What’s it?
The Kelly Criterion is a mathematical components that calculates the optimum quantity to threat on every guess/commerce to maximise long-term development whereas avoiding smash.
Method:
If you understand your:
Then:
Kelly % = [ (b × p) – q ] / b
🎯 Let’s Apply it to a Case
-
You have got $100.
-
The sport has:
-
Excessive Threat
-
Reward vary: 1x to 100x
-
Let’s assume common reward = 10x
-
Let’s say win likelihood p = 0.1 (10%)
-
Then q = 0.9 , and b = 10
-
Kelly % = (10 × 0.1 − 0.9) / 10 = (1 − 0.9) / 10 = 0.1 / 10 = 0.01 = 1%
So you must threat only one% of your capital on every guess.
Why? As a result of risking extra (e.g., 10%, 20%) in a high-variance system will finally blow your account. Kelly ensures long-term compounding with minimal threat of smash.
Ought to You Hold Threat Fixed?
Not all the time. This is the logic:
| State of affairs | Motion | Motive |
|---|---|---|
| You’re profitable | Enhance barely | Capital grows → greater % dollar-wise, preserve % steady or improve barely. |
| You’re dropping | Scale back threat | Keep away from drawdowns turning into smash. Compounding in reverse is lethal. |
| Sport edge modifications | Recalculate Kelly | As likelihood or payout modifications, so does optimum threat. |
Actual-World Merchants Use This:
-
Edward Thorp (inventor of Kelly) turned blackjack earnings right into a hedge fund empire.
-
Renaissance Applied sciences, Soros, Druckenmiller, and plenty of quant funds use Kelly-like fashions.
-
Crypto fund managers scale positions dynamically based mostly on edge + volatility.
Conclusion: Technique Abstract
| Metric | Worth / Logic |
|---|---|
| Capital | $100 |
| Common Win % | 10% |
| Reward/Threat | 10:1 |
| Threat per Wager | 1% (Kelly) |
| Modify per end result? | Sure, adapt barely |
| Purpose | Keep away from smash, develop exponentially over time |