Strategies·November 17, 2026·7 min read

Order-Book Imbalance on Polymarket: Reading Depth Instead of Price

How to measure bid and ask depth on Polymarket, when imbalance predicts short-term pressure, and why thin books, queue position and cancellable orders break the signal.

order bookmarket microstructureexecutionpolymarket
Stocks graph — illustrative image
Photo: 50Fish / StockSnap · CC0Source

Order-book imbalance is the difference between resting buy depth and resting sell depth, normalized by the total. On Polymarket it answers a different question from price. Price says what the market thinks will happen; imbalance says which side is currently paying to get filled. A 50-cent contract with $40,000 of bids and $4,000 of asks is not the same trade as a 50-cent contract with the reverse book, even though the screen shows the same number.

The useful version of the signal is narrow. Imbalance is a short-horizon execution and pressure measure, not a probability. It can tell you where the next few cents are likely to come from and whether your order will fill. It cannot tell you how the event resolves, and it is fragile because displayed orders can be cancelled. Treat it as a reason to look at the book, not a reason to bet.

Why the midpoint hides the book

Polymarket displays a midpoint unless the spread is wider than 10 cents, in which case it shows the last traded price. That convention is useful for a headline and misleading for execution. Two markets can both print 50 cents with completely different books.

Market A has 500 shares bid at 49 and 500 offered at 51. Market B has 20,000 shares bid at 49 and 800 offered at 51. The midpoint is 50 in both. A 2,000-share buy in A walks through the ask and beyond; the same order in B fills against 800 shares and then rests. The price display is identical. The trade is not.

This is why the first step is to replace the midpoint with four numbers: best bid, best ask, depth within one cent, and depth within three cents. Those numbers define what you can do.

Measuring imbalance without fooling yourself

The simplest measure is the top-of-book ratio: bid size divided by ask size. If 5,000 shares sit on the bid and 1,200 on the ask, the ratio is 4.2 to 1. It is easy to manipulate and easy to overread.

A better measure weights depth by distance from the touch: full weight at the best price, half one cent away, a quarter two cents away. Suppose the bid side is 5,000 at 48, 3,000 at 47 and 2,000 at 46. The weighted bid depth is 5,000 + 1,500 + 500 = 7,000. The ask side is 1,200 at 51, 800 at 52 and 400 at 53, so weighted ask depth is 1,200 + 400 + 100 = 1,700. The weighted ratio is 4.1 to 1.

Now convert that ratio into dollars, because shares at different prices are not the same exposure. The bid notional is $2,400 + $1,410 + $920 = $4,730. The ask notional is $612 + $416 + $212 = $1,240. The dollar ratio is 3.8 to 1. When the share and dollar ratios disagree, a large order at a very different price is distorting the share count.

Finally, compute the cost to trade your size. Buying 2,000 shares against that ask ladder costs 1,200 times $0.51 plus 800 times $0.52, or $612 + $416 = $1,028. The average price is 51.4 cents. The best ask was 51, so the slippage is 0.4 cents, or $8. Imbalance told you buyers were stacked, but it did not give you a free fill. You still paid above the touch.

What imbalance can and cannot predict

The academic case for order flow is stronger than the case for static depth. Cont, Kukanov and Stoikov studied the relationship between order book events and price changes and found that short-horizon price moves are driven mainly by order flow imbalance: the net effect of limit orders, market orders and cancellations at the best quotes. A static snapshot is weaker than the change in the book. A bid that has rested for an hour is different from a bid that appeared five seconds ago.

Chordia, Roll and Subrahmanyam documented that order imbalance is related to market returns and liquidity. Their work also makes the opposite point: imbalance can be informed. A seller hitting the bid may be selling because the price is about to fall. If you mechanically buy when the bid side is heavy, you are taking the other side of that information.

What imbalance cannot do is predict resolution. A book can be five to one bid-heavy for a week and the contract can still resolve No. The signal has a horizon of minutes to hours.

Thin books and fake depth

A ratio is meaningless without a depth floor. A 20-to-1 ratio on $300 of resting orders is noise. A 2-to-1 ratio on $300,000 is a statement. Before using any imbalance number, require total two-sided depth of at least five times your order, and preferably ten times. If the book cannot absorb your exit, the entry signal is irrelevant.

Displayed depth is not committed depth. A 20,000-share ask wall at 52 cents can be cancelled the moment price approaches it. That is not necessarily manipulation; it can be a market maker repricing. But it means you should not treat the wall as supply. Watch the lifetime of the order. If it disappears whenever the best bid ticks up, it was never a wall you could trade against. If it stays and trades through, it was real.

Queue position and fill probability

Polymarket matches by price first, then by time at the same price. If 5,000 shares are ahead of your bid at 49 cents and the average trade is 300 shares, you are roughly seventeenth in line. The book can look bid-heavy while your order never fills.

This cuts both ways. If you need to own the contract, cross the spread and pay for certainty. If you want to be paid for waiting, join a queue only when it is short and the imbalance is moving your way. A cancel and replace puts you at the back, so do not chase the touch.

How to execute

  1. Take three snapshots 30 seconds apart. Compute weighted imbalance and total depth for each.
  2. Require the sign to persist across all three. A single snapshot is a screenshot, not a signal.
  3. Require depth of at least ten times your intended order on both sides.
  4. Compute the all-in cost to buy and to sell your size, not the top-of-book price.
  5. Confirm with an independent view of fair value. Imbalance alone is an execution signal, not a reason to take a position.
  6. Enter with a limit order. If the signal is worth only one cent, do not cross three cents to act on it.
  7. Set the exit against the bid-side depth, not the midpoint. If the bid is thin, size down before you enter.
  8. Cancel if the imbalance flips or the large order that created it disappears.

The fee hurdle matters. At 50 cents in a politics market, a taker fee on 100 shares is $1.00, or one cent per share. A one-cent imbalance edge is exactly the fee. At 20 cents the fee is smaller in dollars but larger relative to the stake.

Where the signal fails

The first failure is informed flow. A sudden ask-side build can mean a large seller knows something. Fading it because the book looks too bearish is how a liquidity provider becomes exit liquidity.

The second is spoofing and fleeting orders. A displayed order is a quote, not a contract, until it trades. Build your signal on orders that have survived for at least a minute and on actual trades, not on the biggest number on the screen.

The third is an event. A debate, a data release or a court ruling can reset the book in seconds. Imbalance measured before the event is not information about the book after it.

The fourth is your own impact. A 5,000-share order changes the imbalance it is trying to exploit. Model the post-trade book.

The fifth is resolution. A contract with an ambiguous rule can trade on flow for weeks and then resolve on a sentence nobody read.

A short checklist

  1. Is the depth real, persistent and large enough to absorb my exit?
  2. Is the ratio meaningful in dollars, not just shares?
  3. Is the flow moving in the same direction as the static book?
  4. What does my own order do to the imbalance?
  5. What are the spread and fee hurdles at my size?
  6. Where is the bid I can actually sell into?
  7. What would make me cancel, and will I be fast enough to do it?

Imbalance is a better question than "is this cheap?" It is not a substitute for one. The book tells you who is paying for liquidity right now; the resolution rule and your probability estimate tell you whether you want to be on the other side.

Sources

  1. 1.Prices & Orderbook— Polymarket Documentation
  2. 2.Matching Engine— Polymarket Documentation
  3. 3.Order Imbalance, Liquidity, and Market Returns— Journal of Financial Economics
  4. 4.The Price Impact of Order Book Events— arXiv
  5. 5.Limit Order Books— arXiv

More from the journal

All articles