Key Takeaways — Chapter 1: What Are Prediction Markets?

Core Takeaways

  1. A prediction market is a market for contracts that pay out based on the outcome of future events. The contract price equals the market's consensus probability of the event occurring.

  2. The price-probability link arises from no-arbitrage reasoning. Buyers push the price up when they believe the probability is higher than the current price; sellers push it down when they believe the probability is lower. Equilibrium price = consensus probability.

  3. Prediction markets work because of three reinforcing mechanisms: - Information aggregation — each trader contributes private information via their trades. - Incentive alignment — money on the line discourages wishful thinking and encourages research. - Continuous updating — prices adjust in real time as new information arrives.

  4. There are four main contract types, each suited to different questions: - Binary (Yes/No): Did the event happen? - Categorical (Multi-outcome): Which outcome occurred? - Scalar (Range): What was the numerical value? - Conditional: What will happen given that some precondition is met?

  5. Prediction markets consistently outperform polls and individual experts in liquid markets, but they are not magic. Their accuracy depends on liquidity, diversity of participants, and clear resolution criteria.

  6. Liquidity is the single most important determinant of price quality. A thin market (low volume, few traders, wide bid-ask spread) produces unreliable prices and is vulnerable to manipulation.

  7. Always normalize for overround. When the sum of complementary contract prices exceeds \$1.00, divide each price by the sum to recover the "fair" implied probability.

Key Formulas

Formula Description When to Use
$P_{\text{implied}}(\text{Yes}) = \dfrac{p_Y}{p_Y + p_N}$ Implied probability after removing overround Whenever you read a binary contract's price
$\text{EV} = q - c$ Expected value of buying a binary contract at price $c$ with your probability estimate $q$ Deciding whether a trade is worth making
$E[X] = \sum p_i \cdot m_i$ Expected value from a scalar market (bucket midpoints $m_i$, bucket prices $p_i$) Extracting a point forecast from a scalar market
$\text{Overround} = \left(\sum p_i\right) - 1$ Market maker's margin Assessing the cost of trading

Key Vocabulary

Term Definition
Prediction market An exchange where contracts tied to future events are traded; prices reflect crowd-consensus probabilities.
Implied probability The probability of an event as encoded in the contract's market price, after removing the overround.
Binary contract A contract that pays \$1 if the event occurs, \$0 otherwise.
Overround (vig) The amount by which the sum of all contract prices exceeds \$1.00; represents the market maker's fee.
Market maker An entity that continuously posts buy and sell orders, providing liquidity in exchange for the bid-ask spread.
Liquidity The ease with which contracts can be bought or sold without significantly moving the price.
Resolution The official determination of whether the event occurred, based on pre-specified criteria.
Information aggregation The process by which the market price synthesizes the private information of many traders into a single number.
Wisdom of crowds The phenomenon where the aggregate judgment of a diverse, independent group outperforms individual experts.
Expected value (EV) The average profit or loss from a trade if repeated many times; $\text{EV} = q - c$ for a binary contract.
Calibration A forecaster is well-calibrated if events predicted at X% actually occur X% of the time.
Favorite-longshot bias The tendency for low-probability events to be overpriced and high-probability events to be slightly underpriced.

Remember This

  • Price = Probability (in a frictionless market). A contract trading at \$0.65 means the crowd thinks there is a 65 % chance the event happens.

  • EV = Your probability minus the price. If your number is bigger, buy. If the price is bigger, pass.

  • Three liquidity red flags: volume below \$10,000, fewer than 50 traders, bid-ask spread wider than \$0.05.

  • Normalize before interpreting. If Yes costs \$0.54 and No costs \$0.50, the implied probability of Yes is NOT 54 % -- it is $0.54 / 1.04 \approx 51.9\%$.

  • Prediction markets are a complement, not a replacement. Combine them with models, expert judgment, and polling for the most robust forecasts.

  • A probability is not a certainty. An 80 % forecast means the event fails to happen 20 % of the time. Respect the uncertainty.

End of Key Takeaways -- Chapter 1