Risk Management and Position Sizing

You've built a strategy. You've backtested it. You've validated it with walk-forward analysis. Now the question that will determine whether you survive long enough to profit: how much money do you put on each trade?

This is where most traders — especially systematic ones who did everything else right — blow up. Not because their strategy was bad, but because they sized too aggressively for the drawdowns they inevitably faced.

Risk management isn't a safety net you add at the end. It's the foundation that everything else sits on.

The Only Risk You Can Control

You can't control whether the next trade wins or loses. You can't control whether the market gaps against you overnight. You can't control whether your strategy hits its worst drawdown in month one or month twelve.

What you can control is how much you lose when you're wrong. That's position sizing — and it's the single most important decision you make as a trader.

Consider two traders running the exact same strategy:

• Trader A risks 2% of capital per trade
• Trader B risks 10% of capital per trade

After a streak of 5 consecutive losers (which will happen to any strategy eventually), Trader A is down about 10%. Uncomfortable, but recoverable. Trader B is down 41%. From that hole, they need a 70% gain just to get back to breakeven. Most people quit long before that.

The goal of position sizing isn't to maximize returns. It's to ensure you're still in the game when the edge plays out.

Position Sizing Methods

Fixed Fractional (Percent Risk)

The most widely used method. You risk a fixed percentage of your current capital on every trade. If your account is 10 lakhs and you risk 2% per trade, your maximum loss per trade is Rs 20,000.

To calculate position size, you need to know your stop-loss distance:

Position Size = (Account × Risk%) / Stop Distance

Example:
Account = Rs 10,00,000
Risk per trade = 2% = Rs 20,000
Stop distance = Rs 50 per share

Position Size = 20,000 / 50 = 400 shares

The beauty of fixed fractional sizing is that it automatically scales: you trade bigger when you're winning (account grows) and smaller when you're losing (account shrinks). This makes it very hard to blow up — losses shrink the bet size, creating a natural brake.

Fixed Ratio (Capital Per Unit)

Simpler variant: allocate a fixed amount of capital per lot or per unit. For example, Rs 5 lakhs per NIFTY lot. If your account is 10 lakhs, you trade 2 lots. If it drops to 7 lakhs, you drop to 1 lot.

This is practical for futures and options where lot sizes are fixed. It's less mathematically elegant than percent risk, but easier to implement and reason about.

Kelly Criterion

The Kelly formula tells you the theoretically optimal fraction to bet, based on your win rate and payoff ratio:

Kelly% = W - (1 - W) / R

Where:
W = win rate (e.g., 0.45 for 45%)
R = average win / average loss (e.g., 2.0)

Kelly% = 0.45 - (0.55 / 2.0) = 0.175 = 17.5%

In theory, betting the Kelly fraction maximizes long-term growth. In practice, never use full Kelly. The formula assumes you know your exact win rate and payoff ratio — which you don't. Estimation errors make full Kelly extremely volatile.

Most practitioners use half-Kelly or quarter-Kelly — sacrificing some theoretical growth for dramatically smoother equity curves. Half-Kelly gives about 75% of the growth rate with far less drawdown.

Which Method to Use?

For most retail algo traders, fixed fractional at 1-2% risk per trade is the right answer. It's simple, it scales, and it's conservative enough to survive the inevitable bad streaks. Start there. You can get more sophisticated later — but more sophisticated usually means more ways to fool yourself.

Drawdown: The Number That Matters Most

We covered max drawdown in the backtesting post. Now let's talk about it in the context of live trading, where it becomes a psychological survival test.

Your backtest says the max drawdown is 15%. Here's what that means in practice:

• Live drawdowns will be worse. Expect 1.5-2x your backtest max drawdown. Slippage, timing differences, and market conditions you haven't backtested will make reality harder than simulation.
• Drawdowns feel much worse than they look. A 15% drawdown on a chart is a small dip. A 15% drawdown on your real account — watching Rs 1.5 lakhs disappear from Rs 10 lakhs — tests your conviction at a visceral level.
• Drawdowns cluster. You won't lose 1% per day for 15 days. You'll have days that lose 3-4%, followed by recovery, followed by another sharp drop. The emotional whipsaw is brutal.

The Drawdown Survival Test

Before going live, do this exercise:

1. Take your backtest max drawdown
2. Multiply it by 2 (your realistic worst case)
3. Multiply by your planned account size
4. Look at that rupee amount and ask: can I watch this much money disappear without turning off the strategy?

If the answer is no, reduce your position size until it becomes yes. This isn't weakness — it's the most rational thing you can do. A strategy you abandon during a drawdown gives you the worst of both worlds: you eat the losses but miss the recovery.

The best position size isn't the one that maximizes returns. It's the largest size you can hold through the worst drawdown without blinking.

Loss Limits: Circuit Breakers for Your Account

Even with proper position sizing, you need hard limits — predetermined points where you stop trading and reassess. Think of them as circuit breakers.

Daily Loss Limit

Set a maximum loss per day. If you hit it, stop trading for the rest of the session. This prevents a single bad day from becoming catastrophic, and it prevents revenge trading — the instinct to "make it back" that leads to the worst blowups.

A reasonable daily limit might be 1-2% of your account, or a fixed point value for futures traders.

Weekly Loss Limit

Same idea, broader timeframe. If your week is bad enough to hit this limit, step back. Review whether the losses are within normal strategy behavior or whether something has changed.

Drawdown Review Trigger

Set a drawdown level that triggers a formal strategy review. Not an automatic shutdown — but a commitment to stop and evaluate. Is the drawdown within historical norms? Has the market regime changed? Is the strategy still executing correctly?

And set a hard stop — a drawdown level where you halt the strategy entirely. This should be beyond your expected worst case but short of account-threatening. Think of it as the "something is fundamentally broken" threshold.

Example Framework

Daily loss limit:     Stop trading for the day
Weekly loss limit:    Stop trading, review over weekend
Drawdown review:      Reduce size, investigate
Drawdown hard stop:   Halt strategy, full review before restarting

The specific numbers depend on your strategy, your account, and your risk tolerance. The key is to define them before you start trading — not when you're in the middle of a losing streak and your judgment is compromised.

The Compounding Trap

New algo traders often run backtests with compound position sizing — reinvesting all profits into larger positions. The equity curves look spectacular. "Look, my Rs 10 lakh account turned into Rs 1 crore!"

The problem is that compounding amplifies drawdowns just as much as it amplifies gains. A 30% drawdown on a compounded account means you're giving back months of accumulated profits in days. And because position sizes grew with the account, the rupee amount of the drawdown can be staggering.

A more realistic approach:

• Start with fixed position sizes. Trade 1 lot regardless of account growth. This lets you validate the strategy without compounding risk.
• Scale up in steps. After hitting predefined milestones (e.g., 3 months profitable, drawdown within limits), add 1 lot. Don't scale continuously — step up in discrete, deliberate increments.
• Withdraw periodically. Taking profits off the table isn't just prudent — it's psychologically stabilizing. It's much easier to ride a drawdown when you've already locked in some gains.

Correlation and Portfolio Risk

If you're running multiple strategies, position sizing gets more nuanced. Two strategies that both go long when markets rally aren't really diversifying — they're doubling your exposure to the same risk.

Think about:

• Do your strategies trade the same instrument? Two strategies on NIFTY have correlated drawdowns.
• Do they trade in the same direction? A long-biased strategy and a mean-reversion strategy might partially offset.
• What's the combined worst case? If both strategies hit max drawdown simultaneously (which they will, eventually), can your account handle it?

The combined position size across all strategies should still respect your overall risk limits. Running 3 strategies at 2% risk each doesn't mean 2% total risk — it means 6% if they all lose on the same day.

The First Three Months

When you take a validated strategy live for the first time, treat it as an experiment, not a money-making operation. Here's a practical framework:

1. Paper trade first. Run the strategy in real-time with no money at risk. Verify that signals fire correctly, orders would execute at reasonable prices, and the strategy behaves as expected during live market hours. Two weeks minimum.
2. Go live at minimum size. One lot, smallest position possible. Your goal isn't profit — it's confirming that live results roughly match your walk-forward OOS expectations.
3. Track everything. Log every trade, every signal, every deviation from expected behavior. Compare live performance to backtest and paper trading.
4. Scale up only after validation. If after 2-3 months your live performance is within expected range of your OOS results (accounting for the 30-40% degradation that's normal), you have evidence to increase size.

This is slow. It's supposed to be. The traders who survive aren't the ones who made money fastest — they're the ones who avoided the catastrophic mistakes that take most people out of the game.

Summary

1. Position sizing determines survival. A great strategy with bad sizing will blow up. A mediocre strategy with good sizing will survive long enough to improve.
2. Start with 1-2% fixed fractional risk. Simple, effective, hard to ruin. Graduate to more complex methods only after you've proven you can handle the basics.
3. Expect drawdowns 1.5-2x worse than backtests. Size your positions for the realistic worst case, not the backtest worst case.
4. Set loss limits before you start trading. Daily, weekly, and drawdown hard stops. Define them when your head is clear, not when you're losing money.
5. Start at minimum size and scale slowly. The first 3 months are for validation, not profit. One lot, track everything, scale up only with evidence.
6. The best position size is the one you can hold through the worst drawdown. If you'd turn off the strategy during a bad streak, you're sized too large.

What's Next

You now have a complete framework: design a strategy, backtest it, validate with walk-forward analysis, and size your positions to survive real-world trading. The next post will cover connecting AmiBroker to a broker for automated execution — the engineering that turns a validated strategy into a hands-free trading system.