Trading Betting Game (EV / Market-Making)

The core QR/QT interview round: a live betting game over many rounds. Starting from \$1000, you choose how much to wager across several sub-games (coin flips, dice sums, card products) given randomized odds each round, plus a market-making sub-game on the sum of three cards. The interviewer probes your sizing and reasoning; speed and decision-making under pressure matter more than ending profit. Requirements A timed game (≈25–45 minutes) on an assessment platform, played over multiple rounds with a starting bankroll of $1000 (often unlimited rounds within a fixed total time). Three recurring betting sub-games, each with several wagers whose odds are randomly generated each round: Coin flips — e.g. H>T, T>H, HTH/THT, HHH/TTT. Two dice — bet on the sum (=2 or 3, =4, =10, =11 or 12, odd, even, etc.). Three cards — bet on the product of the first two (>20, >100, odd, even). A market-making sub-game: a market on the sum of three cards where you can buy or sell at quoted prices that move; you profit by buying below or selling above the true sum. Each round you have ~2–4 minutes to compute EVs and allocate; the interviewer asks you to justify each decision and follows up on sizing and the probability of profit. Notes Whether you end up profitable is widely reported as not the main signal — the interviewer evaluates speed, calibrated sizing, and clear reasoning under time pressure. The stated intent is finding people who spot market opportunities fast in high-pressure settings. Bet only positive-EV wagers; when an arbitrage (guaranteed profit) appears, size it aggressively / all-in rather than applying Kelly. Interviewers frequently hint that candidates are under-betting good edges. Kelly is a useful framing but several interviewers explicitly say it "doesn't apply here" for the arbitrage cases — don't over-fit to Kelly. A common pitfall is spending too long computing exact EVs early; pick up the pace after the first round. A related estimation variant scores you on confidence intervals: each round you give a closed interval [L, U] for a numeric answer and score L/U if the true value lands inside it, 0 otherwise — incentivizing the tightest honest interval. Preparation Drill fast EV computation for dice/coin/card bets until it's reflexive; pre-build intuition for odds-vs-probability so you can decide bets in seconds. Practice recognizing arbitrage / guaranteed-profit setups and sizing them decisively; rehearse articulating why you sized a bet a given way, since every decision draws a follow-up. Practice quoting two-sided markets on a random variable (e.g. sum of three cards) and updating quotes as the price moves.