Chapter 13 Quiz: Finding and Quantifying Your Edge

Instructions: Select the best answer for each question. Answers are provided at the end.

Question 1

What is the formal definition of "edge" in prediction market trading?

(a) The amount of money you expect to make on a trade (b) The difference between your estimated probability and the market's implied probability (c) The number of trades you win divided by total trades (d) The maximum amount you are willing to lose on any single trade

Question 2

A YES contract is priced at $0.40. You estimate the true probability at 55%. What is the expected value of buying one YES contract?

(a) $0.40 (b) $0.55 (c) $0.15 (d) -$0.15

Question 3

Which of the following is an example of an "analytical edge"?

(a) You have access to a leaked internal memo from a company (b) You build a superior statistical model from publicly available data (c) You trade faster than other participants after news breaks (d) You exploit the favorite-longshot bias

Question 4

The Kelly criterion for buying YES contracts at price $c$ with estimated probability $p$ is:

(a) $f^* = p - c$ (b) $f^* = (p - c) / c$ (c) $f^* = (p - c) / (1 - c)$ (d) $f^* = (c - p) / (1 - p)$

Question 5

A market prices YES at $0.60. You estimate the true probability at 75%. What is the full Kelly fraction?

(a) 15% (b) 25% (c) 37.5% (d) 75%

Question 6

What happens when you bet at exactly 2x the Kelly criterion?

(a) Your growth rate doubles compared to full Kelly (b) Your growth rate is approximately zero (same as not betting) (c) You are guaranteed to go bankrupt (d) Your growth rate is 50% of full Kelly

Question 7

Why is half Kelly widely recommended over full Kelly?

(a) It produces exactly half the growth rate of full Kelly (b) It achieves about 75% of full Kelly's growth rate with significantly lower drawdown risk (c) It is always optimal regardless of circumstances (d) It eliminates the possibility of any losses

Question 8

You estimate a probability of 60% for an event priced at 60%. What does the Kelly criterion recommend?

(a) Bet 10% of your bankroll (b) Bet half your bankroll (c) Do not bet (d) Short the market

Question 9

Which type of edge typically decays the fastest?

(a) Behavioral edge (b) Analytical edge (c) Information edge (d) Calibration edge

Question 10

In the edge decomposition framework, what is "timing edge"?

(a) Being the first to discover new information (b) Entering trades at better prices than the average (e.g., below VWAP for YES buys) (c) Predicting when markets will close (d) Trading only during business hours

Question 11

You are deciding whether to trade a market. Which of the following is the STRONGEST warning sign that you lack edge?

(a) The market is highly liquid with many active traders (b) You cannot articulate a specific reason why the market price is wrong (c) The event will resolve in more than 6 months (d) The market price is close to 50%

Question 12

What is the break-even probability for buying a YES contract priced at $0.65?

(a) 35% (b) 50% (c) 65% (d) 100%

Question 13

In reference class forecasting, what is the main tension you face when selecting a reference class?

(a) Cost vs. benefit (b) Speed vs. accuracy (c) Relevance vs. sample size (d) Simplicity vs. complexity

Question 14

The "extremization" technique for combining probability estimates involves:

(a) Taking the most extreme estimate and discarding the others (b) Pushing the combined average away from 50% to correct for underconfidence (c) Removing outlier estimates before averaging (d) Doubling the distance between the estimate and the market price

Question 15

For full Kelly betting, what is the probability that your bankroll will at some point drop to 50% of its starting value?

(a) 10% (b) 25% (c) 50% (d) 75%

Question 16

For half Kelly betting, what is the probability that your bankroll will at some point drop to 50% of its starting value?

(a) 10% (b) 25% (c) 50% (d) 75%

Question 17

The Brier score is used to measure:

(a) The profitability of a trading strategy (b) The calibration and accuracy of probability forecasts (c) The volatility of market prices (d) The speed at which a trader reacts to news

Question 18

A prediction market has these properties: thin liquidity, mostly recreational traders, and large apparent mispricings. Based on the market efficiency trajectory described in this chapter, this market is most likely in which stage?

(a) Mature stage (b) Growth stage (c) Early stage (d) Decay stage

Question 19

You have been trading for 6 months with 80 resolved trades. Your probability estimates have a Brier score of 0.21, while the market prices for the same events have a Brier score of 0.19. What does this tell you?

(a) You are more accurate than the market and have edge (b) The market is more accurate than you, and you likely do not have edge through calibration (c) You have a 2% edge on every trade (d) You should increase your position sizes

Question 20

When computing Kelly for a prediction market trade, what should you do if your 80% confidence interval for the true probability includes the market price?

(a) Use full Kelly based on your point estimate (b) Double your position size to make up for uncertainty (c) Reduce your position size or do not trade, since your edge is not statistically robust (d) Use the upper bound of your confidence interval

Question 21

P&L attribution is useful because it helps you:

(a) Predict which markets will be profitable in the future (b) Identify which sources of edge and market categories drive your returns (c) Calculate the exact Kelly fraction for each trade (d) Eliminate all losing trades from your portfolio

Question 22

Which of the following is NOT a recommended component of a trade log?

(a) Your probability estimate and rationale (b) The Kelly fraction you computed and the fraction you actually used (c) The names and positions of other traders in the market (d) A post-mortem analysis of what you got right and wrong

Question 23

The expected value of buying NO when the market price of YES is $p$ and your estimated probability of YES is $q$ is:

(a) $q - p$ (b) $p - q$ (c) $(1-q) - (1-p)$ (d) Both (b) and (c) -- they are equivalent

Question 24

A trader's monthly edge has declined from 8% to 2% over 12 months. Trading costs are approximately 1.5% per trade. What should the trader do?

(a) Double their position sizes to compensate for the declining edge (b) Continue trading normally since they still have positive edge (c) Consider retiring this strategy soon, as the net edge (after costs) is approaching zero (d) Switch to full Kelly to maximize the remaining edge

Question 25

Which of the following best describes why most traders in prediction markets do not have a consistent edge?

(a) Prediction markets are rigged against individual traders (b) Trading is a negative-sum game after fees, so the average trader must lose money; having edge requires being better than average (c) All information is instantly reflected in market prices (d) Prediction markets only work for large institutional traders

Answer Key

Question 1: (b) Edge is formally defined as the difference between your estimated probability and the market's implied probability. This is expressed as $E = q - p$.

Question 2: (c) EV = q - p = 0.55 - 0.40 = $0.15 per contract.

Question 3: (b) An analytical edge comes from processing publicly available information more skillfully, such as building a better model. Options (a) is information edge, (c) is speed edge, and (d) is behavioral edge.

Question 4: (c) The Kelly fraction for buying YES is $f^* = (p - c)/(1 - c)$. This is derived from the general Kelly formula adapted for prediction market payoffs.

Question 5: (c) $f^* = (0.75 - 0.60)/(1 - 0.60) = 0.15/0.40 = 0.375 = 37.5\%$.

Question 6: (b) At 2x Kelly, the expected growth rate drops to approximately zero, the same as not betting at all. Beyond 2x Kelly, growth becomes negative.

Question 7: (b) Half Kelly achieves approximately 75% of full Kelly's geometric growth rate while dramatically reducing the probability and magnitude of drawdowns.

Question 8: (c) When your probability estimate equals the market price, the Kelly fraction is $(0.60 - 0.60)/(1 - 0.60) = 0$. Kelly correctly recommends no bet when there is no edge.

Question 9: (c) Information edges decay fastest because once private information becomes public, the edge vanishes immediately.

Question 10: (b) Timing edge measures whether you consistently enter trades at better prices than the average market price (e.g., VWAP) during the period.

Question 11: (b) The inability to articulate a specific reason for disagreeing with the market is the strongest warning sign. It suggests your trading is based on feeling rather than analysis.

Question 12: (c) The break-even probability equals the market price. Buying YES at $0.65 breaks even when the true probability is exactly 65%.

Question 13: (c) The core tension is between relevance (a narrow, specific reference class) and sample size (a broader class with more data points).

Question 14: (b) Extremization pushes the averaged estimate away from 50% to correct for the tendency of averaged forecasts to be underconfident.

Question 15: (c) For full Kelly, the probability of ever reaching fraction $x$ of your bankroll is $x$ itself. So the probability of a 50% drawdown is 50%.

Question 16: (b) For half Kelly, the probability of reaching fraction $x$ is approximately $x^2$. So the probability of a 50% drawdown is $0.5^2 = 0.25 = 25\%$.

Question 17: (b) The Brier score measures the mean squared error between probability forecasts and actual outcomes, capturing both calibration and resolution.

Question 18: (c) Thin liquidity, recreational traders, and large mispricings are hallmarks of an early-stage prediction market.

Question 19: (b) A higher Brier score means worse accuracy. Your 0.21 is worse than the market's 0.19, indicating the market's probability estimates are more accurate than yours.

Question 20: (c) If the market price falls within your confidence interval, your edge is not statistically robust. You should reduce your position size or skip the trade.

Question 21: (b) P&L attribution helps you understand which market categories and edge sources are driving (or dragging on) your returns, allowing you to focus on your strengths.

Question 22: (c) Information about other traders' identities and positions is generally unavailable and not a useful component of a trade log. All other options are recommended components.

Question 23: (d) EV of buying NO = $p - q = (1-q) - (1-p)$. Both expressions are algebraically identical.

Question 24: (c) With edge at 2% and declining, and costs at 1.5%, the net edge is only 0.5% and shrinking. The strategy should be retired soon before it becomes net-negative.

Question 25: (b) Prediction markets are zero-sum before fees and negative-sum after fees. The average participant must lose an amount equal to the fees. Having a consistent edge requires being more accurate than the collective market.