A prediction-market trader sitting on a profitable position shortly before an election faces a deceptively difficult question: take the available price now, or wait for the result and pursue the remaining upside?
The question is usually framed as a forecasting problem. If a contract trades at 70 cents, is the true probability higher or lower than 70%? But an exit decision has a second dimension that receives much less attention: will the market still be liquid when the trader wants to sell?
A close look at Wisconsin’s 2026 Democratic gubernatorial primary suggests that this execution question can be decisive. Before the votes arrived, traders could transact against a deep and tightly quoted order book. Once counting began, near-market depth collapsed and spreads widened. Liquidity returned only after David Crowley’s victory was effectively known, when traders no longer needed it to manage the event risk.
A Natural Experiment in Election-Night Liquidity
The market examined here was the Kalshi contract KXGOVWINOMD-26-DCRO, which paid $1 if Milwaukee County Executive David Crowley won the Democratic nomination for governor of Wisconsin. Crowley was a long shot entering election day but ultimately defeated Francesca Hong in a razor-thin upset.
Wisconsin Public Radio reported that the Associated Press called the race at 2:34 a.m. Central on August 12. The analysis uses Predexon’s sub-cent historical order-book feed from 8:00 p.m. on August 10, exactly 24 hours before polls closed, through 2:45 a.m. on August 12.
The dataset contains 27,848 full order-book snapshots. To avoid overweighting periods with especially frequent quote updates, the time-series analysis retains the last observed book in each minute. Because the article asks when a trader can sell a YES position, liquidity is measured using the bid-ask spread and the displayed YES bids available within 1 cent and 5 cents of the best bid. These sell-side bands approximate the quantity a long trader could exit without moving far from the prevailing price.
| Moment | Midprice | Spread | YES Bid Depth Within 1 Cent | YES Bid Depth Within 5 Cents |
|---|---|---|---|---|
| 24 hours before polls closed | 5.05 cents | 0.10 cents | 29,854 | 338,477 |
| Polls closed/counting began | 4.50 cents | 1.00 cents | 2,200 | 56,555 |
| AP race call | 99.55 cents | 0.10 cents | 272,085 | 276,832 |
The Market Barely Moved. Liquidity Did.
Twenty-four hours before polls closed, Crowley YES traded around 5.05 cents. The spread was only 0.10 cents, while buyers displayed 29,854 contracts within 1 cent of the best bid and 338,477 contracts within 5 cents.
At the moment polls closed, the midpoint was still only 4.50 cents. In other words, the market’s assessment of Crowley’s chance had barely changed. The execution environment, however, had changed dramatically. The spread widened tenfold to 1 cent. YES bid depth within 1 cent fell 93% to 2,200 contracts, and bid depth within 5 cents fell 83% to 56,555 contracts.
READ THE LATEST PREDICTION MARKET NEWS
The deterioration persisted during counting. Median 1-cent bid depth during the count was 8,667 contracts, roughly 43% below the 24-hour pre-count median. Median 5-cent bid depth fell from 312,693 contracts before counting to 20,118 during counting, a 94% decline.
The most severe moment arrived around 8:34 p.m. Central, when Crowley’s midpoint was near 55 cents. Only 102 YES bid contracts were displayed within 1 cent of the best bid, and the total remained 102 even across a 5-cent band. The spread reached 6 cents. A trader who had waited for the dramatic repricing could be directionally correct and still face a poor exit: cross a wide spread, accept substantial slippage, or leave a limit order exposed while the result continued to change.
By the AP call, the midpoint had reached 99.55 cents, the spread had tightened to 0.10 cents, and YES bid depth within 1 cent had surged above 272,000 contracts. That apparent improvement needs careful interpretation. It was settlement-like liquidity near a near-certain outcome, not evidence that the market had remained continuously liquid as information arrived.
Why Order Books Thin When Information Begins to Arrive
This pattern is consistent with a central result in financial market microstructure: liquidity providers are most vulnerable when the trader on the other side may know more than they do.
In the classic model developed by Lawrence Glosten and Paul Milgrom, bid-ask spreads compensate market makers for the risk of trading against better-informed participants. When the probability of informed order flow rises, rational liquidity providers widen their quotes or reduce the size they are willing to display. Glosten and Milgrom (1985).
Albert Kyle’s framework reaches a related conclusion from the perspective of price impact: market depth is endogenous to the amount of uninformed trading and the information advantage embedded in order flow. When information is arriving quickly, the same order can move the price much more than it would in an ordinary period. Kyle (1985).
Election-night prediction markets fit these models unusually well. Once votes begin to be reported, some participants have faster election feeds, better county-level models, local reporting, or automated systems that can reprice results before a manual trader reacts. A resting limit order becomes a free option for the better-informed trader: execute against it when it is stale and ignore it when it is not. The natural response is to cancel the order, quote a smaller size, or demand a wider spread.
This also explains why headline liquidity figures can mislead. At polls close, the Crowley book still displayed more than 1.1 million contracts in total, but a YES seller could see only 2,200 bid contracts within 1 cent of the best bid. Very large orders near 0 or 100 cents may make a book look deep without offering meaningful capacity at the price a trader actually wants.
The Case for Selling Before the Event
A practical decision rule: Wait only when the expected value of the remaining information edge exceeds the expected cost of wider spreads, greater slippage, and the risk that an exit will not fill.
Selling before the event can therefore be rational even when a trader believes the contract has additional upside. The trader is not simply giving up expected value; the trader is purchasing execution certainty.
Your Edge Has Already Played Out
A position may have been entered because polling, endorsements, fundraising, or early-vote data were mispriced. If the market has already moved toward that thesis, holding through the event converts a successful pre-event trade into a new bet on the final outcome. The second bet may have a worse risk-return profile and a worse exit environment.
Your Position Is Large Relative to Usable Depth
A trader holding 20,000 contracts might appear small relative to a million-contract book. But if only 2,200 YES bids are displayed within 1 cent, or 102 during a shock, the position is large relative to the liquidity that matters. Selling in advance can prevent the trader from becoming the market’s largest urgent seller at the worst possible moment.
You Need Certainty of Execution
A limit order can protect the price but not guarantee a fill. A market order can guarantee immediacy but not the execution price. Before the event, traders often have time to work an order, split it into smaller pieces, and wait for passive fills. Once results arrive, that flexibility shrinks.
A Partial Exit Can Be the Cleanest Compromise
Selling enough before the catalyst to recover principal or lock in a target return can reduce the urgency of any later decision. The remaining position can then be held through the event without depending on election-night liquidity.
When Holding Is Still Rational
The data do not imply that every trader should sell before every event. A small position held to settlement does not require exit liquidity. A trader with a genuine informational advantage during the count may benefit from precisely the volatility that causes others to retreat. Some markets may also attract additional market makers as the event approaches, resulting in better liquidity rather than worse.
The Crowley market is one case study, not a universal law. It was an unusually volatile, upset-driven primary in which the favorite changed rapidly and the final margin was narrow. The results should be treated as evidence that liquidity can be state-dependent and can disappear during information shocks, rather than as a precise estimate for all prediction markets.
A Better Way to Plan an Exit
Prediction-market traders routinely record an entry price and a probability estimate. They should also record an exit-liquidity plan. Before a major catalyst, that means comparing position size with depth near the market, tracking the spread rather than total displayed liquidity, and deciding in advance how much of the position must be sold before the event.
The most important lesson from Wisconsin is that being right about the event and monetizing that view are separate problems. Before polls closed, Crowley traders faced deep, tight markets but little certainty about the outcome. During the count, certainty increased while liquidity deteriorated. By the time liquidity surged again, the market was already near 100 cents.
Sometimes the best sale is not the one made at the highest theoretical value. It is the one completed while there is still a real order book on the other side.
Methodology and Limitations
Predexon’s sub-cent Kalshi history records full order-book snapshots with decimal-cent prices and fractional contract sizes. The analysis window runs from August 10, 2026, at 8:00 p.m. Central through August 12 at 2:45 a.m. Central. Exact snapshots are used for the three reference moments; stage medians use the last observed snapshot in each minute. The sell-side analysis measures displayed YES bids only. It cannot observe hidden trading interest, orders that could appear after a market order arrives, or execution quality for a specific trader. It is descriptive and does not establish that vote counting alone caused every quote change.
Prediction markets involve risk and are not suitable for everyone. While many of the best prediction platforms offer tools to make informed trades, outcomes are never guaranteed, and users should never risk more than they can afford to lose. Always trade responsibly. Additionally, platform availability and legal status vary by region. It is your responsibility to check local laws and verify that you are legally allowed to use a given platform before participating.
Read our full affiliate & risk disclosure.
References
Predexon: Kalshi Orderbook History (Sub-Cent)
Wisconsin Public Radio: In Shocking Comeback, David Crowley Wins Democratic Primary for Governor
Glosten, L. R., and P. R. Milgrom. “Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders.” Journal of Financial Economics 14 (1985): 71–100.
Kyle, A. S. “Continuous Auctions and Insider Trading.” Econometrica 53 (1985): 1315–1335.
Kalshi public market record: KXGOVWINOMD-26-DCRO
