. Thin liquidity makes this worse, since a partial fill on a thin book is more likely than on a deep one, and exiting a partial position on an illiquid market can itself be costly — you may need to sell into the same shallow book that caused the problem in the first place. Resolution risk is separate from liquidity risk and should be sized for independently — the two compound rather than cancel out. Size for the depth you can verify, not the depth you hope is there.
Related articles
- Sports Arbitrage on Prediction Markets: Kalshi vs Polymarket
- Kalshi vs Polymarket: Fees, Liquidity & Which Is Better for Arbitrage (2026)
- Prediction Market Arbitrage Strategy: The Complete 2026 Guide
- How Much Can You Make with Prediction Market Arbitrage?
- How Prediction Market Arbitrage Works — Complete Guide 2026
Key terms: Orderbook · Liquidity Pool · Kelly Criterion