Option selling backtests: capital at risk, drawdown, margin
What to measure in an option selling backtest on NIFTY or SENSEX: capital at risk, drawdowns, stops that never fire, and why broker margin is separate.
- By
- Veloce team
- Published
- Reading time
- 9 min read
On this page
Selling options collects premium up front and pays out if the market moves far enough. That shape makes the usual backtest headline, net profit and win rate, a poor guide. What matters is how much the positions could have lost at any moment, how deep and long the drawdowns were, whether the stops could actually fire, and how much capital the broker would have blocked. This guide sets out what to measure in an option selling backtest, works through an iron condor with Veloce’s own risk code, and is explicit about what Veloce does not model: it does not compute SPAN margin. It is about measurement, not about whether selling options is profitable.
The shape of a selling strategy
A short option position earns a little most of the time and loses a lot occasionally. In a backtest, that shows as:
- a high win rate,
- a smooth equity curve between losses,
- a few large losing trades or days that undo weeks of gains.
None of those is a problem in itself. The problem is reading the first two without the third. A strategy that wins 80% of the time with an average win of ₹1,500 and an average loss of ₹6,500 loses ₹100 a trade on average before costs: 0.8 × 1,500 − 0.2 × 6,500 = 1,200 − 1,300.
So the first measures to read together are:
| Measure | What it shows |
|---|---|
| Win rate | How often trades close at a profit |
| Average win and average loss | The size of the typical outcome on each side |
| Largest loss against average win | How many winners one bad trade erases |
| Profit factor | Gross profit divided by gross loss, after charges |
| Exit reasons | How many trades hit the target, the stop, or the time exit |
Capital at risk
Capital at risk is the most the open positions could lose if every option expired at the worst possible price. Veloce computes it after every fill:
- Positions are grouped by underlying and expiry.
- For each group, it finds the worst loss at expiry across every possible settlement price, given the premiums paid and received.
- Long futures add their cost. A short future, or a sold call not covered by a bought call, makes the loss unlimited, and the capital at risk is reported as unlimited at that moment.
- The report charts it over time, and a sweep reports the peak capital at risk of each run, or “unlimited”.
The figure uses fill prices, not charges, so it is the payoff risk of the position as held.
A worked iron condor
Take a NIFTY weekly iron condor as Veloce’s reference spec builds it: sell the put and the call four strike steps from the money, buy the put and the call eight steps away. NIFTY strikes are 50 points apart, so with NIFTY at 26,000 the legs are:
| Leg | Strike | Premium (open) | Fill with one tick slippage |
|---|---|---|---|
| Buy put | 25,600 | ₹36.90 | ₹36.95 |
| Buy call | 26,400 | ₹46.25 | ₹46.30 |
| Sell put | 25,800 | ₹83.20 | ₹83.15 |
| Sell call | 26,200 | ₹98.20 | ₹98.15 |
The premiums are illustrative, priced with a Black-Scholes model at 12% volatility with seven days to expiry. The lot is 65 units.
| Measure | Value |
|---|---|
| Net credit per unit | 83.15 + 98.15 − 36.95 − 46.30 = ₹98.05 |
| Net credit per lot | ₹6,373.25 |
| Wing width | 200 points |
| Capital at risk (Veloce’s exposure code) | (200 − 98.05) × 65 = ₹6,626.75 |
| Charges on the four entry fills | ₹119.47 |
Compare three structures built from the same illustrative chain:
| Position | Capital at risk per lot |
|---|---|
| Iron condor above | ₹6,626.75 |
| Short straddle at 26,000, hedged with the 25,800 put and 26,200 call | ₹2,382.25 |
| Short straddle at 26,000 with no hedges | unlimited |
| Condor with the call hedge removed | unlimited |
Removing one hedge does not make the risk a bit bigger; it makes it unbounded on that side. Veloce shows that as a gap in the capital-at-risk chart and “unlimited” in a sweep.
The ratio that matters
The condor risks ₹6,626.75 to keep at most ₹6,373.25. With the reference spec’s take profit at half the credit, it aims to make ₹3,186.63 against a worst case of ₹6,626.75. If every losing trade lost the maximum, the target would need to be hit in 6,626.75 ÷ (6,626.75 + 3,186.63) ≈ 68% of trades just to break even before charges. Real losses are usually smaller than the maximum, which is exactly what a backtest measures: the distribution of actual losses against this worst case.
A stop that can never fire
The reference iron condor spec has a stop at two times the credit. On this condor, that is a loss of 2 × ₹6,373.25 = ₹12,746.50, but the position can lose at most ₹6,626.75. The stop can never trigger; the condor’s maximum loss is reached first.
The general rule for a condor with wing width W points and credit C points: a stop at k times the credit can fire only if k × C is less than W − C, that is if C < W ÷ (k + 1). For a stop at two times the credit and 200-point wings, the credit must be below 66.67 points.
That depends on volatility and days to expiry. At lower volatility or further strikes, the credit is smaller and the same stop becomes reachable. A backtest across months will contain both kinds of weeks. Check the exit reasons: if no trade ever closed on the stop, find out whether the market never got there or the stop could not.
Drawdown
Maximum drawdown is the largest fall in account equity from a peak. For selling strategies, read it in three ways:
- Against the average credit. A drawdown of ₹25,000 on a strategy collecting ₹6,000 a week is about four weeks of full credits.
- Against peak capital at risk. If the drawdown exceeds the peak capital at risk of any single position, losses came in a cluster: several bad weeks in a row, or a re-entry after a loss that lost again.
- In time. How many weeks from the peak to the recovery? A selling strategy can spend months recovering from one bad expiry.
Gaps and fills
Stops in a backtest are only as good as the prices they fill at. Veloce checks exits on each candle and fills at the next candle’s open. If NIFTY gaps overnight, the first candle of the morning opens at the gapped price and the exit fills there, beyond the stop level. Positions held overnight show their gap losses in full. Look at the largest single-day losses and check how many came from overnight gaps rather than intraday moves.
Margin: what Veloce does not model
Brokers block initial margin for short options: the SPAN margin computed by NSE Clearing’s portfolio risk model plus an exposure margin. SPAN evaluates a set of price and volatility scenarios across the whole portfolio, so hedged positions need less than naked ones, and the figure moves every day with volatility. Zerodha Varsity’s iron condor chapter shows the margin benefit of hedges against a short strangle; its figures are dated, but the effect is the point.
Veloce does not compute SPAN or exposure margin. In a backtest:
| Rule | What it does |
|---|---|
| Cash | Buys only what the cash on hand pays for. Sells are not limited by it. |
| Futures margin | Blocks a flat share of each futures lot’s value (10% by default) out of equity. Futures only. |
| Max loss at expiry | Cuts each new option position so its worst loss at expiry stays within a share of equity, such as 2%. |
| Daily loss limit | After a set rupee loss in a day, only closing trades are allowed. |
So the capital you set for a backtest is the account the P&L is measured against, not a margin requirement. Returns in the report are on that capital. A broker might block more or less than the position’s capital at risk, and that number changes with market conditions. Before relying on a size, check the margin for the same legs in your broker’s calculator on a few dates, and compare it with the backtest’s peak capital at risk.
Sizing from capital at risk
A max-loss-at-expiry rule of 2% on ₹10,00,000 sets a budget of ₹20,000. The rule checks each new option leg against everything already held in the same expiry, at every strike where the payoff can turn, and cuts the leg’s lots so the combined worst loss at expiry stays within the budget. It also never sells more calls than the calls already bought in that expiry, which is why specs list hedges first. For the illustrative condor at ₹6,626.75 a lot, a ₹20,000 budget covers three condors’ worth of worst-case loss (₹19,880.25), not four. The rule works from the payoff alone; it does not know what margin a broker would charge.
Costs that weigh on selling
Selling pays securities transaction tax on the premium sold, 0.15% from 1 April 2026 (0.1% before). A condor sells two legs on entry and, if closed early, sells the two long legs on exit. Charges on the four entry fills of the example came to ₹119.47, and a full round trip of all four legs costs more. For a strategy taking half the credit as profit, charges and slippage take a visible share of each winning trade. The charges guide itemises them.
What to measure, in one list
| Measure | Why |
|---|---|
| Peak capital at risk, and any unlimited periods | The real worst case of the positions you held |
| Net P&L divided by peak capital at risk | Return on the risk actually taken |
| Largest loss against average win | How fragile the win rate is |
| Exit reasons | Whether stops and targets fire, and how often |
| Whether each stop is reachable | Credit and wing width decide it |
| Maximum drawdown in rupees, against credits and in weeks | What you would have sat through |
| Largest single-day losses, and gaps among them | Overnight risk |
| Charges as a share of gross profit | STT on sells and slippage on four legs |
| Results across volatility regimes | Credits and stop reachability change with volatility |
| Broker margin on sample dates | Not modelled; check it separately |
Questions traders ask
Is a weekly or monthly expiry safer to sell? Neither is safer in general. Weekly options decay faster but react more sharply near expiry; monthly options collect more premium per trade but are held through more news. Test both with the same capital-at-risk budget and compare drawdowns, not just totals.
Should I hold to expiry? Holding to expiry exposes the position to the final day’s moves and to settlement. Veloce’s reference iron condor exits one day before expiry. The stop, trailing stop and time exit guide compares the choices, and the December session guide explains how expiry settlement changes from 7 December 2026.
Does a high win rate mean low risk? No. It often means the opposite: the strategy is paid for taking rare large losses. Read the largest loss and the capital at risk.
Can I compare a straddle with a condor on return alone? Compare them on return per rupee of peak capital at risk and on drawdown. A naked straddle’s capital at risk is unlimited, so no return figure makes the comparison fair.
Try it in Veloce
Veloce backtests option selling on NIFTY 50 weekly, BANK NIFTY monthly and SENSEX weekly options on one-minute data, with fills at the next candle’s open, itemised charges, a capital-at-risk chart after every fill, and peak capital at risk in every sweep row. The reference iron condor and hedged short straddle are ready to run, and the risk and exits page describes each rule.
Create a free Veloce account and read the capital-at-risk chart before the net P&L on your next selling backtest. For sizing, see the drawdown and position sizing guide.
Frequently asked questions
- How much capital do I need to sell options on NIFTY?
- There are two different numbers. The broker blocks margin set by the clearing corporation's risk model, which changes with volatility and hedges. The strategy's capital at risk is the most its open positions could lose at expiry. A backtest can measure the second exactly; the first you check with your broker's margin calculator.
- What is capital at risk in Veloce?
- After every fill, Veloce computes the most the open positions could lose if every option expired at the worst possible price, per underlying and expiry, plus the cost of any long futures. If a sold call or a short future leaves the loss unlimited, it reports the capital at risk as unlimited for that moment.
- Does Veloce model SPAN margin?
- No. Veloce does not compute SPAN or exposure margin. Its cash rule limits only what you buy, a futures margin rule blocks a flat share of each futures lot's value, and a max-loss-at-expiry rule caps each new option position's worst loss as a share of equity.
- Why is the win rate of option selling misleading?
- Selling collects many small gains and occasionally takes a large loss. A 75% win rate says nothing on its own: if the average loss is four times the average win, the strategy loses money. Read win rate together with average win, average loss and the largest loss.
- Can a stop-loss on a credit spread fail to trigger?
- Yes, if it is set beyond the most the spread can lose. On an iron condor, a stop at two times the credit can only fire if the credit is less than a third of the wing width. Above that, the maximum loss is reached first and the stop never triggers.
- What is a good drawdown for an option selling strategy?
- There is no good number in general. Compare the maximum drawdown with the average credit collected, with the peak capital at risk and with the capital you would actually commit, and check how long the strategy took to recover.
Sources
- NSE: Equity derivatives margins (SPAN and exposure margin)nseindia.com, accessed
- Zerodha Varsity: Iron condor, new margin frameworkzerodha.com, accessed
- Union Budget 2026-27, Memorandum explaining the provisions of the Finance Bill (STT on options and futures)indiabudget.gov.in, accessed
- NSE circular NSE/FAOP/70616: revision in market lot of derivative contracts on indices (October 3, 2025)nsearchives.nseindia.com, accessed