The uncomfortable opening number

SEBI's "Comparative Study of Growth in Equity Derivatives Segment vis-à-vis Cash Market After Recent Measures" (released 7 July 2025) found that individual traders lost a net ₹1,05,603 crore in FY25 — a 41% jump over FY24's ₹74,812 crore. About 91% of individual F&O traders ended the year in the red, with an average loss of roughly ₹1.1 lakh per person. The data covered nearly 96 lakh unique traders across the top 13 brokers, so this is not a fringe sample.

Most commentary on these numbers blames "speculation," "leverage," or "lack of education." Those are symptoms. The actual disease is simpler and more fixable: position sizing. Indian retail traders consistently bet too much on each trade, and the math of compounding losses takes care of the rest.

Why position sizing matters more than your entry

On 21 July 2025, Zerodha's Nithin Kamath shared a now-widely-cited post about an experiment by Elm Wealth called the "Crystal Ball Challenge" (Haghani & White, September 2024). 118 young adults trained in finance were given the front page of the next day's Wall Street Journal — 36 hours before the news broke — and asked to trade the S&P 500 and 30-year Treasuries with that information. They had a literal information edge.

Roughly half lost money. One in six went bust. Their direction calls were right 51.5% of the time, slightly better than a coin flip. A separate group of five seasoned macro traders — the head of trading at a top-five US bank, the founder of a top-ten macro hedge fund, a senior trader at a top-ten macro fund, a former senior government-bond trader at a top-three US primary dealer, and a former senior Jane Street trader — played the same game and grew their starting wealth by 130% on average, with a median gain of 60%. The experienced group also picked direction better (63% vs 51.5%), but as the Elm Wealth authors put it, "mostly we ascribe the dramatically different results to the much more rational trade-sizing displayed by the experienced traders."

The implication is sharp. Even if you somehow knew the news, you would still lose money if you sized positions like most retail traders do. Sizing is the first lever, not the last.

The math of ruin — why "small" feels too small

"Risk of ruin" is the probability that a string of normal losses takes your account below the point of practical recovery. It depends on three things: your per-trade risk, your win rate, and your reward-to-risk ratio.

Two strategies with identical positive expectancy can have wildly different survival profiles depending on how much you risk per trade. Consider a system with a 55% win rate and a 1.5:1 average win-to-loss ratio:

A 10-trade losing streak sounds extreme, but with a 55% win rate it has a roughly 1-in-300 probability over any 30-trade window. Over a year of active trading, that is not a tail event — it is an expected one. A 50% drawdown requires a 100% gain to recover, which is why aggressive sizing is mathematically self-defeating even with a real edge.

The first job of position sizing is not to maximise returns. It is to keep you in the game long enough for your edge to show up.

Three frameworks for Indian retail traders, ranked by where they actually work

1. Fixed fractional (the 2% rule) — start here

Risk a fixed percentage of equity per trade, usually 1–2%. On a ₹5,00,000 account, 1% means ₹5,000 of risk per trade — not a ₹5,000 position. The position can be much larger; only the distance to your stop-loss caps the actual money at risk.

Worked example: you want to swing-trade Reliance at ₹2,400 with a stop at ₹2,350. Risk per share is ₹50. With ₹5,000 of allowed risk, you can buy 100 shares (₹2,40,000 of exposure on a ₹5L account). If Reliance hits your stop, you lose 1% of equity. If it doubles, you make many multiples of that.

The fixed fractional approach is the default professional rule because it adapts automatically: a 10% drawdown reduces your next bet, and a 20% gain increases it. It also forces you to define a stop before you enter — which by itself eliminates half of the typical retail mistakes.

2. Volatility-based sizing (ATR method) — better for derivatives

The flaw in fixed-percentage sizing is that it treats every stop the same. A ₹50 stop on Reliance and a ₹50 stop on a small-cap with a daily range of ₹15 are not comparable trades. Volatility-based sizing fixes this by anchoring your stop to the asset's Average True Range (ATR).

The formula:

Position size = (Account × Risk %) ÷ (ATR × Multiplier)

Practical Bank Nifty example. Capital ₹10,00,000, willingness to risk 1% (₹10,000) per trade. Bank Nifty 14-day ATR is around 600 points. Using a 1.5× multiplier gives a stop distance of 900 points. With a current lot size of 30, the rupee risk per lot is 900 × 30 = ₹27,000 — so you cannot take even a single lot inside your 1% rule. The honest answer is to skip the trade or wait for the ATR to contract. That refusal to overtrade is the entire point.

For a Nifty futures position: with the post-revision lot size of 65 and a normal-volatility 100-point ATR-based stop, a single lot risks 65 × 100 = ₹6,500. On a ₹5L account at 1% risk (₹5,000), even one lot is too much — meaning either your stop has to be tighter or you skip.

Zerodha's "In the Money" newsletter spells out the dynamic version in its sizing series: trade 10 lots when volatility is normal and your stop is 100 points away — risk ₹65,000 — and when volatility rises and the stop widens to 200 points, automatically cut to 5 lots so the rupee risk per trade stays constant. That is the discipline that institutional desks have built in and most retail traders ignore.

3. Kelly Criterion — useful as a ceiling, dangerous as a floor

The Kelly Criterion calculates the mathematically optimal bet size for maximum long-run capital growth:

Kelly % = W − (1 − W) / R

where W is win rate and R is win/loss ratio.

For a strategy with W = 0.55 and R = 1.5: Kelly % = 0.55 − 0.45/1.5 = 25%. That is 25% of capital per trade. Mathematically correct. Practically suicidal — a single 5-trade losing streak at 25% risk compounds to roughly a 76% drawdown.

This is why almost every serious practitioner uses fractional Kelly, typically one-quarter to one-half. Quarter Kelly on the same strategy gives 6.25% per trade — still aggressive, but survivable. The deeper point is that Kelly is exquisitely sensitive to the inputs: if your real win rate is 50% instead of 55%, full Kelly drops from 25% to under 17%. Most retail traders overestimate their edge, so Kelly's output is almost always too high.

Use Kelly the way professional gamblers do: as a hard upper bound. Whatever your sizing method says, do not exceed half of what Kelly says.

A concrete Bank Nifty intraday example

Setup: ₹8,00,000 capital, willing to risk 1% (₹8,000) on one intraday short straddle. A Bank Nifty 0-DTE ATM short straddle requires roughly ₹3.25 lakh of margin per lot. From your own backtest, the realistic intraday max drawdown per lot is ₹20,000. Applying the common "double the backtest drawdown" rule of thumb that practitioners use to account for slippage and regime shifts, plan for ₹40,000 per lot.

Bare-minimum capital allocation per lot: ₹40,000 (drawdown buffer) + ₹3,25,000 (margin) = ₹3,65,000. On an ₹8L account, this realistically supports one lot, not two — even though margin alone says you could take two.

This is the discipline most retail F&O traders skip: they look at available margin (which would say "you can trade two lots") instead of risk-of-ruin math (which says "one lot, or skip the day"). It is also consistent with what SEBI's July 2025 study reported — "the number of overall unique traders dealing in F&O saw a 20 per cent decline compared to the previous year" once tighter lot-size rules and risk-coverage requirements forced this same math on everyone.

A concrete equity swing trade example

Capital ₹3,00,000. Target: buy HDFC Bank at ₹1,640 with a swing-low stop at ₹1,595. Per-share risk: ₹45.

• 1% fixed risk: ₹3,000 ÷ ₹45 = 66 shares. Position size ₹1,08,240 (36% gross exposure).
• 2% fixed risk: ₹6,000 ÷ ₹45 = 133 shares. Position size ₹2,18,120 (73% gross exposure).
• ATR-adjusted: if HDFC Bank's 14-day ATR is ₹30 and you use a 1.5× multiplier, the volatility-true stop is ₹45 — which matches. The two methods converge here, which is a sign the trade is well-structured.

Notice what happens at 2%: a single trade ties up almost three-quarters of your capital. If you take three simultaneous trades at this level, you are effectively running 200%+ gross exposure — exactly the kind of correlated bet that turns one bad earnings night into an account-ending event.

Mistakes Indian retail traders make repeatedly

• Confusing position size with risk. A ₹2 lakh position is not a ₹2 lakh risk if your stop is 2% away.
• Sizing by available margin. Brokers happily extend leverage you should never use. The exchange tells you what is allowed; your sizing rule tells you what is sensible.
• Doubling down on losers. Averaging into a losing trade is the fastest known route to ruin, because it violates the one principle that keeps you alive: smaller bets when the evidence weakens.
• Ignoring correlation. Five "different" trades on five Nifty IT stocks is one trade. Size accordingly.
• Letting wins inflate the next bet. Recompute your rupee risk against current equity after every closed trade, not against the original capital.
• Skipping the trade-log. You cannot run Kelly or even an honest 2% rule without an accurate win-rate and average-win-to-loss number. Without a log, every sizing method is a guess.

Where AI-driven backtesting fits in

The reason these frameworks are now usable by retail traders — and not just hedge funds — is that no-code backtesting platforms can compute your real win rate, expectancy, and drawdown distribution on actual NSE/BSE data with realistic transaction costs and slippage.

WelthWest's backtesting engine and AI Screener, for example, let you test a strategy across Nifty, Bank Nifty, and 500+ Indian stocks across 10+ years and read off the Sharpe, Sortino, and max drawdown that your sizing rule needs as inputs. Pair that with our notes on AI risk management and the existing piece on five common backtesting mistakes, and you have the ingredients for a sizing rule grounded in your data, not someone else's anecdote.

One practical workflow: backtest your strategy, take the realistic drawdown number, and assume your live drawdown will be 1.5× to 2× greater than the backtested drawdown (QuantVPS's backtesting research summarises this multiplier from slippage and regime shifts). Then size positions so that doubled drawdown represents no more than 15–20% of equity. That single discipline puts you ahead of most retail F&O participants in India today.

Frequently asked questions

How much should an Indian retail trader risk per trade?

For most accounts, 1% of equity per trade is the right starting point. New traders or anyone in a drawdown should drop to 0.5%. Only traders with a logged, live win-rate above 55% and a 2:1 average win-to-loss ratio over at least 100 real trades should consider the full 2%.

Is the Kelly Criterion safe for F&O traders?

Full Kelly is not safe for anyone, and particularly not for options traders, where the actual win and loss distributions have fat tails that the formula does not capture. Use quarter-Kelly as a ceiling and combine it with a fixed-percentage rule for the floor.

How is ATR-based sizing different from the 2% rule?

The 2% rule fixes how much money you risk; ATR-based sizing fixes how many "units of volatility" you risk. ATR sizing automatically takes smaller positions in volatile assets (small caps, expiry-day Bank Nifty) and larger positions in calm ones. For derivatives traders, this is the better default.

Can I use the same sizing rule for equities and options?

Not directly. Equity stops translate linearly to rupee loss; option premiums do not. Theta, vega, and IV crush can wipe far more premium than the underlying's ATR implies. For options, size off premium-at-risk and assume your stop will slip by 20–30% in stressed conditions.

What is risk of ruin and how do I calculate it?

Risk of ruin is the probability your account drops below the threshold where recovery is unrealistic — often defined as a 50% drawdown. The simplest closed-form formula uses win rate, payoff ratio, and risk per trade; Monte Carlo simulators give more realistic estimates because they account for losing streaks. Free calculators from BacktestBase and Switch Markets can model your specific numbers in under a minute.

What's the single most important rule?

Never let one trade exceed your planned per-trade risk. Not because of a "magic strategy" but because every account that has been wiped out in Indian markets in the last decade had at least one trade that broke this rule.

The bottom line

The SEBI numbers tell us 91% of Indian F&O traders lose money. The Elm Wealth experiment tells us they would still lose if they could read tomorrow's newspaper. The common thread is not bad analysis. It is bad sizing.