🧠 Position Sizing Science: The Kelly Criterion Applied to Indian Stock Portfolios

Hook:
Imagine having a winning stock tip—a company you’re 80% sure will double in a year. How much of your portfolio should you bet? 10%? 50%? All in?
Most Indian investors guess. They bet too little on their winners (killing potential) or too much on their risks (inviting ruin). Enter The Kelly Criterion: the mathematical formula used by legendary investors like Warren Buffett and Jim Simons to scientifically determine the perfect bet size.

The Gambler’s Ruin vs. The Scientist’s Edge 🎲📐

In the bull markets of 2020-2024, many retail investors in India confused “conviction” with “concentration.” They went heavy into small-caps, only to see portfolios vanish when volatility hit.

The Kelly Criterion solves a specific problem: What is the optimal percentage of capital to allocate to a trade to maximize long-term wealth growth without going bankrupt?

It balances two opposing forces:

1. Betting too little: You leave money on the table; compounding is too slow.

2. Betting too much: A string of losses wipes you out (Geometric Drag).

The Formula: Adapting Kelly for Dalal Street 📝

While the classic Kelly formula works for coin tosses (win/loss), the stock market has continuous outcomes. For equity portfolios, we use the Continuous Approximation of the Kelly Criterion.

To avoid confusion with symbols, here is the formula in plain text:

The Equity Formula

Optimal Allocation % = (Mean Return – Risk Free Rate) / (Volatility Squared)

Where:

• Optimal Allocation %: The fraction of your portfolio to invest.

• Mean Return: Your expected annualized return (e.g., 0.25 for 25%).

• Risk Free Rate: The safe return available (use India’s 10-Year G-Sec Yield, approx. 0.07 or 7%).

• Volatility Squared: The Annualized Volatility (Standard Deviation) multiplied by itself. This represents the variance (risk).

🇮🇳 Practical Case Study: The “Trent” vs. “HUL” Dilemma

Let’s apply this to two distinct Indian stock profiles to see what the math recommends.

Scenario A: The High-Growth Momentum Play (e.g., Trent/Zomato)

• Expected Return: Bullish thesis suggests 25% CAGR (0.25).

• Volatility: High beta stock, historical volatility is 40% (0.40).

• Risk-Free Rate: 7% (0.07).

Calculation:

1. Excess Return: 0.25 – 0.07 = 0.18

2. Variance (Volatility Squared): 0.40 x 0.40 = 0.16

3. Optimal Allocation: 0.18 / 0.16 = 1.125

Result: 112.5% Allocation.
The formula suggests you should go all in and even use leverage (margin) to buy this stock.
Wait, isn’t that insane? Yes. We will address this in the “Half-Kelly” section.

Scenario B: The Defensive Compounder (e.g., HUL/ITC)

• Expected Return: Steady compounder, expecting 15% CAGR (0.15).

• Volatility: Low beta, volatility is 18% (0.18).

• Risk-Free Rate: 7% (0.07).

Calculation:

1. Excess Return: 0.15 – 0.07 = 0.08

2. Variance (Volatility Squared): 0.18 x 0.18 = 0.0324

3. Optimal Allocation: 0.08 / 0.0324 = 2.46

Result: 246% Allocation.
Because the volatility is so low relative to the excess return, the math screams “Leverage up!”

⚠️ The “Full Kelly” Suicide Trap

If you followed “Full Kelly” (allocating 100%+ as calculated above), you would mathematically maximize wealth if and only if:

1. Your expected return estimates are perfectly accurate.

2. Stock returns follow a perfect normal distribution (they don’t; they have “fat tails” or Black Swan events).

3. You have the emotional robot-brain to withstand a 90% drawdown.

The Volatility Tax: Full Kelly portfolios are wildly volatile. In the Indian market, a “Full Kelly” portfolio would have been wiped out during the March 2020 Covid crash.

The Solution: “Half-Kelly” (The Sanity Filter) 🛡️

Smart practitioners never use Full Kelly. They use Fractional Kelly (usually Half-Kelly or Quarter-Kelly).

Half-Kelly Formula:
Safe Allocation = Optimal Allocation / 2

Why Half-Kelly?

1. Model Error: You are likely overestimating your return and underestimating volatility. Halving the bet provides a “Margin of Safety.”

2. Psychology: Half-Kelly captures 75% of the growth of Full Kelly but with 50% of the volatility.

Re-sizing Our Indian Examples (Half-Kelly)

Wait, 56% is still too high for one stock!
Correct. In a concentrated portfolio of 5-10 stocks, Kelly might suggest huge weights. But for most retail investors, Kelly should be applied to Asset Classes or used as a Conviction Multiplier rather than a raw number.

🛠️ Applying Kelly to Your Portfolio (Step-by-Step)

Don’t use the raw formula to put 50% in one stock. Use it to rank and weight your ideas.

Step 1: The “Edge” Check

If Expected Return is less than or equal to the Risk-Free Rate (7%), Kelly is zero. Don’t invest.

• Example: buying a PSU stock solely for a 4% dividend yield when G-Sec gives 7%. Kelly says: Exit.

Step 2: The Relative Sizing

If Stock A has a Kelly score of 1.12 and Stock B has 0.5, Stock A deserves roughly 2x the weight of Stock B, assuming diversification constraints allow.

Step 3: The “Quarter-Kelly” Rule for Retail

For Indian retail investors, I recommend Quarter-Kelly (Optimal Allocation / 4):

• High Growth Stock: 112% / 4 = 28% (Still aggressive, but manageable).

• Constraint: Never exceed 20% in a single stock, regardless of what Kelly says.

📉 Behavioral Constraints: Can You Stomach the Drop?

Kelly optimization assumes you act like a casino: cold, rational, and infinite time horizon.
Real Scenario:

• Ravi (IT Professional): Allocates 40% to a Small-Cap fund using Kelly logic.

• Event: Market corrects 20%. His portfolio is down ₹8 Lakhs.

• Reaction: Ravi panics, forgets the math, and sells.

• The Kelly Failure: The formula worked, but the human broke.

Modification: Always scale down your Kelly size by your “Sleep Point”—the allocation size where you can sleep soundly if the stock drops 20% overnight.

Key Takeaways 🏁

1. Don’t Guess, Calculate: Use the Equity Kelly Formula (Return – Risk Free) / Volatility Squared to objectively check if your conviction matches the math.

2. Volatility Punishes Size: High volatility stocks naturally require smaller position sizes, even if returns are high. The math proves that “YOLO-ing” into volatile stocks is mathematically suboptimal.

3. Use Fractional Kelly: Always divide your result by 2 (aggressive) or 4 (conservative). This protects you from your own overconfidence (hubris) and market surprises.

4. Respect the Zero: If your expected return isn’t significantly above 7% (India’s Risk-Free Rate), Kelly dictates a 0% allocation. Cash is better than a bad bet.

5. Directional Guide: Use Kelly not for exact percentages, but to determine relative conviction. If Kelly says “Bet Big,” investigate why—is the volatility low, or is your return assumption unrealistic?