- Price as probability: a contract at 63¢ means the market assigns roughly a 63% chance. It is a forecast backed by money, not a poll of opinions.
- Markets vs polls: markets partly consume polls, so their edge is real but less independent than it sounds. Best where dispersed private information exists.
- Favourite-longshot bias: the most repeatable pattern in political markets — long shots overpriced, near-certainties mildly underpriced.
- Resolution beats prediction: more losses come from misreading which contract you bought than from being wrong about the winner.
- Liquidity decays near the event: spreads widen exactly when you most want to exit. Plan the exit before the news, not during it.
1. What an Election Market Actually Is
An election prediction market is a set of contracts that pay out based on an electoral outcome. On Polymarket the mechanics are the same as any other market: you buy a YES or NO position at some price between 0 and 100 cents, and if the contract resolves in your favour it pays $1.00 per share.
The difference from a poll is the incentive structure. A poll asks people what they think or expect. A market asks people to put money behind their expectation. Anyone who believes the price is wrong has a direct financial reason to trade against it, and the price moves until the disagreement is exhausted.
| Feature | Opinion poll | Prediction market |
|---|---|---|
| Measures | Stated preference or expectation | Money-weighted expectation |
| Cost of being wrong | None | Direct financial loss |
| Updates when | At sampling intervals | Continuously, with each trade |
| Who participates | Sampled respondents | Self-selected, capital-holding traders |
| Main weakness | Sampling and response bias | Liquidity, thin markets, whale moves |
The core intuition: polls suffer from people having no reason to be careful. Markets suffer from participation being limited to those willing to trade. Each weakness is real, and neither instrument is unconditionally superior to the other.
2. Reading Price as Probability
A contract trading at 63¢ is the market's way of saying "roughly 63% likely." That mapping is straightforward, but three refinements matter in practice.
| Price | Stated probability | What it means in practice |
|---|---|---|
| 95¢ | ~95% | Near-certain. Modest return, exposed to tail surprises. |
| 80¢ | ~80% | Strong favourite. Small mispricings here compound reliably. |
| 50¢ | ~50% | Maximum uncertainty. Widest disagreement, often widest spread. |
| 20¢ | ~20% | Underdog. Payoff is large; base rate is easy to overestimate. |
| 5¢ | ~5% | Long shot. Historically the most systematically overpriced band. |
The three refinements
- The price includes a spread. What you pay to buy is the ask, not the midpoint. A "63¢" market might be 62¢ bid / 64¢ ask — you are paying 64¢ to express that view.
- The price reflects the marginal trader, not the average. In a thin market a small number of participants set the price. That is the market's view, not necessarily the crowd's.
- The price is a distribution summary, not a point truth. 63% does not mean "will happen." It means that if you re-ran this exact situation many times, the outcome occurs about 63 times in 100. Single outcomes tell you almost nothing about calibration.
The calibration trap: after an election, everyone judges the market by the result. A 90% favourite that loses is described as a market failure. That is statistically illiterate — a 90% favourite is expected to lose one time in ten. Judge markets on long-run calibration across many events, never on one outcome.
3. Markets vs Polls: An Honest Comparison
The popular claim is that prediction markets beat polls. The honest version is more interesting.
Markets do not operate in isolation from polling. Traders read polls, so poll information is already embedded in the market price. When a market "beats" a poll, it is often not outperforming on independent information — it is aggregating polls plus turnout models, historical base rates, and early-voting data in a way that a single poll cannot.
| Condition | Markets tend to do well | Markets tend to struggle |
|---|---|---|
| Liquidity | High volume, many participants | Thin markets dominated by a few traders |
| Information spread | Many participants hold private signals | All participants read the same polls |
| Time horizon | Weeks to days out, information accumulates | Minutes after a shock, spreads blow out |
| Market type | Winner markets with clear rules | Exotic markets with ambiguous criteria |
Practical conclusion: treat the market price as a well-informed prior, not as ground truth. Your edge comes from information the price has not yet absorbed — local knowledge, faster reading of primary returns, or a structural bias you can identify. See our prediction market accuracy study for the calibration evidence behind this.
4. The Favourite-Longshot Bias — Your Most Repeatable Edge
Across betting and prediction markets, the same pattern recurs: low-probability outcomes trade at prices higher than their true probability, while high-probability outcomes trade slightly lower. This is the favourite-longshot bias, and it is the most durable structural inefficiency available to a disciplined trader.
| Contract band | Typical bias | What a disciplined trader does |
|---|---|---|
| 1–5¢ long shots | Overpriced — pays less than the true probability implies | Look for NO positions; the lottery appeal inflates YES |
| 5–20¢ underdogs | Mildly overpriced | Require a genuine informational edge before buying |
| 40–60¢ coin flips | Roughly fair, widest spreads | Trade only with strong conviction; pay attention to cost |
| 80–95¢ favourites | Mildly underpriced — modest but persistent edge | Small, repeated positions; the least glamorous profit on the platform |
Why does this persist? Because payoffs are psychologically asymmetric. A 5¢ contract that could pay $1 offers a 20x return and is fun to own. A 90¢ contract that pays $1 offers an 11% return and feels pointless. That preference is not a rational assessment of probability — it is a preference about the shape of the payoff, and it systematically pushes long-shot prices up and favourite prices down.
Why this matters for elections specifically: political markets are full of novelty candidates, withdrawal scenarios, and dramatic long-shot stories. That is exactly the environment where the bias is strongest and most reliable.
5. Liquidity and the Cost of Exiting
The most practical thing to understand about election markets is that liquidity is not constant — it decays as the event approaches.
Market makers and larger traders know that information arrives faster near the event than they can reprice. Their response is to widen spreads and reduce quoted size. The result is that the moment you most want to act — when the news lands — is the moment trading is most expensive.
| Time to event | Typical liquidity | Spread | Implication |
|---|---|---|---|
| Months out | Moderate, opinion-driven flow | 2–5¢ | Cheap to enter, slow to reprice |
| Weeks out | High, news cycle active | 1–3¢ | Usually the best combination |
| Days out | Very high volume | 0.5–2¢ | Best execution, most competition |
| Final hours | Thinning sharply | 2–10¢+ | Exits become expensive exactly when needed |
| After event, pre-resolution | Thin, resolution-dependent | Wide, sometimes no quotes | You may be locked in until resolution |
Plan the exit before the event. Decide in advance whether you are holding to resolution or scaling out into strength. The traders who get hurt are the ones who assumed they could exit instantly on the news and discovered the order book had emptied.
6. Resolution Traps in Political Markets
This is where more money is lost than in any misjudgement of the electorate. Political markets are unusually prone to resolution ambiguity because political events have messy shapes.
| Trap | How it catches you | What to check |
|---|---|---|
| Source ambiguity | Two networks call a race at different times | Which named source is authoritative |
| Popular vote vs winner | You are right about votes, wrong about the contract | Which measure the contract actually settles on |
| Withdrawal / substitution | Your candidate drops out; what happens to the contract? | Explicit rules for withdrawal and replacement |
| Timing edge cases | "By election day" does not include certification | Exact dates, timezones, and process steps |
| Pledged vs bound delegates | You assume a mechanism the rules do not require | The precise procedural definition used |
| Recounts and legal challenges | Outcome is clear but certification is contested | How disputes are resolved and by whom |
The expensive lesson: being right about politics and wrong about the contract pays nothing. Read the full resolution description before you price anything. Our resolution guide walks through how disputes actually arise.
7. A Risk Framework for Election Trading
- Write your probability down first. Before looking at the price, state your own estimate. Then compare. If you looked at the price first, you have anchored yourself to it — a well-documented effect that makes genuine mispricing much harder to spot.
- Demand a real gap. If your estimate is 55% and the market says 52%, that is noise, not edge. Spread costs and resolution risk will consume it. Look for differences large enough to survive both.
- Read the resolution criteria fully. Every time. Then read the named source and the deadline.
- Size for total loss. Assume any single election position can go to zero, because it can. Position sizing is what converts a good process into a survivable one — see bankroll management.
- Check the spread before you commit. Cost is part of your entry price. A 3-cent spread on a 5-cent edge is not a trade.
- Decide the exit in advance. Holding to resolution or scaling out — pick one before you enter, not while the news is breaking.
- Keep a record of your estimates. Score them after the fact. The only way to know whether your political judgement is actually good is to track it honestly over many events.
8. Five Recurring Mistakes
| Mistake | Why it happens | The fix |
|---|---|---|
| Betting your opinion, not your edge | Political views feel like knowledge | Demand a specific informational advantage, not agreement with your priors |
| Buying long shots for the payoff | Lottery appeal, narrative appeal | Compare the price to the base rate for that class of outcome |
| Ignoring the spread | Midpoint price looks like the cost | Price your edge net of the actual ask |
| Judging markets by single results | Outcome bias after every election | Score calibration across many events, not one |
| Assuming you can exit on the news | Liquidity looks fine weeks out | Plan the exit before the event; check depth, not just price |
Frequently Asked Questions
What is an election prediction market?
A market where contract prices represent the crowd's probability estimate for an electoral outcome. A contract at 63¢ pays $1 if the outcome occurs, implying roughly a 63% assessment. Because participants risk real money, disagreement is resolved by trading rather than by opinion, which is the key structural difference from a poll.
Are prediction markets more accurate than polls?
On average they have compared favourably, but the advantage is less independent than usually claimed — markets consume polling data, so they are partly downstream of the same source. Their edge is clearest where many participants hold dispersed private information and thinnest in low-liquidity races where a few traders set the price. Treat the price as a strong prior, not a verdict.
Why are long-shot candidates often overpriced?
The favourite-longshot bias. Traders systematically overpay for low-probability, high-payoff contracts because the large multiple is attractive, while underpricing near-certainties that offer modest returns. This is one of the most durable patterns in both betting and prediction markets, and political markets are especially prone to it because they generate dramatic long-shot stories.
What is the biggest risk in election markets?
Resolution criteria, not forecasting error. Many losses come from buying a contract whose actual settlement condition differed from the headline. Source ambiguity, popular-vote versus winner markets, withdrawal rules, and certification timing all cause outcomes that contradict a correct political judgement. Read the full resolution description before trading.
Should I hold an election position until resolution?
It depends on where your edge sits. If your edge is in the final outcome, holding to resolution avoids paying the spread repeatedly. If your edge is in a repricing that should occur before election day, exiting earlier captures it with less exposure to late news. Both are legitimate — the mistake is defaulting to holding without considering which one applies to you.
Why does liquidity dry up near the event?
Market makers narrow their exposure when information arrives faster than they can reprice, so they widen spreads and pull size. That is precisely when you are most likely to want to trade, which is why planning your exit in advance matters more than reacting in the moment.
How should I size an election position?
Assume the position can go to zero, because it can. Then apply standard bankroll logic: no single event should be able to materially damage your total capital. For long-shot positions specifically, remember that sizing should reflect the base rate for that class of outcome, not the size of the potential payoff.