Poker Bankroll Probability and Risk Management Models: A Tool Guide
5 views
This article explains the principles of calculating bankruptcy probability in poker and practical risk management models, including the Kelly criterion, safe betting methods, and mathematical formulas. With real-world examples, it shows how to adjust bet sizes based on win rate, odds, and bankroll size to help players manage their bankroll scientifically and reduce the risk of going broke.
Tool Purpose
Bankroll probability calculation and risk management models are mathematical tools used by poker players to determine a reasonable bankroll size and control bet amounts to avoid going broke in the long run. The core objective is: given a win rate, odds, and bankroll size, maximize bankroll growth while keeping the risk of ruin within an acceptable range (e.g., <5%).
Formula Principles
1. Risk of Ruin Formula (Classic Random Walk Model)
Assuming a fixed win/loss probability and fixed profit/loss per hand, the risk of ruin can be approximated as:
$$P(\text{Ruin}) = \left( \frac{1 - \frac{b}{a}}{1 + \frac{b}{a}} \right)^{B}$$ (when win probability p < 0.5)
Where:
- a = amount won when winning
- b = amount lost when losing
- B = initial bankroll (in units of b)
2. Kelly Criterion
Kelly optimal bet fraction:
$$f^* = \frac{bp - aq}{ab} = \frac{bp - a(1-p)}{ab}$$
where p is the win probability, q = 1-p is the loss probability, a is the net profit multiplier when winning, and b is the net loss multiplier when losing (typically a=b=1 gives f* = 2p-1).
3. Fractional Kelly
To reduce volatility, a fraction of full Kelly is often used (e.g., 1/2 Kelly or 1/4 Kelly). This significantly lowers the risk of ruin while slightly reducing long-term growth.
Usage Steps
- Estimate your true win rate: Record at least 100,000 hands and calculate your win probability p.
- Determine typical win/loss ratio: For example, in No-Limit Hold'em, the ratio of average pot won to average pot lost (a:b).
- Choose a risk tolerance: Usually set an acceptable risk of ruin <5%.
- Calculate maximum bet size: Use the Kelly formula or fractional Kelly, based on your current bankroll.
- Adjust dynamically: Update your bankroll after each hand or session and recalculate bet sizes.
Practical Example
Scenario: You play in an NL200 cash game with a bankroll of $5,000. Your average win probability per hand is p = 55%, you win an average of $150 when you win, and lose an average of $100 when you lose.
Calculation:
- a = 150/100 = 1.5, b = 1 (losing unit is $100)
- Kelly fraction f* = (1.50.55 - 10.45) / (1.5*1) = (0.825 - 0.45)/1.5 = 0.25
- This suggests you should bet 25% of your bankroll per hand? But the bet amount cannot exceed your total bankroll; the actual bet size should be ≤ 0.25 * $5,000 = $1,250. However, in a cash game, you cannot always bet that large a proportion; you can only bet a portion of the current pot. A more reasonable application is to treat the Kelly fraction as the proportion of your total bankroll to risk per hand, but adjust it based on pot size. A common recommendation is to use 1/2 Kelly = 12.5%, meaning you risk at most $625 per hand.
Risk of ruin: If you consistently use full Kelly, the risk of ruin is extremely low; with 1/4 Kelly, the risk of ruin is about 0.1%.
Common Questions
Q: Is the Kelly criterion applicable to tournaments? A: Tournaments have a different payout structure; the ICM model is more appropriate, but Kelly provides a conservative estimate. It is generally recommended to use 1/4 Kelly or an even smaller fraction.
Q: My win rate fluctuates a lot. What should I do? A: Use a conservative estimate, such as your historical low win rate, or employ fractional Kelly (e.g., 1/4 Kelly).
Q: What are the assumptions of the risk of ruin formula? A: It assumes each hand is independent and identically distributed, and that the player's bankroll is infinitely divisible. In practice, bet discreteness must be considered.