AI for option trading in India: use it to inspect risk assumptions

Begin with the position you already defined

A useful educational request asks the assistant to explain the quantities, premium signs, expiry assumptions and missing information in a supplied worksheet. Keep the task bounded. Asking it to select a position would mix the calculation with a market recommendation, which this guide does not provide.

For option maths, small input mistakes change the answer: a premium per unit is different from a cash total, a lot count is different from units, and an expiry payoff is different from a price before expiry. Have the assistant restate those definitions before calculating anything.

The options strategy payoff tool is a separate calculation page. Here we inspect a fictional worksheet so every result can be independently recomputed. No brokerage, statutory charge, tax or margin estimate is embedded in the example.

A fictional credit spread worksheet, before valuation

This assumed four-leg iron condor is only an arithmetic fixture. All legs have the same fictional expiry and equal quantities. The quantity of 50 units is a teaching assumption, not taken from any NSE lot-size circular. On each side, the short and long strike prices are a made-up 40 points apart; no actual security, strike or expiry is selected.

Short-side premiums received per unitAuthor assumptions; no observed market quote.

22 + 20 = 42

Protective-side premiums paid per unitAuthor assumptions for the two outer legs.

16 + 14 = 30

Opening credit

42 − 30 = 12 per unit

Gross maximum expiry gain

12 × 50 = ₹600

Gross maximum expiry lossRequires the complete equal-width position to be maintained through the modelled settlement.

(40 − 12) × 50 = ₹1,400

Payoff range under the simplified assumptions

−₹1,400 to +₹600, before costs

Source: [5]

Explain the expiry formula without inventing probabilities

The short iron condor described by the Options Industry Council has inner short options and protective outer options sharing an expiry. For the equal-width structure in this worksheet, net credit sets the gross maximum gain and width less credit sets the gross maximum loss at expiry. We use only the generic payoff relationship; the source's US exercise examples are not imported as Indian settlement rules.

In the fictional centre region, all legs expire without intrinsic liability, leaving the credit of 12 per unit. At an assumed expiry price 30 points beyond an inner boundary, but before the outer boundary, the affected spread has an intrinsic liability of 30 per unit. The worksheet result is then (12−30) × 50, or −₹900. Beyond the outer boundary the equal-width wing's liability is capped at 40 per unit in this simplified complete-position model.

That formula does not supply the chance of landing in any region. The opening credit is not an expected return, and the distance between boundaries is not a probability measure. An assistant needs an explicit model and inputs to estimate a distribution; inventing a percentage would add an unsupported claim.

Nor does ₹1,400 mean the broker will reserve exactly that amount. Required margin, execution sequencing and settlement obligations are separate questions. This page deliberately avoids treating a payoff limit as a verified funding requirement.

Expiry payoff and pre-expiry repricing answer different questions

All observations below are author assumptions. No prices were fetched, and no implied-volatility pricing model was fitted.

Fictional scenario	Declared assumption	Gross worksheet P&L	What remains unknown
At expiry in centre region	No intrinsic spread liability	12 × 50 = +₹600	No probability of this outcome is estimated.
At expiry beyond an inner boundary	Affected spread liability 30 per unit	(12 − 30) × 50 = −₹900	This is a terminal-payoff assumption, not a live mark.
Calm pre-expiry / lower-IV illustration	Assume total package closes for 5 per unit	(12 − 5) × 50 = +₹350	The close price of 5 per unit is assumed; it is not derived from an IV change.
Adverse pre-expiry illustration	Assume package close cost 30 per unit	(12 − 30) × 50 = −₹900	Bid/ask, fill availability and residual legs are absent.
Complete-position terminal outer region	Wing liability 40 per unit	(12 − 40) × 50 = −₹1,400	Costs and failures to maintain the modelled legs are excluded.

Payoff method: payoff source. Every scenario input and rupee result is illustrative, calculated by the author. The two −₹900 rows use different kinds of inputs despite sharing a cash result.

Source: [5]

Why an IV-crush story needs a valuation model

Implied volatility, or IV, is a valuation input inferred from option prices. A sentence saying volatility falls does not uniquely determine a package's closing value. To calculate one, preserve the underlying observation, each contract's identity, remaining time, model convention and relevant financing or distribution assumptions.

For that reason the lower-IV row above explicitly assumes a closing cost of 5 per unit. It demonstrates the cash arithmetic if that price were available. It does not prove that a given IV fall would produce that price, or that every leg could be closed together. The same discipline applies when a gap scenario is described with a single package number.

Ask an assistant to label each output as an input assumption, a formula result or a model estimate. If it gives a precise close value without stating a model, retain it as unverified rather than quietly feeding it into the next calculation. A spreadsheet can multiply an invented price perfectly.

A full historical test would need licensed observations old enough for an educational example and consistent timestamps across every leg. We have not run that test. A fictional fixture lets us audit arithmetic without pretending to have reconstructed an actual trading session.

A separate local Greek-sensitivity exercise

These assumed portfolio Greeks are signed per-unit sensitivities for a separate hypothetical position, not Greeks estimated for the condor above. Use a small-move Taylor approximation only as an educational calculation. Assumptions: delta −0.10, gamma −0.002, vega −0.50 per one IV percentage point, theta +0.40 per elapsed calendar day; quantity 50.

Assumed underlying moveAuthor-chosen perturbation, not a prediction.

+20 arbitrary price units

Delta term

−0.10 × 20 = −2.00

Gamma term

½ × (−0.002) × 20² = −0.40

Vega term for assumed IV rise of 5 percentage points

−0.50 × 5 = −2.50

Theta term for one elapsed calendar day

+0.40 × 1 = +0.40

Sum per unit

−2.00 − 0.40 − 2.50 + 0.40 = −4.50

Approximate signed change at assumed quantity

−4.50 × 50 = −₹225

Source: [6]

Keep Greek units beside every coefficient

NSE's derivatives workbook defines delta as sensitivity to the underlying price, gamma as the change in delta, theta as sensitivity to elapsed time and vega as sensitivity to volatility. The unit conventions matter when applying those definitions. A volatility fraction and an IV percentage-point change are not interchangeable inputs.

In our separate exercise, the vega convention is explicitly per percentage point and theta is per elapsed calendar day. The approximation includes a quadratic gamma term; it omits changing coefficients, cross-effects and other valuation inputs. It should not be extended into a large-gap loss forecast.

Recompute a test with the sign reversed as a diagnostic. Under the same assumed coefficients and a −20 underlying move, the delta term becomes +2.00 while the gamma term stays −0.40 because the move is squared. With the other terms unchanged, the total becomes −0.50 per unit, or −₹25 for 50 units. This is a useful sign check; it is not a favourable trading scenario being proposed.

If an assistant reports a completely symmetric outcome after that reversal, inspect its formula. If it changes the assumed coefficients during the check, ask it to show the change explicitly. The exercise tests the explanation and arithmetic under frozen assumptions.

An illustrative IST research routine, outside live trading

1. 08:30 IST · before review

   Prepare an old or fictional worksheet

   Choose the research date window, identify units and preserve source observations. Ask for missing fields and formula definitions. These times are an author-designed routine, not market-session rules.

2. 12:00 IST · during review

   Challenge the assumptions

   Compare expiry payoff with an explicitly modelled or assumed pre-expiry value. Change one input at a time. Keep the original response so a later explanation cannot erase an earlier sign or quantity error.

3. 18:00 IST · after review

   Record corrections and unresolved questions

   Save the worksheet version, independent calculations and uncertainty. Review whether the assistant followed the task. Do not convert a retrospective explanation into a live instruction.

Intraday speed adds a timestamp problem

A correct formula can still use incompatible observations. An underlying close and an option premium from different moments do not describe one coherent state. A response generated after a delay is not automatically refreshed because its prose refers to the current market.

For retrospective research, record observation times and retrieval times separately. Preserve whether a value is last traded, a bid, an ask or an assumed package mark. If the source cannot establish those fields, the analysis should carry that limitation into its conclusion. Do not use a general assistant's confident wording to fill missing market data.

Chart interpretation has a related evidence problem. The AI chart analysis guide explains why unreadable labels, cropped timeframes and missing volume cannot support precise conclusions. Combining a screenshot explanation with an option worksheet does not repair those missing inputs.

The AI prompt guide provides broader research prompts. For this task, a useful instruction is to preserve the supplied contract descriptions and units, identify unavailable inputs, show the formula, and stop short of selecting a security or a trade.

Keep a reconciliation column for each quantity conversion. If a worksheet starts with a number of contracts, the multiplier belongs in the calculation explicitly; if it starts with already multiplied cash amounts, multiplying again would double-count exposure. Our fixture starts with per-unit premiums and an assumed unit quantity. It contains no contract multiplier or actual lot count, so those cannot be inferred from the rupee totals.

Requests that exceed this evidence

• Choose an expiry or strike from the example

  The worksheet contains no current contract or market selection. Its invented boundaries serve only the payoff arithmetic.

• Predict an expiry settlement or a profitable day

  No distribution or forecast has been estimated. A Greek is a sensitivity under assumptions, not a prediction of the input change.

• Guarantee the maximum loss as an execution result

  The simplified terminal model assumes the complete position survives as specified. Real costs and mismatched execution need separate treatment.

• Call an assumed IV-crush price a measured result

  The close value of 5 per unit was chosen by the author. A historical claim requires dated leg-level observations and a reproducible valuation method.

• Treat a review assistant as an order control

  The educational workflow does not enforce broker limits or execute trades. A written exit plan and an implemented control are separate mechanisms.

What to retain after checking an AI answer

Keep the input worksheet, the original output, your independent calculation and a short correction note. Distinguish an assistant that found a missing assumption from one that merely restated your preferred outcome. An accurate explanation should remain accurate when the example loses money.

The existing AI options strategy tool page and AI expiry planner page are linked for their separate topics. Their names do not turn this fictional example into a trade suggestion. No tool accuracy or return is claimed here.

SEBI's 17 August 2026 PR No.48/2026 cautions about live strategy sessions and says a person engaged solely in education should not use the preceding thirty days of market-price data or indicate future prices or give security-related advice. Our worksheet uses no observed market data at all. This regulatory description is a plain-English summary, not legal advice.

The useful outcome of this exercise is a transparent account of what the inputs imply: a gross expiry payoff range, separately assumed closing scenarios, and a local sensitivity calculation with declared units. Actual margin, valuation, fees and execution remain different questions. Each needs its own evidence before a real position can be assessed.

Sources

Every figure on this page comes from one of these. Each was checked on the date shown.

1. [1]SEBI: historical FY22–FY24 individual F&O loss study release · checked 14 Oct 2026
2. [2]SEBI: FY25–FY26 profitability study landing page · checked 14 Oct 2026
3. [3]SEBI: PR No. 50/2026, FY25–FY26 equity derivatives studies, 20 August 2026 · checked 14 Oct 2026
4. [4]SEBI: live-strategy caution, PR No.48/2026 · checked 14 Oct 2026
5. [5]OCC Options Industry Council: short iron condor mechanics · checked 14 Oct 2026
6. [6]NSE Academy: derivatives workbook, section 6.4 · checked 14 Oct 2026