Option Strategies & Hedging Exposures

Ring: Finance-Plus (Derivatives branch) · Node DV1.03 · CROWN · ~18 focused hours · Mastery gate ≥ 90%

You now own the machinery. DV1.01 gave you forwards, futures and swaps from cost-of-carry; DV1.02 gave you put-call parity, the binomial tree, Black-Scholes-Merton and the Greeks. All of that is a price. This module is about a different verb: use. A trading desk, a corporate treasury, a pension plan and a concentrated private investor reach for the same six or seven structures and the same three hedge-sizing formulas, and the whole difference between a professional and an enthusiast is whether they can state, before trading, exactly what the position pays at every price, exactly how many contracts it takes, and exactly how much risk is left over when the hedge is on. That last quantity (the residual) is what this node is about. There is no perfect hedge outside a textbook; there is only a hedge whose leftover risk you have measured and accepted.

This Ring is additive institutional/CFA breadth, it presents the toolkit and the plumbing alongside, not in place of, the program's process-over-P&L, concentration-over-diversification philosophy. Nothing here asks you to trade options. Phase 9 taught that survival is the first-order goal and that a concentrated book of understood businesses beats a diversified book of guesses; this module hands you the instruments a serious analyst must be able to read, on a counterparty's balance sheet, in a treasury policy, in a fund factsheet, in the footnote where M1.09 put hedge accounting, and, occasionally, use. A collar on a concentrated legacy holding is a Phase-9 sizing decision executed with a Phase-DV instrument. That is the honest relationship.

The module in one sentence: a hedge does not remove risk, it exchanges one risk for a smaller, differently-shaped one, and the entire professional skill is measuring what you exchanged it for. The equity hedger swaps market risk for basis and rounding risk. The delta hedger swaps direction for gamma; the delta-gamma hedger swaps gamma for vega and theta. The FX hedger swaps translation risk for carry, or for the onshore-offshore basis. In every case a number can be put on the leftover, and in every case here you will put one there: computed, not asserted.

One standing warning (R10). This is the highest AI-over-trust region in the corpus. Option arithmetic looks like something a language model does well (formulaic, symbolic) and a model will produce a confident payoff table with breakevens to two decimals in one second. It is also arithmetic where a sign error, a multiplier error, or a strike-versus-forward confusion yields an answer that is plausible, internally consistent, beautifully formatted, and off by a factor of a hundred. A wrong hedge ratio does not fail loudly like a wrong DCF; it sits quietly in the book until the market moves. Every number here is one you must be able to produce by hand, and the gate is 90% for exactly that reason. AI is an amplifier and nothing else: it can build the grid, it cannot own the sign.

Case Lab: run case CL-DV.1 (Hedging a Nifty Book Through Earnings) in the app's Cases tab, read a short-straddle book's Greeks, size a whole-lot Nifty futures delta hedge, and reconcile what the hedge did and didn't cover once earnings hit, the same delta- and delta-gamma-hedging machinery Worked Examples 4–6 build below. Attempt it after Worked Example 6, blank sheet, before checking the model answer.

Learning objectives

By the end you can:

1. Write the payoff and profit function of every core structure from memory, long/short call and put, covered call, protective put, collar, bull and bear verticals (debit and credit), straddle, strangle, butterfly, and for each compute the breakeven(s), maximum profit and maximum loss as explicit arithmetic, not a remembered shape.
2. Map a market view onto a structure using the direction × volatility grid: what you own when bullish and volatility is cheap is not what you own when bullish and volatility is rich.
3. Use put-call parity as the grammar of strategy, an equal-strike collar is a synthetic forward; a bull call spread and a bull put spread differ only by financing; a covered call is a synthetic short put, and use each identity as an arithmetic cross-check on your own payoff table.
4. Convert an option's Greeks into a position's Greeks, applying sign, quantity and multiplier correctly, and read the result as a description of what the book is actually exposed to.
5. Delta-hedge a book and run it through a price move, computing the exact residual P&L, decomposing it into gamma and theta terms, and showing numerically that a delta hedge is first-order and fails second-order, and how fast.
6. Delta-gamma hedge with a second option, solving the two-equation system, shrinking the residual by an order of magnitude on the same move, and then naming and pricing the vega and theta you took on in exchange.
7. Size an equity hedge with index futures (beta adjustment, contract rounding, target-beta transformation) and decompose the realized outcome into index, beta-error, idiosyncratic, rounding and basis components.
8. Size an interest-rate hedge from DV01/BPV, using a futures conversion factor and a yield beta, or a swap notional, and quantify the residual left by convexity, yield beta and rounding.
9. Hedge an FX exposure four ways (forward, bought option, zero-cost risk reversal, participating forward) computing the CIP forward, the effective floor net of premium carry, the zero-cost strike, and the crossover rate.
10. Quantify basis risk. Derive and apply h* = ρ·σ_S/σ_F, show it is numerically the same object as beta, compute hedge effectiveness = ρ² and residual volatility σ_S√(1−ρ²), and state in currency what the leftover costs.
11. Price the hedge (the annualized premium drag of a standing put program, the opportunity cost of a covered call) so that "should we hedge?" becomes a comparison of two numbers rather than a temperament.
12. (Productivity objective: R10 duality.) Build the payoff grid, the Greeks table and the hedge-ratio regression in Python/pandas once the by-hand skill is installed, and verify every machine figure against arithmetic you can do yourself.

The duality, stated once (R10). The gated skill is the by-hand one: payoff algebra, hedge-ratio arithmetic, residual decomposition. The productivity payoff is the vectorized version. The order is not a matter of taste, this is the region where a confident wrong number is least likely to look wrong. Verification is not optional here; it is the skill.

Prerequisites & connections

Builds on. DV1.02 is the direct parent and all of it is assumed: parity, risk-neutral valuation, BSM and its assumptions, and the five Greeks computed and interpreted. Nothing here re-derives a Greek, you arrive able to produce delta, gamma, vega, theta and rho, and this node spends its time on what you do with them. DV1.01 supplies the cost-of-carry forward, futures mark-to-market and covered interest parity the FX and index hedges lean on. FI1.02 owns duration, DV01/PVBP and convexity; the rate section uses BPV as a known quantity and quantifies the hedge residual. M3.02 gave you beta and the volatility definition of risk, which the beta-adjusted hedge operationalizes; M9.01 gave you sizing, the impairment definition of risk, N_eff, and the anti-ruin overlay, it decides whether to hedge a concentrated position, this one only decides how. M1.09 owns hedge accounting: designation, effectiveness testing, and whether gains land in OCI or P&L; this module computes the economics and points there for the accounting. AI0.01/AI0.06 supply the Primary-Source Guardrail governing every verify step below.

Feeds forward. FI1.04 is the fixed-income twin: a callable bond's issuer option is a short call you did not choose, and its negative convexity is the same second-order problem gamma hedging solves here. AA1.02/AA1.03 inherit hedge effectiveness as a risk-budget and attribution input. PW1.01 uses the collar as one of two mechanical routes out of a concentrated position and adds the tax layer this module deliberately leaves alone. AL1.01 reads option overlays in a fund's return stream. The quant-code region builds the vectorized payoff engine and hedge-ratio regression sketched below.

Two sibling nodes own the material this one deliberately stops short of, and both are worth reading next. DV1.04 owns swaptions, the Black model for options on futures and rates, caps and floors, and the currency and equity swaps that sit alongside the rate swap. Every structure here is written on an equity or an index; the moment the underlying is a rate or a swap, the pricing convention changes, and DV1.04 is where it changes. DV1.05 owns the exchange-traded layer this module keeps thin. That means the margin regime, expiry-day mechanics and settlement, the taxation of derivative gains in both markets, portfolio-level currency management, and volatility treated as an asset class rather than as an input. Section 12b's calendar spread and section 13's contract mechanics both hand off there.

What you need in hand. A calculator or spreadsheet, the BSM formula from DV1.02, comfort with max(·, 0), and the habit of writing a position line by line before writing the total.

Core teaching content

4. Put-call parity as the grammar: and as your cross-check

With a continuous dividend yield, c − p = S₀e^{−qT} − Ke^{−rT}. Three consequences do most of the work.

(a) An equal-strike collar is a forward. Set K_p = K_c = K: you end with exactly K whatever happens. Net premium is p − c = Ke^{−rT} − S₀e^{−qT}, so outlay today is S₀ + p − c = Ke^{−rT} + S₀(1 − e^{−qT}), the second term being precisely the present value of the dividends you will collect. Net of dividends you paid the present value of K and will receive K with certainty: a risk-free bond. Worked example 1 checks this to four decimals, and it is the single best proof your payoff table is right, if a zero-width collar does not price to the risk-free rate in your spreadsheet, the spreadsheet is wrong.

(b) A bull call spread and a bull put spread are the same risk. Subtracting parity at K₁ from parity at K₂ gives call-spread debit + put-spread credit = (K₂ − K₁)e^{−rT}. Identical payoff shapes; they differ only in whether you pay now and receive later or the reverse, which is worth exactly the interest on the strike width, so their breakevens differ by (K₂ − K₁) − [D + C].

(c) A covered call is a synthetic short put. Long stock plus short call at K has the profit shape of a short put at K, plus financing. If you would not sell the naked put at that strike, do not sell the covered call. "Covered" describes the collateral, not the risk.

Use parity as a check, always. Before trusting any payoff table, collapse it to a case where the answer is known: set both strikes equal, set a strike to zero, send volatility to zero. If the degenerate case is wrong, the general case is wrong.

8. Hedging an equity exposure: sizing, rounding, target beta

To hedge a portfolio of value V_P and beta β_P with an index future at price F and multiplier m:

N = − β_P · V_P / (F · m)

One line of derivation: the portfolio moves β_P for every 1% of index, so its index-equivalent exposure is β_P·V_P; each contract carries F·m; divide. To move to a target beta, N = (β_target − β_P) · V_P / (F · m), the same expression, with β_target = 0 recovering the full hedge and β_target > β_P giving a positive N, i.e. going long futures to lever market exposure without buying stock.

Three residuals survive, and a professional names all three before trading:

• Rounding. N is rarely an integer. Worked example 6 needs 120.86 contracts and trades 121, leaving 0.14 contracts: ₹2.68 lakh of notional: unhedged.
• Beta error and idiosyncratic return. β_P is a historical estimate and the portfolio is not the index. In WE6 the index falls 12.00%, so a beta-1.22 book "should" fall 14.64%; it falls 16.10%. That −1.46% gap is ₹27.0 lakh, and no index hedge can touch it, it is stock-specific risk, the risk you presumably wanted to keep.
• Basis. The futures price is not the index. Basis starts at +100 points and ends at +26, a −74 point move flowing straight into hedge P&L: 121 × 75 × 74 = ₹6.72 lakh. Basis converges to zero at expiry, which is why a hedge held to expiry carries less basis risk than one lifted early, and why rolling quarterly means eating the basis four times a year.

9. Hedging an interest-rate exposure: BPV, conversion factors, yield beta

Rate hedges are sized in basis-point value, never notional: BPV = ModifiedDuration × MarketValue × 0.0001 (FI1.02). The hedge instrument's BPV is computed the same way, with one wrinkle for bond futures, the contract tracks the cheapest-to-deliver bond scaled by its conversion factor:

BPV_futures = BPV_CTD / CF and N = − BPV_portfolio / BPV_futures

Worked example 7 hedges a $250mm portfolio of modified duration 6.8 (BPV $170,000/bp) with a contract whose CTD BPV is $73.63 and CF 0.8215, giving futures BPV $89.63 and a hedge of 1,897 contracts short. On a +75bp parallel shift the linear P&Ls offset to within $2,144 on a $12.75mm move. Then reality arrives in three pieces, each a number:

• Convexity. A straight bond portfolio has positive convexity, so the linear BPV estimate overstates the loss on a large move in either direction. At 86.25bp with convexity 62, the portfolio gains ½ × 62 × 0.008625² × $250mm = $576,527 versus linear. On a hedge, positive convexity is a gift, and it is exactly what a callable bond or MBS lacks (FI1.04).
• Yield beta. Corporate yields do not move one-for-one with Treasuries. At a yield beta of 1.15 the unadjusted hedge leaves a −$1,333,829 residual on a 75bp Treasury move: 0.53% of the portfolio, from a parameter that never appears in the textbook formula. Scale the hedge (N = −β_y·BPV_P/BPV_F, short 2,181 rather than 1,897) and the residual becomes +$575,295, which is the convexity gain of $576,527 less $1,232 of rounding. That tie-out is what a verified hedge looks like: every dollar of residual has a name.
• The swap alternative. A pay-fixed swap does the same job with no conversion factor and no CTD switching. A 7-year semiannual swap at a flat 4.00% par rate has annuity 6.0531 years, so PV01 = $605.31 per $1mm per bp; neutralizing $170,000 needs $281mm notional. Cleaner arithmetic, but bilateral or cleared documentation, margin and counterparty exposure. That trade-off, not the arithmetic, is the decision.

For a floating-rate borrower the instruments change, the logic does not. A cap at K pays Notional × max(r − K, 0) × (days/basis) per period: insurance with a premium. A collar funds the cap by selling a floor, trapping the effective rate in [K_floor, K_cap]. A payer swap fixes the rate outright at no premium. The pattern is what you should expect: the cap costs premium and preserves the benefit of falling rates; the collar is free and gives that benefit away below the floor; the swap is free and gives it away entirely (practice item P14).

10. Hedging an FX exposure: four structures and the crossover arithmetic

An exporter with a dollar receivable is long USD/INR. Four ways to deal with it, in increasing cleverness and decreasing certainty:

1. Sell forward. Locks the covered-interest-parity rate F = S × (1 + r_INR·t)/(1 + r_USD·t) on money-market conventions (DV1.01). Worked example 8 computes F = 86.8665 from a spot of 86.40, a 2.17% annualized premium that exists because rupee rates exceed dollar rates. This is the most misunderstood number in Indian corporate treasury: the forward premium is interest arbitrage, not a forecast.
2. Buy a USD put / INR call. A floor with upside retained, at a premium. The honest comparison is K minus the future value of the premium versus the forward: an 86.50 put costing 0.7595 carries to 0.7720, so the effective floor is 85.7280: 1.1385 below the forward, and the option beats the forward only above 87.6385. State that crossover before you buy, or you are not comparing the alternatives.
3. Zero-cost risk reversal (collar). Buy the 86.00 put, sell a call struck to fund it exactly: 87.7848. A band of 86.00–87.78 at zero premium, with everything above 87.78 given away. In FX this is the default corporate structure and is quoted as a single "risk reversal" number.
4. Participating forward. Buy a put at K and sell a call at the same K on part of the notional, netting to zero. At 50% participation the strike solves to 86.2235: floored on the whole amount, keeping half of any move above. A collar whose ceiling is a slope, not a wall.

And the venue choice. The same exposure can go through NSE USD/INR futures ($1,000 per contract, so $6mm is 6,000 contracts), a bank forward, or the offshore NDF. These do not price identically: WE8 puts the NDF at 86.95 against an onshore forward of 86.83, a 12-paise basis worth ₹7.20 lakh on $6mm, often larger than the entire strike-selection decision, and invisible in any treatment that assumes one price.

11. Basis risk, quantified: the minimum-variance hedge ratio

When the hedge instrument does not track the exposure (a mid-cap book hedged with a large-cap index, jet fuel with crude, corporates with Treasuries) you are cross-hedging and the right ratio is not 1. With σ_S, σ_F and correlation ρ, hedging at ratio h gives

σ²_hedged = σ_S² + h²σ_F² − 2hρσ_Sσ_F

Minimizing over h gives the minimum-variance hedge ratio h* = ρ · σ_S / σ_F, and substituting back gives minimum variance σ_S²(1 − ρ²), so:

Hedge effectiveness = ρ² (fraction of variance removed) · Residual volatility = σ_S√(1 − ρ²)

Stare at h* = ρσ_S/σ_F until you recognize it. It is beta, the OLS slope of the exposure's returns on the hedge's returns is Cov(S,F)/Var(F) = ρσ_S/σ_F, and the regression R² is your hedge effectiveness. The minimum-variance ratio and §8's beta hedge are the same object from two directions, which tells you how to estimate the ratio (run the regression) and how to report its quality (report the R²) in one move.

Worked example 9: a 24%-volatility mid-cap portfolio hedged with a 14%-volatility index at ρ = 0.72 gives h* = 1.2343, effectiveness ρ² = 51.84%, residual volatility 16.66%. A hedge a factsheet would call "market-neutral" removes barely half the variance and leaves two-thirds of the volatility standing: ₹3.08 crore of one-standard-deviation risk on ₹18.5 crore, against ₹4.44 crore before. Say that number out loud before calling anything hedged.

Note also the flatness near the optimum: at h = 1.0 residual volatility is 16.98% against 16.66% at 1.2343. The cost of a modestly wrong hedge ratio is second-order; the cost of a wrong correlation is first-order, because ρ sets the floor and no h beats it. Estimating and stress-testing ρ therefore matters far more than fine-tuning h, and the crisis behaviour of ρ (toward 1 among risky assets, toward the wrong sign among hedges) is what actually breaks hedges. M9.01 §4.10 made this point about diversification; it is identical mathematics.

13. Contract mechanics you must look up, never remember

Every figure here is perishable and exchange-set. Treat each as an illustration of a shape and get live values from the exchange's own contract-specification page before sizing anything.

• India (NSE F&O), index options. Nifty and Bank Nifty index options are European-style and cash-settled. Lot sizes are exchange-set and revised by circular, so they are perishable by construction: the worked examples here use a Nifty lot of 75 and a Bank Nifty lot of 30, which are the figures the tables below were built on and which you must verify on the NSE contract-specification page before sizing anything. NSE has since revised the Nifty lot to 65 with effect from the January 2026 monthly cycle (circular FAOP70616, cited in DV1.02's market-anchor block); the arithmetic here is kept on 75 for continuity with the tables, and every rupee figure scales linearly, so multiply by 65/75 to restate any of it on the current lot.
• India (NSE F&O), single-stock options. Single-stock options are European-style, but they are not cash-settled: NSE moved the entire single-stock derivatives segment to compulsory physical settlement from the October 2019 expiry, and an in-the-money position that survives to expiry becomes a delivery obligation. Three consequences change how a strategy is sized. First, delivery margins ramp through the last five trading sessions of the expiry, typically in steps from a fraction of the delivery value to the full value by expiry day, so a spread that was margin-efficient on Monday can demand several times the margin on Thursday. Second, an in-the-money long carries a full-contract-value obligation: exercising a 20-lot in-the-money call on a ₹2,000 stock with a 250-share lot means finding ₹1 crore of cash, not settling a difference. Third, the exchange offers a do-not-exercise facility for options that are in the money only marginally, where the settlement cost would exceed the intrinsic value; using it abandons the intrinsic value deliberately, and forgetting it exists is how a trader accidentally takes delivery of a position they never intended to own. DV1.05 owns the margin regime, the expiry-day mechanics and the tax treatment in full.
• India, other contract terms. USD/INR futures carry $1,000 per contract. Securities transaction tax, exchange charges and stamp duty are real: a strategy whose edge is smaller than its taxes is not a strategy.
• United States. Standard equity options cover 100 shares and are American (early exercise matters, especially around dividends). SPX options are European, cash-settled, $100 multiplier; SPY options are American on 100 ETF shares, not interchangeable despite tracking the same index. Three-month SOFR futures have DV01 of exactly $1,000,000 × 0.25 × 0.0001 = $25.00 per bp; one-month SOFR futures, $5,000,000 × (1/12) × 0.0001 = $41.67. Treasury futures deliver a basket with conversion factors and a cheapest-to-deliver that can switch.
• The universal rule. Before sizing any hedge write down five lines: multiplier, settlement style, exercise style, expiry date and time, margin regime. A hedge sized without them is a guess with a spreadsheet attached.

Modern Data Analysis. All of the above is closed-form, which makes it ideal for vectorization once the by-hand skill is installed. The Python/pandas way: build a NumPy price grid S = np.linspace(0.6*S0, 1.4*S0, 801), write each leg as a vectorized payoff array (np.maximum(S - K, 0) for a long call), stack the legs as columns of a pandas.DataFrame, and let df.sum(axis=1) be the structure, an eight-line payoff engine that never makes the same sign error twice. Breakevens then come from np.interp on the sign change of the profit column rather than from algebra, which generalizes instantly to ratio spreads and butterflies where the algebra gets tedious. Greeks come from a vectorized BSM using scipy.stats.norm.cdf, so a full strike-by-expiry Greeks surface is a single broadcast. The minimum-variance hedge ratio is literally an OLS slope: h_hat = np.cov(dS, dF)[0,1] / np.var(dF), with R² as effectiveness, on 1,000 simulated daily returns with a true h* = 1.2343 and ρ² = 0.5184 the sample estimates come back 1.2559 and 0.4845, which is itself the lesson that your hedge ratio is an estimate with sampling error, not a constant. And §6's discrete-rebalancing error distribution is a fifteen-line Monte Carlo: loop over time steps, re-compute delta, trade the difference, accrue cash. The quant-code region builds all of this properly; this callout is the seam. Do it by hand first, then never do it by hand again.

14. Where this leaves you

You can now take any structure, write its payoff, solve its breakevens, and put a currency number on its maximum gain and loss; size an equity, rate or FX hedge and name every residual it leaves; delta-hedge a book, watch it fail second-order, fix it with a second option, and account for the theta surrendered; and say what the hedge costs per year, which is the number that decides whether it should exist.

Carry the discipline, not the catalogue. Three questions before any hedge: What exactly am I exposed to? (name the risk factor, not the instrument). What does the hedge leave behind? (rounding, beta error, basis, correlation, convexity, vega, theta, compute each). What does it cost per year, and is that less than owning less? Analysts who ask those three rarely put on a bad hedge; analysts who reach straight for the instrument routinely do.

Common mistakes & how experts think differently

1. Reading a payoff diagram instead of deriving it. The diagram is a picture of the algebra, not a substitute; the moment a structure has an unequal ratio or a leg you have not seen, remembered shapes fail silently. Experts write the table (one row per price region, one column per leg) which costs ninety seconds and is why they do not make sign errors.
2. Forgetting the multiplier, or applying it twice. The highest-frequency real-world option error, and not a beginner's error: per-share and per-contract numbers look identical on a page. Experts carry the unit at every step ("₹156.51 per share × 12,000 = ₹18.78 lakh") and treat any number without a unit as unverified.
3. Calling a covered call "income." Selling a call is selling the right tail; the premium is that tail's price, not a yield. The correct frame is the synthetic short put: if you would not sell the naked put, do not sell the covered call. Experts judge it on whether the implied volatility sold exceeds the volatility expected, a forecast, and they say so.
4. Buying a straddle because "there's an event." Everyone knows about the event, so it is in the price: the straddle's breakeven is the market's expected move. Experts compare their own forecast to the implied one first, knowing that a correctly-forecast direction plus a volatility crush is a losing straddle (WE3: +8.3% realized, −$249 per contract).
5. Believing a delta hedge makes you flat. It makes you flat to first order and converts the position into a bet on realized versus implied volatility. Experts state the gamma exposure explicitly ("short 350 shares of gamma per point") and know their break-even realized volatility before the open.
6. Gamma hedging without pricing the theta given up. Neutralizing gamma in WE5 cut theta from $1,084/day to $29/day. That is the price of the protection and it belongs in the decision; any "risk-free" gamma-hedge pitch without the theta line is incomplete.
7. Hedging at a ratio of 1 because it feels natural. When the instrument is not the exposure, h* = ρσ_S/σ_F, routinely well above or below 1. More importantly, experts report ρ² rather than h*, because ρ² is what tells a risk committee how much risk actually went away, and a 0.72 correlation, which sounds high, removes only 51.8% of variance.
8. Ignoring the residual because the hedge "worked." A hedge offsetting 98% of a move looks perfect and leaves 2% of a large number. Experts decompose every outcome into named components and are suspicious of any residual they cannot name, because an unnamed residual is an unmodelled exposure that will be larger next time.
9. Assuming flat volatility when pricing a collar. Skew moved WE1's zero-cost call strike from ₹2,023 to ₹1,939 and roughly halved the retained upside. Experts quote collars from the market's own risk reversal, not from one volatility in a spreadsheet.
10. Treating the FX forward premium as a forecast. WE8's 2.17% annualized premium is the interest differential and nothing else. "The forward says the rupee will weaken" is a category error, and a treasurer who declines to hedge because "the forward is expensive" is declining because their own currency has higher interest rates.
11. Hedging permanently because volatility is uncomfortable. A standing 5%-out-of-the-money put program costs about 11.3% of NAV a year at 28% volatility. Experts know that before they start, which is why they hedge episodically, or collar, or (this program's preference) own less (M9.01).
12. Letting an AI produce the hedge ratio. These formulas are exactly what a language model reproduces fluently and gets subtly wrong: a sign, a multiplier, a conversion factor in the wrong place. Experts use AI to build the grid and never to own the number, and check the degenerate case on every table before trusting the general one.

The thread through all twelve: an expert's model of a hedge is not "the risk is gone" but "the risk has been transformed into a specific, smaller, named residual with a known price." Every mistake above is a failure to name the residual.

Worked examples

Every figure below was recomputed in Python before it entered this page (R3). Companies are explicitly synthetic; market levels, volatilities and rates are stated assumptions chosen for arithmetic clarity, not quotations. Contract specifications are illustrative and must be verified on the exchange's own page before use.

Worked example 1: A collar on a concentrated Indian holding (India, ₹)

Konkan Speciality Chemicals (synthetic). You hold 12,000 shares at ₹1,850 (₹2,22,00,000) acquired long ago at a low basis, so selling triggers a large gain. You want to survive the next three months without an open-ended downside and will give up some upside. Assume T = 0.25, risk-free 6.50%, dividend yield 1.00%, a flat 28% implied volatility, and an illustrative lot of 500 shares (so 12,000 shares is exactly 24 contracts).

Solution. Step 1: price the legs (BSM, DV1.02). The ₹1,750 put is ₹49.67; the ₹2,000 call is ₹56.18. Net premium D = p − c = −₹6.51: a credit of ₹6.51 per share, ₹78,126 across the position.

Step 2, why the credit exists, which is the most instructive number here. The strikes look lopsided against spot (₹100 below, ₹150 above), but options are symmetric around the forward, not the spot, and the three-month forward is 1,850 × e^{(0.065−0.010)×0.25} = ₹1,875.61. From there the put is 125.6 below and the call 124.4 above, nearly symmetric, with the call marginally closer and therefore marginally dearer. A collar that looks like it gives away more upside than downside is often, correctly measured, symmetric. Whenever a collar's premiums surprise you, re-measure the strikes from the forward.

Step 3, the payoff table. Profit per share = (S_T − 1,850) + max(1,750 − S_T, 0) − max(S_T − 2,000, 0) + 6.51:

Step 4, the three numbers. BE = S₀ + D = ₹1,843.49. Max profit = K_c − S₀ − D = ₹156.51/share = ₹18,78,126 (+8.46%). Max loss = K_p − S₀ − D = −₹93.49/share = −₹11,21,874 (−5.05%). For comparison at the same strikes: the covered call alone breaks even at ₹1,793.82 with max gain ₹206.18 but unbounded downside; the protective put alone breaks even at ₹1,899.67 with a floor of −₹149.67 and unbounded upside. The collar is the only one of the three with both ends closed.

Step 5, the parity cross-check (do this every time). Set both strikes to ₹1,900: put ₹116.0905, call ₹92.0968, net debit ₹23.9937. Parity predicts Ke^{−rT} − S₀e^{−qT} = 1,869.3745 − 1,845.3808 = ₹23.9937: exact to four decimals. Total outlay 1,850 + 23.9937 = ₹1,873.9937 decomposes as Ke^{−rT} = ₹1,869.3745 plus S₀(1 − e^{−qT}) = ₹4.6192, the present value of the dividends you collect. Net of dividends you paid the PV of ₹1,900 for a certain ₹1,900: the equal-strike collar is a risk-free bond. If your spreadsheet cannot reproduce this, fix it before trusting anything else it says.

Step 6, the zero-cost version, and the skew that ruins it. At a flat 28%, the call strike that exactly funds the ₹1,750 put is ₹2,023.27, a band of −5.4%/+9.4%. Price the put at 33% and the call at 25% (a modest, realistic skew): the put costs ₹65.65 and funds only a ₹1,938.91 call: −5.4%/+4.8%. Skew cut the retained upside nearly in half. The lesson is not that flat-volatility is a poor approximation; it is that the upside strike of a zero-cost collar is set by the skew, so it must be quoted from the market's risk reversal.

Read it. The collar did what a concentrated holder needs: a −5.05%/+8.46% band for a small credit, with no sale and no tax event. What it did not do is remove risk, it exchanged unlimited downside for a capped band plus the risk that the stock triples and you own none of it above ₹2,000, plus margin mechanics, plus the tax treatment of the option legs (M1.09 and a tax adviser, not this module). Name the exchange, then decide.

Worked example 2: Vertical spreads and the parity tie-out (US, $)

SPX, assumed index level 5,600, three months, r = 4.30%, q = 1.30%, flat 15% implied volatility, $100 multiplier. You are mildly bullish and want defined risk.

Solution. Step 1: price four options. c(5600) = 188.03, c(5750) = 120.56, p(5600) = 146.33, p(5750) = 227.25.

Step 2: build both spreads. Bull call spread: long 5600 call, short 5750 call → debit 67.47 → $6,747 per contract at risk. Bull put spread: short 5750 put, long 5600 put → credit 80.93 → $8,093 received.

Step 3, the parity tie-out. 67.47 + 80.93 = 148.3961, and (K₂ − K₁)e^{−rT} = 150 × e^{−0.043×0.25} = 148.3961: exact. Same risk; the 150 − 148.3961 = 1.6039 difference is precisely the interest on the strike width, and it is why the breakevens differ.

Step 4, the numbers.

Step 5, read what the spread bought. The outright call needs the index up 3.36% to break even; the spread needs 1.21%. Risk-neutral probability of finishing above the spread's breakeven is 46.1% versus 35.3% for the outright, far more likely to make something. What you gave up is everything above 5,750. Note the exact identity: the breakeven improved by 120.56 points, precisely the premium of the call you sold. Every point of breakeven improvement was purchased with a point of short-leg premium: arithmetic, not judgement.

Read it. Take the credit version when you want cash now and can post margin, the debit version when you want the maximum loss to be cash already paid. They are the same position at a 1.60-point financing difference ($16,039 on 100 contracts) invisible unless you tie them out through parity. WE1's skew caveat applies here too: at a real SPX skew the 5,600 put prices above 15% volatility and the credit is larger.

Worked example 3: Buying event volatility, and losing while being right (US, $)

Cascadia Diagnostics (synthetic) trades at $84 with a binary regulatory decision in 21 days. The 85-strike options are bid to 62% implied volatility; r = 4.30%, no dividend. You buy the 85 straddle. The decision lands, the stock jumps to $91, and implied volatility collapses to 34% with 19 days left.

Solution. Step 1, price it and state the breakevens before trading. Call $4.62 + put $5.41 = $10.02, 11.93% of spot. Breakevens $95.02 (+13.12%) and $74.98 (−10.74%). Greeks: net delta +0.0091 (flat), gamma 0.0639, vega 16.08 per 1.00 of volatility (divide by 100 for one volatility point, so 0.161 per point), theta −$0.236/day.

Step 2: sanity-check against theory. Expected absolute move = S·σ·√(2T/π) = 84 × 0.62 × √(2×0.0575/π) = $9.97, 11.87% of spot, against a straddle price of $10.02. The straddle is priced at the expected absolute move, to within five cents. That is what "fairly priced" means, and it says immediately that this trade has no edge unless your volatility forecast differs from 62%.

Step 3: mark it after the event. At S = 91, σ = 34%, T = 19/365: call $6.86, put $0.67, straddle $7.53. P&L = −$2.49 per share = −$249 per contract. The stock moved +8.33%, large, and correctly anticipated, and the position lost a quarter of its cost.

Step 4, the two breakevens that actually mattered. The expiry breakeven was +13.12%; you got +8.33%. But you did not hold to expiry, so: at what post-event volatility would this have been flat? 54.12%. Implied volatility had to stay above 54%, and events by definition destroy the uncertainty supporting the 62%. Alternatively, at 34% volatility the stock needed $94.29 (+12.25%). Two conditions, either of which saves it; neither met.

Step 5, cheaper expressions of the same view. The 80/90 strangle costs $5.81 with breakevens $74.19 / $95.81: 42% cheaper, needing a bigger move. The 80/85/90 call butterfly costs $0.77 for a maximum profit of $4.23 at $85, a 5.5 : 1 payoff, the opposite view, and what to own if you thought 62% was too high.

Read it. You were right about the event, right about direction, roughly right about magnitude, and lost money, because the price already contained the move and the volatility crush took the rest. Buying an option is never a bet on an event; it is a bet on the event being bigger than the price says. Computing the breakeven percentage before trading, and asking honestly whether you can forecast a move larger than that, avoids this entirely.

Worked example 4: Delta-hedging a short call book through a price move (US, $)

Fairbanks Instruments (synthetic) trades at $150. Your desk is short 200 contracts of the three-month 150 call: 20,000 share-equivalents. r = 4.50%, q = 1.50%, σ = 30%, T = 0.25, multiplier 100.

Solution. Step 1, the option and the position. Price $9.4703, delta 0.5477, gamma 0.017527, vega 29.577 per 1.00 of volatility, theta −$19.784/yr = −$0.0542/day. Premium received $189,405. Position Greeks (short, so signs flip):

Step 2: hedge the delta. Buy 10,954 shares for $1,643,042. The book is delta-neutral, and the remaining line items read honestly: short 350.54 shares of gamma per point, short $591,534 of vega per 1.00 of volatility (−$5,915 per point), long $1,084 a day of theta.

Step 3, one trading day at unchanged volatility (option revalued at T = 0.25 − 1/252; "approx" is ½Γ_pos(ΔS)² + Θ_pos·Δt):

Step 4, read the table. The "hedged" book loses on both sides: $39,207 down and $34,585 up, and the loss scales as the square of the move (±5% → ±10% multiplies the gamma term by four). At zero move it earns +$1,576, essentially the day's theta. The second-order estimate reproduces actual P&L within ~8% at ±10% and ~1% at ±2%: good enough to think with, not to book.

Step 5: put the identity to work. Break-even daily move: ½ × 350.54 × (ΔS)² = 1,084 → ΔS = $2.49, 1.66% of spot, against the 30%/√252 = 1.89% the implied volatility charges for. In volatility terms, ½Γ_pos S²(σ_impl² − σ_real²)Δt gives +$782/day at 20% realized, $0 at 30%, −$1,095 at 40%, −$1,731 at 44.79%. The delta-hedged short book is a short position in realized volatility, and 30% is its strike.

Step 6, five days of actual rebalancing along 150 → 153 → 149.50 → 155 → 152 → 158:

The trade column buys high and sells low every single day, that is what short gamma is, mechanically. The path realizes 47.45% annualized against 30% implied, and the book loses $8,595 over five days, in line with Step 5's identity.

Step 7, the irreducible noise. Monte Carlo over 20,000 paths, short one at-the-money call (premium $9.4703):

Each quadrupling of frequency roughly halves the error (1/√n), and even daily rebalancing leaves noise worth 5.3% of the premium. Against transaction costs rising linearly in frequency, this is why desks rebalance on delta bands.

Read it. A delta hedge is not a hedge in the everyday sense. It removes the directional risk you did not want and installs a volatility risk you must then have a view on. If you cannot state your book's break-even realized volatility, you do not know what you own.

Worked example 5: Delta-gamma hedging the same book, and what it costs (US, $)

Same book: position delta −10,953.6, gamma −350.54, vega −$591,534 per 1.00 of volatility, theta +$1,084/day. The hedging instrument is the three-month 160 call: price $5.4662, delta 0.3787, gamma 0.016860, vega 28.452, theta −$18.529/yr.

Solution. Step 1: gamma first. n₂ = 350.54 / 0.016860 = 20,790.8 share-equivalents = 207.91 contracts to buy, costing $113,647. Check: −350.54 + 20,790.8 × 0.016860 = 0.000000.

Step 2: delta second. The purchase brought +7,872.7 of delta, so net delta is −3,080.9; buy 3,080.9 shares ($462,137), down from 10,954, because the long calls now carry most of the positive delta. Stock has zero gamma, so this cannot disturb Step 1. Order matters: gamma, then delta.

Step 3, the same one-day moves.

The residual falls by roughly an order of magnitude at ±10% and changes character, no longer uniformly negative. What remains is third-order (the rate of change of gamma) plus the two options' higher-order mismatch.

Step 4, now price what you paid. Residual vega is $0, on either scaling. Position vega was −$591,534 per 1.00 of volatility; 20,790.8 × 28.452 = +$591,534. They cancel, because for two options on the same underlying and expiry, vega/gamma = S²T is identical at every strike. Gamma-neutralizing with a same-expiry option is automatically vega-neutralizing, a theorem rather than a coincidence, and it also means you cannot separate the two risks without going to a different expiry. Residual theta: $29/day, down from $1,084/day, you surrendered 97% of the theta income. Verify with a vol shock: at 30% → 35% with the stock unchanged, the delta-only book loses $27,761 while the delta-gamma book makes +$305.

Read it. Delta-gamma hedging is not "a better hedge"; it is a different position, a short-volatility book earning $1,084/day traded for a nearly flat book earning $29/day. If your view was that realized volatility would come in under 30%, you have just thrown away the trade. The question is never "should I be gamma-neutral" but "do I want to be short volatility at 30%, and how much", the hedge is the answer to that question, not a substitute for it.

Worked example 6: Hedging an Indian equity portfolio with Nifty futures (India, ₹)

Your book is ₹18,50,00,000 (₹18.5 crore) with beta 1.22 to the Nifty 50. The index is 24,800, the near-month future 24,900 (basis +100), assumed lot 75. You want to neutralize market risk for a quarter without selling anything.

Solution. Step 1: size it. Contract value = 24,900 × 75 = ₹18,67,500. N = −1.22 × 18,50,00,000 / 18,67,500 = −120.86 → short 121 contracts. Rounding leaves 0.1432 contracts = ₹2,67,500 unhedged (0.14% of the book). Name it now; it reappears later.

Step 2, the generalization. To target beta 0.50 instead of zero: N = (0.50 − 1.22) × 18,50,00,000 / 18,67,500 = −71.33 → short 71 contracts. Same formula, one parameter.

Step 3, the quarter happens. The index falls 12.00% to 21,824; the future falls to 21,850 (−12.25%), so basis goes +100 → +26, a −74 point change. Your portfolio falls 16.10%.

Step 4, the P&L. Portfolio 18,50,00,000 × (−16.10%) = −₹2,97,85,000. Futures (−121) × 75 × (21,850 − 24,900) = +₹2,76,78,750. Net −₹21,06,250, or −1.14% of the book, against an unhedged −16.10%.

Step 5, decompose the residual, every rupee named.

Read the signs. The beta error dominates (4× the basis effect and 82× the rounding) and it is stock-specific risk, which an index hedge is not supposed to remove. The basis moved in your favour here (the future cheapened while you were short it); over a quarter ±74 points is ordinary and can go either way. Rounding is trivial on this book and would not be on a small one: a ₹2 crore portfolio needs 13.07 contracts, and rounding to 13 leaves 0.5% mis-hedged.

Step 6, the cost, stated honestly. Run the same hedge through a +12% quarter. A pure-beta portfolio gains 1.22 × 12% × ₹18.5cr = +₹2,70,84,000; the short futures lose −₹2,62,81,200 at an assumed converging basis; net +₹8,02,800, or +0.43% against +14.64% unhedged. That is what the hedge costs when you are wrong about direction: 14.2 percentage points of a good quarter. A hedge is a position, and it can be the wrong one.

Read it. The formula is the easy part. The professional content is Step 5 (every component named, signed and sized) so that when a hedge "doesn't work" you know within five minutes whether it was beta, basis, rounding, or a data error, and only one of those is fixable.

{
  "skill": "Sizing a futures hedge (beta-adjusted equity, DV01 rate, or FX) and decomposing the realized residual",
  "use": "Use an assistant to lay out the hedge-sizing arithmetic step by step and to build the post-trade attribution table (index / beta error / basis / rounding), so you have a structured second pass over a calculation whose sign and denominator errors are easy to make and hard to see.",
  "tools": ["Chat assistants with code execution — an assistant that writes and RUNS the arithmetic in Python rather than predicting it", "Spreadsheet AI — Excel Copilot over your own sizing grid", "Exchange contract-specification pages (NSE, CME) — the only acceptable source for lot size, multiplier and conversion factor"],
  "prompt": "Portfolio: <value, beta or modified duration>. Hedge instrument: <futures price, multiplier/lot, conversion factor if any>. Show the hedge-ratio formula, substitute the numbers, and give the fractional and rounded contract count. Then build an attribution table for this realized outcome: <index move, portfolio move, futures move>. Show units at every step and flag any assumption you had to make.",
  "verify": "Recompute the contract count by hand (one division) and confirm the units cancel to 'contracts'. Independently verify the multiplier/lot size and conversion factor on the EXCHANGE's own contract-specification page — these are perishable, exchange-set values that models reproduce from stale training data with complete confidence. Check that the attribution components sum to the actual residual; an unexplained gap means an unmodelled exposure, not a rounding issue. Cite the spec page and the date you checked it (Guardrail 1-5, especially #1 Trace, #4 Date-check).",
  "diy": "You must be able to derive N = -beta*V/(F*m) and N = -BPV_p/BPV_f from the definitions, and name the four residual components unprompted. The formulas are the gate; the assistant only saves keystrokes on the table.",
  "market": "IN"
}

Worked example 7: Hedging a US bond portfolio: DV01, conversion factor, yield beta (US, $)

You run a $250,000,000 investment-grade corporate portfolio, modified duration 6.8, convexity 62. You expect rates to rise and want to neutralize duration for a quarter without selling into a thin market. The hedge is a Treasury note future whose cheapest-to-deliver bond trades at 96.25 with modified duration 7.65 on $100,000 face and a conversion factor of 0.8215.

Solution. Step 1, the two BPVs. Portfolio = 6.8 × 250,000,000 × 0.0001 = $170,000/bp. CTD = 7.65 × 96,250 × 0.0001 = $73.6313. Futures = 73.6313 / 0.8215 = $89.6302/bp.

Step 2: size it. N = −170,000 / 89.6302 = −1,896.68 → short 1,897 contracts.

Step 3, the naive check at +75bp. Portfolio (linear) −170,000 × 75 = −$12,750,000; futures +1,897 × 89.6302 × 75 = +$12,752,144. Net +$2,144 on a $12.75 million move, textbook-perfect, and misleading, because two things were assumed away.

Step 4: put convexity back. +½ × 62 × 0.0075² × $250mm = +$435,937, so the portfolio actually loses $12,314,062 and the hedge shows a residual gain of ~$438,000. Positive convexity is a gift on a large move either way, and exactly what a callable bond or MBS lacks (FI1.04).

Step 5, put yield beta back, and watch the hedge break. At yield beta 1.15, a 75bp Treasury move drags corporate yields 86.25bp, so the portfolio loses −170,000 × 86.25 + ½ × 62 × 0.008625² × 250mm = −14,662,500 + 576,527 = −$14,085,973 while the hedge still gains only $12,752,144. Residual: −$1,333,829: 0.53% of the portfolio, from a parameter absent from the textbook formula. That is quantified basis risk, and it is the most important number here.

Step 6: fix the sizing. N = −1.15 × 170,000 / 89.6302 = −2,181.18 → short 2,181 contracts, gaining $14,661,268 against a $14,085,973 loss, net +$575,295, which is the convexity gain of $576,527 less $1,232 of rounding. Every dollar of residual accounted for. That tie-out, not the hedge, is the deliverable.

Step 7, the swap alternative. A 7-year semiannual pay-fixed swap at a flat 4.00% par rate has annuity 6.053124 years, so PV01 = $605.31 per $1mm per bp; neutralizing $170,000 needs 170,000 / 605.31 × $1mm = $280,846,699, rounded to $281mm (PV01 $170,093). No conversion factor, no CTD switch, no quarterly roll, but documentation, initial and variation margin, and counterparty exposure. The arithmetic favours the swap; operational reality often favours the futures. Both are correct answers to different questions.

Read it. Three residuals: rounding $1,232, convexity +$576,527, yield beta −$1,333,829 if ignored, $0 if included. Note the ranking, the parameter not in the formula beats rounding by three orders of magnitude. That is the general lesson of hedge sizing: the errors that matter are almost never in the arithmetic, they are in the assumptions the arithmetic quietly makes.

Worked example 8: An Indian exporter's dollar receivable, four ways (India, ₹ / US, $)

Vindhya Software Services (synthetic) will receive $6,000,000 in 91 days. Assume spot USD/INR 86.40, INR 6.60% (Act/365), USD 4.35% (Act/360), implied volatility 5.5%. Every level here is an assumption; live quotes come from the bank or the exchange.

Solution. Step 1, the forward (CIP, DV1.01). F = 86.40 × (1 + 0.0660 × 91/365)/(1 + 0.0435 × 91/360) = 86.8665, a premium of 0.4665, 2.17% annualized, existing only because rupee rates exceed dollar rates. Selling forward locks ₹52,11,99,147.

Step 2, the option hedge, priced properly. An 86.50 USD put costs 0.7595/USD (₹45,57,050), delta −0.4266. Carried forward: 0.7595 × (1 + 0.0660 × 91/365) = 0.7720. Effective floor = 85.7280, which is 1.1385 below the forward, and the option beats the forward only above F + FV(premium) = 87.6385. State that before you buy: it converts "should we buy protection?" into "do we think the rupee will be weaker than 87.64?", a question with an answer.

Step 3, the zero-cost risk reversal. Buy the 86.00 put (0.5622) and sell the call struck to fund it exactly: 87.7848. A band of 86.0000 – 87.7848 at zero premium: floor ₹51.60 crore, cap ₹52.67 crore. Note the cap sits almost exactly at Step 2's crossover of 87.6385; the structures are pricing the same thing.

Step 4, the participating forward. Buy a put at K, sell a call at the same K on half the notional, with p(K) = ½·c(K): K = 86.2235 (put 0.6455 = ½ × 1.2909). Floored on the full $6mm, keeping half of anything above, a collar whose ceiling is a 50% slope.

Step 5, the comparison (₹ crore realized on $6mm).

Step 6, read it as a decision, not a menu. The forward wins outright at or below 86.87 and never above it. The put costs ₹68 lakh versus the forward in the down cases and overtakes above 87.64. The risk reversal is the best down-case protection after the forward at zero premium, capping at 87.78. The participating forward is never best and never worst. There is no dominant structure, only one that matches your view and your tolerance for regret. The first question is not "which is optimal" but "which regret can this company survive," and for a company whose margins depend on the receivable the answer is usually the forward, because certainty is the product being purchased.

Step 7, the venue basis the textbook omits. The same exposure can go through NSE USD/INR futures ($1,000 per contract, 6,000 contracts for $6mm), a bank forward, or the offshore NDF. Assume the NDF at 86.95 against an onshore forward of 86.83: a 12-paise basis worth ₹7,20,000 on $6mm. Against the CIP-fair 86.8665, the onshore forward is 3.65 paise cheap. The venue is a real decision worth more than most of the strike selection above.

Read it. Four structures, one exposure, and three numbers separate them: the CIP forward (86.8665), the premium carried to expiry (0.7720), and the crossover (87.6385). Everything else is presentation. The accounting treatment (cash-flow hedge designation, effectiveness testing, OCI versus P&L) is M1.09, which owns it.

Worked example 9: Cross-hedging: the minimum-variance ratio and what is left (India ₹ / commodity)

A ₹18.5 crore mid-cap Indian portfolio has annual volatility 24%; the Nifty future has 14%; correlation 0.72. How many contracts, and what is left?

Solution. Step 1, the ratio. h* = 0.72 × 0.24 / 0.14 = 1.2343. You must short 1.2343 rupees of index exposure per rupee of portfolio: 23% more than notional, because the portfolio is far more volatile than the index. A naive 1:1 hedge under-hedges.

Step 2, how much risk actually goes away. Effectiveness = ρ² = 51.84% of variance; residual volatility = 24% × √0.4816 = 16.66%; volatility reduction = 30.6%. In rupees, one-standard-deviation risk falls from ₹4,44,00,000 to ₹3,08,12,448, a hedge a factsheet would call "market neutral" left ₹3.08 crore of annual risk standing.

Step 3, how much the ratio itself matters.

The objective is flat near the optimum: a 20% error in h costs 0.3–0.4 volatility points. The correlation, not the ratio, is the binding constraint: ρ sets the 16.66% floor and no h beats it. Effort therefore belongs in estimating and stress-testing ρ (remembering it moves toward 1 among risky assets exactly when you need the hedge), not in fine-tuning h.

Step 4, the same mathematics on a commodity cross-hedge. An airline (illustrative) must buy 900,000 barrels of jet fuel and can hedge only with crude. Jet fuel volatility 34%, crude 26%, correlation 0.88. h* = 0.88 × 0.34 / 0.26 = 1.1508 → 1,035,692 barrels → at 1,000 barrels per contract, 1,036 contracts. Effectiveness = 77.44%; residual volatility = 16.15%, against 16.62% for a naive 1:1. Even at a correlation of 0.88, very high by any standard, more than a fifth of the variance survives, and it survives as crack-spread risk: a genuine, named, historically volatile exposure.

Step 5, where the ratio comes from in practice. Regress the exposure's returns on the hedge's: the slope is h*, the R² is effectiveness. On 1,000 simulated daily returns with true h* = 1.2343 and ρ² = 0.5184, the sample estimates come back 1.2559 and 0.4845. Your hedge ratio is an estimate with sampling error, not a constant, one more reason Step 3's flatness is a comfort and the correlation is the worry.

Read it. "Hedged" is not a binary state. It is a number: ρ² of variance removed, σ_S√(1−ρ²) left standing, and a specific named residual (stock selection here, the crack spread there). Report both or you have not reported the hedge.

Worked example 10: A calendar spread valued across the grid, and a diagonal beside it (US, $)

A stock trades at $100 with r = 4% and no dividend. The one-month (30-day) 100 call is offered at 24% implied volatility and the three-month (90-day) 100 call at 21%. Build the long calendar, value it at the front expiry across a price grid, then compare it with the long diagonal that sells the 30-day 105 call against the same back-month leg.

Solution. Step 1, price the legs and state the entry. Short 30-day 100 call = $2.9072; long 90-day 100 call = $4.6470. Net debit $1.7399, which is also the maximum loss, since both legs are calls at the same strike and the long one outlives the short one. Net Greeks at entry: delta +0.0256, gamma −0.0199, vega +8.1992 per 1.00 of volatility (0.082 per point), theta +$0.0226 per day.

Step 2, value it at the front expiry. At the front expiry the short leg is worth its intrinsic value and the long leg is a 60-day option. Holding back-month implied volatility at 21%:

The shape is a tent centred on the strike, which is the signature of a short-gamma position, and the peak of 1.9825 is a 114% return on the debit. The two breakevens sit a little outside 95 and a little inside 105, so the trade needs the underlying to end the month within roughly five percent of where it started.

Step 3, price the volatility risk separately, because it is a different risk. Repeat the middle three rows with back-month implied volatility collapsing from 21% to 17%:

A four-point fall in back-month implied volatility costs 0.6420 at the pin, which is 32% of the profit, and turns the 95.00 outcome from a small loss into a large one. That is the vega leg of the trade doing exactly what the Greeks table said it would, and it is the reason this structure cannot be sized off the price grid alone.

Step 4, the diagonal for comparison. Keep the long 90-day 100 call and sell the 30-day 105 call at 23.5% implied instead, for $1.0261. Net debit $3.6209, delta +0.2978, gamma −0.0104, vega +10.2888 per 1.00 (0.103 per point), theta +$0.0107 per day. At the front expiry:

The tent has become a ramp. The diagonal keeps the calendar's long-vega, long-theta character but adds a directional view worth 0.30 of delta, and it pays for that view with a larger debit and a worse downside. Choose between them on whether you have a directional opinion, not on which payoff table looks nicer.

Read it. Neither table is a payoff diagram and neither is a promise. Every figure in the front-expiry columns is a model value of a live option, so it inherits the model's assumptions and, in particular, an implied-volatility input that has to be forecast rather than observed. The tent in Step 2 will be drawn confidently in any options-education deck; the honest version is Step 3's, which shows that a third of the profit at the pin depends on a volatility level nobody has quoted you. A calendar spread sized without the Step 3 grid is a volatility position taken by accident.

Practice set

P1–P5 are guided, P6–P10 independent, P11–P15 exam-condition (12 minutes each, closed book). Numeric answers to ±1% unless stated. All entities are synthetic or illustrative; contract specifications are assumptions. Write your full answer before reading the solution.

P1 (guided: covered call). You own a share at ₹2,400 and sell a three-month ₹2,600 call for ₹62. Give (a) the breakeven, (b) the maximum profit in rupees and percent, (c) the static return if the share is unchanged, annualized, (d) the P&L at ₹2,200 and ₹2,800.

Solution. (a) 2,400 − 62 = ₹2,338. (b) 2,600 − 2,400 + 62 = ₹262, or 10.92%. (c) The call expires worthless and you keep ₹62: 62/2,400 = 2.58% for the quarter; (1.0258)⁴ − 1 = 10.74% annualized, and note this "yield" is earned only in the state where the share does not move. (d) At ₹2,200: −₹138. At ₹2,800: capped at +₹262, you forfeited ₹138 of the ₹200 above the strike.

P2 (guided: protective put). You own a share at $148 and buy a $140 put for $4.85. Give the breakeven, the maximum loss in dollars and percent, and the insurance cost as a percentage of spot.

Solution. BE = 148 + 4.85 = $152.85. Max loss = 140 − 148 − 4.85 = −$12.85, or −8.68%. Cost = 4.85/148 = 3.28% of spot for the period. Read the last number: an 8.68% floor bought for 3.28% means you paid over a third of the protected loss for the protection, which is what short-dated insurance costs, and why permanent put programs are expensive (§12).

P3 (guided: zero-cost collar). A share trades at ₹640. You buy a ₹580 put and sell a ₹720 call, each at ₹21.50. Give the breakeven, maximum gain and maximum loss (rupees and percent), and state the one thing this did to your position.

Solution. D = 0, so BE = ₹640 (spot). Max gain 720 − 640 = ₹80 (+12.50%); max loss 580 − 640 = −₹60 (−9.38%). What it did: converted an open-ended position into a −9.38%/+12.50% band at zero cash cost, keeping the dividends and votes and creating no taxable disposal. What it did not do: reduce exposure inside the band, or protect you if the share triples.

P4 (guided: bull call spread). Strikes 100/110; premiums 6.20 and 2.45. Give the debit, breakeven, maximum profit, maximum loss, risk/reward, and the P&L at 104 and 115.

Solution. Debit 3.75; BE 103.75; max profit 10 − 3.75 = 6.25; max loss 3.75; risk/reward 1.667 : 1. At 104: 4 − 3.75 = +0.25. At 115: capped at +6.25.

P5 (guided: straddle and its breakeven percentage). A share is at ₹1,200; the ₹1,200 straddle costs c = 58, p = 51. Give the cost, both breakevens, and the percentage move required, then say what that number is a forecast of.

Solution. Cost ₹109; breakevens ₹1,091 / ₹1,309; required move 109/1,200 = 9.08% either way. That percentage is the market's own estimate of the expected absolute move over the option's life, buying the straddle bets the actual move exceeds 9.08%, not that the share moves.

P6 (independent: strangle versus straddle). Same share at ₹1,200. A 1,150/1,250 strangle costs p = 29, c = 33. Give the cost, the breakevens, the required percentage moves, and one sentence on when you would prefer it to P5's straddle.

Solution. Cost ₹62 (43% cheaper). Breakevens ₹1,088 and ₹1,312: moves of −9.33% and +9.33%. Prefer the strangle when you expect a large move and want more leverage per rupee; prefer the straddle when the move might be moderate, since the strangle pays nothing between 1,150 and 1,250. Note the relative price here: the strangle's breakevens are barely wider (9.33% vs 9.08%) for 43% less cost, exactly the comparison to run before buying either.

P7 (independent: butterfly). Calls at 480/500/520 cost 34.50 / 20.10 / 10.40. Build the long butterfly: debit, maximum profit, breakevens, payoff ratio.

Solution. Debit 34.50 − 2(20.10) + 10.40 = 4.70. Max profit 20 − 4.70 = 15.30 at exactly 500. Breakevens 484.70 and 515.30. Ratio 3.26 : 1. It pays if the underlying finishes within ±3.06% of 500, a short-volatility view with a hard maximum loss of 4.70, which is why a butterfly is the acceptable way to be short volatility under M9.01's anti-ruin overlay while a naked straddle is not.

P8 (independent: delta hedge and gamma re-hedge). You are short 150 contracts (100 shares each) of a call with delta 0.62 and gamma 0.021. Give (a) the position delta and initial hedge, (b) the new delta after a +$4 move and the re-hedge trade, (c) the approximate gamma P&L over that move.

Solution. (a) Position delta = −150 × 100 × 0.62 = −9,300 shares; buy 9,300 shares. (b) Position gamma = −315 per $1. After +$4 the option delta is 0.62 + 0.021 × 4 = 0.704, so position delta is −10,560; buy 1,260 more shares, buying after the price rose, the mechanical signature of short gamma. (c) ½ × (−315) × 16 = −$2,520, before theta.

P9 (independent: delta-gamma solve). A book is short 50 contracts of option 1 (Δ₁ = 0.55, Γ₁ = 0.030), multiplier 100. Hedging option 2 has Δ₂ = 0.32, Γ₂ = 0.048. Solve the system.

Solution. Position gamma −150; position delta −2,750. Gamma first: n₂ = 150/0.048 = 3,125 share-equivalents = 31.25 contracts, long. Delta second: −2,750 + 3,125 × 0.32 = −1,750, so buy 1,750 shares. Delta-first would require redoing the delta step, because option 2 carries delta; stock carries no gamma, so gamma-then-delta converges in one pass.

P10 (independent: beta hedge and target beta). A ₹42,00,00,000 portfolio has beta 0.88; the index future is 51,500 with lot 15. (a) Contracts to fully hedge. (b) Contracts to move beta to 1.30, and in which direction.

Solution. Contract value = ₹7,72,500. (a) N = −0.88 × 42,00,00,000 / 7,72,500 = −478.45 → short 478 (leaving 0.45 contracts ≈ ₹3.48 lakh unhedged). (b) N = (1.30 − 0.88) × 42,00,00,000 / 7,72,500 = +228.35 → long 228, levering market exposure without buying a share, from the same one-line formula.

P11 (exam: DV01 hedge). A $120,000,000 portfolio has modified duration 5.4; the futures BPV is $63.80. (a) Size the hedge. (b) Net P&L on a +40bp parallel shift, ignoring convexity and yield beta, then state what you ignored.

Solution. (a) BPV_P = 5.4 × 120,000,000 × 0.0001 = $64,800; N = −64,800/63.80 = −1,015.67 → short 1,016 contracts. (b) Portfolio −$2,592,000; futures +$2,592,832; net +$832, the rounding residual. Ignored: positive convexity (a gain on any large move) and yield beta, potentially far larger than $832, and the first thing to fix (WE7).

P12 (exam: minimum-variance hedge ratio). σ_S = 32%, σ_F = 26%, ρ = 0.65. Give h*, hedge effectiveness, residual volatility, and the residual volatility of a naive 1:1 hedge.

Solution. h* = 0.65 × 0.32/0.26 = 0.8000. Effectiveness = ρ² = 42.25% of variance. Residual volatility = 0.32 × √0.5775 = 24.32%. Naive h = 1: √(0.32² + 0.26² − 2 × 0.65 × 0.32 × 0.26) = 24.87%. Read it: a 0.65 correlation removes only 42% of variance, so 24.32 of the original 32 volatility points survive an optimal hedge, this exposure is essentially unhedgeable with this instrument, and the honest recommendation is to reduce the position rather than hedge it.

P13 (exam: FX crossover). An exporter faces spot 84.50 and a three-month forward of 85.30. An 85.00 USD put costs 0.92; the domestic rate is 6.60%. Give (a) the effective floor net of premium carry, (b) the spot above which the option beats the forward, (c) a one-sentence recommendation.

Solution. (a) 0.92 × (1 + 0.066 × 0.25) = 0.9352; floor = 85.00 − 0.9352 = 84.0648: 1.24 below the forward. (b) The option wins only above 85.30 + 0.9352 = 86.2352. (c) Unless you specifically expect spot above 86.24 (a +2.06% move in three months), the forward is the better hedge, and if the company's margin depends on the receivable, the forward is right regardless, because certainty is the product.

P14 (exam: interest-rate cap). A borrower on $25,000,000 buys a quarterly SOFR cap struck at 4.75%. Give the payoff and effective rate at realized SOFR of 4.00%, 4.75% and 6.00% (use 0.25 for the quarter).

Solution. Payoff = 25,000,000 × max(r − 4.75%, 0) × 0.25. At 4.00%: $0, effective 4.00%. At 4.75%: $0, effective 4.75%. At 6.00%: 25,000,000 × 0.0125 × 0.25 = $78,125, effective 4.75%. The cap is one-sided insurance: it truncates rate upside and leaves the benefit of falling rates intact, exactly what its premium buys, and exactly what a collar or swap gives away.

P15 (exam, the covered call's real cost, and hedge effectiveness). (a) A share at ₹3,200; you sell a ₹3,400 call for ₹86; it finishes at ₹3,750. Compute your P&L, the outright holder's, and the forgone amount, then describe what happened correctly. (b) A hedge gained $12,750,000 against a hedged-item loss of $14,662,500: give the dollar-offset ratio, say whether it is inside the commonly-cited 80–125% band, and name (do not explain) the module that owns what happens next.

Solution. (a) Covered call +₹286; outright +₹550; forgone ₹264 per share, 8.25% of the position. The correct description is not "the call was assigned" but "you sold the move from ₹3,400 to ₹3,750 for ₹86, three months in advance, and it turned out to be worth ₹350", whether that was a bad trade depends on whether ₹86 was fair for that risk at the time, not on the outcome (M9.01: process over outcome). (b) 12,750,000/14,662,500 = 86.96%, inside the 80–125% band though not comfortably. M1.09 owns designation, documentation, effectiveness testing and recognition; this module owns only the economics that produced the two numbers, and note that a single observation says almost nothing about ρ² over a series.

P16 (calendar spread: read the Greeks, then the risk). A stock trades at $50, r = 5%, no dividend. You sell the one-month 50 put at 30% implied volatility and buy the three-month 50 put at 26%. State the debit, the net vega and the net theta, and name the two things that must both go right.

Solution. Short one-month 50 put = $1.6214; long three-month 50 put = $2.2765; net debit $0.6551, which is the maximum loss. Net vega = +4.1107 per 1.00 of volatility, or +0.0411 per volatility point. Net theta = +$0.0142 per day. The structure is long vega and long theta together, paid for with short gamma, exactly as §12b describes. Two things must go right at once. The underlying has to finish the month near 50, because a large move in either direction takes the short leg to intrinsic value while the long leg's time value is only partly recovered. And back-month implied volatility has to hold, because the position's whole vega sits in the leg you bought. Getting the price right and the volatility wrong still loses money.

Applied mini-project

Build a hedge memo for one real, live exposure, and price its residual.

This produces a durable artifact: a two-page hedge memo an investment committee or treasurer could act on, feeding your decision journal (M0.04) and, for an equity exposure, the sizing memo from M9.01. You will pick one real exposure, size a hedge three ways, compute the residual for each, and recommend one, or recommend not hedging, which is legitimate and frequently correct.

Choose one exposure. (India, ₹) a concentrated equity position you or a family member holds, an Indian portfolio you track, or a real foreign-currency exposure. (US, $) a US-listed holding or index exposure, or a floating-rate liability whose reference rate you can source. If you have no live exposure, use a published one: an Indian listed company's annual-report note on foreign-currency exposure and derivative contracts (BSE/NSE), or a US issuer's Item 7A market-risk disclosure (SEC EDGAR). Both are primary sources containing a real, sized exposure.

Produce, in order:

1. Name the exposure precisely: amount, currency, direction, horizon, and the risk factor, not the instrument. "Long ₹2.2 crore of a mid-cap chemical" is an instrument; "long the Indian mid-cap factor, long the chemicals cycle, long one company's execution" is the exposure. Say which of those you want to keep.
2. Verify the hedge instrument's specification: multiplier or lot size, settlement style, exercise style, expiry, margin regime: five lines, each traced to the exchange's own contract-specification page, with the date you checked it. Not from memory, a broker's marketing page, or an AI.
3. Size three alternatives. Equity: beta-adjusted futures, protective put, collar. FX: forward, bought option, zero-cost risk reversal. Rates: futures/DV01, swap, cap. Show the arithmetic with units and give fractional and rounded contract counts.
4. Draw the payoff as a table (not a picture) spanning at least ±25%, one column per alternative plus unhedged, with breakeven, max gain and max loss for each. Run the parity cross-check on at least one structure.
5. Quantify each residual in currency, not adjectives, rounding; beta error or yield beta; basis (including venue basis); h*, ρ², σ_S√(1−ρ²) if it is a cross-hedge; convexity for rates; vega and theta if you used options. Each alternative must end with: "After this hedge I still carry X of Y risk."
6. Price the hedge per year: premium drag, forward points given up, or the sold call's opportunity cost, annualized as a percentage of the exposure, and compare it against owning less (M9.01). State which is cheaper.
7. Recommend, with a trigger: one alternative (or "do not hedge"), one sentence of reasoning, and one pre-committed trigger that would change the answer. Write it before the outcome is known and do not edit it (M0.04).
8. Name the seams: one line each pointing at M1.09 for hedge accounting and at a tax adviser for tax treatment. Attempt neither; demonstrate you know the boundaries of what you computed.

Deliverable: two pages. Page 1: exposure, three sizings, payoff table. Page 2: residual table, annual cost, recommendation and trigger. Every market figure sourced or explicitly labelled an assumption; every derived figure recomputed by you.

Scoring rubric (0/1/2 per criterion; 16 possible; ≥13 to pass, this is a crown module):

Self-check before submitting: does the memo contain a single number you could not re-derive in front of a skeptical reader in under two minutes? If yes, delete it or re-derive it. That test, applied honestly, is the whole gate.

Reading & resources

Everything below is to learn from and verify against. Nothing here, and no AI output, is a primary source for a contract specification or a market quote; those come from the exchange or the counterparty, dated.

The core texts:

• John C. Hull, Options, Futures, and Other Derivatives, the standard reference. For this module: the trading-strategies chapter (spreads, combinations, payoff tables), the Greeks chapter (delta hedging, the gamma-theta relationship, portfolio insurance), and the hedging-with-futures chapter (minimum-variance hedge ratio, cross-hedging, basis risk, tailing). If you own one derivatives book, own this one. [Paid] [Intermediate→Advanced]
• Sheldon Natenberg, Option Volatility and Pricing: the practitioner's book on the volatility dimension: why buying a straddle before an event usually loses, what skew is and why it exists, how a desk actually thinks about gamma and theta. Read after Hull's mechanics land. [Paid] [Intermediate]
• CFA Program Level I & II. Derivatives readings, and Level III. Risk Management Applications of Derivatives (equity/rate/FX hedging, target beta and target duration, currency hedging). This module's institutional vocabulary is theirs; the Level III applications readings are the closest published analogue to §§8–10. [Paid] [Intermediate→Advanced]

Free and authoritative, the specification sources:

• NSE India: F&O and currency-derivatives contract specifications and the circulars revising lot sizes (nseindia.com). The primary source for every Indian contract figure here. [Free] [Beginner→Advanced]
• CME Group, contract specifications for Treasury futures, SOFR futures and equity index futures/options, plus conversion-factor and cheapest-to-deliver documentation. The primary source for every US contract figure here. [Free] [Beginner→Advanced]
• Cboe: SPX/VIX product specifications and index-option settlement education. Settles the SPX-versus-SPY, European-versus-American, cash-versus-physical distinctions definitively. [Free] [Beginner→Intermediate]
• CCIL, FBIL and FIMMDA (India), benchmark rates, INR OIS/MIBOR fixings and market conventions; the reference points for any INR rate hedge. [Free] [Intermediate]
• RBI, master directions on foreign-exchange derivative contracts and hedging of foreign-currency exposure (rbi.org.in). What an Indian resident entity is actually permitted to do, which constrains §10 far more than pricing does. [Free] [Intermediate→Advanced]
• BIS Quarterly Review and the BIS Triennial Survey, size, structure and stress behaviour of FX, rate and equity derivatives markets, including the onshore/NDF relationship. [Free] [Advanced]

Going deeper on the mathematics:

• Emanuel Derman & Michael Miller, The Volatility Smile (2016), the honest, readable account of why §5's flat-volatility assumption is wrong and what practitioners do about it. Read it when the skew paragraph starts to bother you, which it should. [Paid] [Advanced]
• Espen Gaarder Haug, The Complete Guide to Option Pricing Formulas, the reference when you need a structure this module did not cover. [Paid] [Advanced]
• Riccardo Rebonato, Volatility and Correlation: the rigorous treatment of the correlation instability §11 flags as the binding constraint. [Paid] [Advanced]

Case reading: hedges that failed, and why:

• Metallgesellschaft (1993), the stack-and-roll rollover-basis failure. The canonical demonstration that a hedge sound at maturity can be fatal in the interim through margin calls and roll losses; the post-mortems (Culp & Miller; Mello & Parsons) argue both sides and are worth reading against each other. [Free: journal articles] [Advanced]
• Long-Term Capital Management (1998): correlation-goes-to-one in its purest form, and why §11 insists ρ is the binding constraint. Pair with M9.01's ruin arithmetic. [Free/Paid] [Intermediate→Advanced]
• Corporate hedging disclosures you can read today, any large Indian IT or pharma exporter's annual-report note on foreign-currency exposure and derivative contracts (BSE/NSE), and any US issuer's Item 7A in a 10-K (SEC EDGAR). Two real ones are worth more than a chapter. [Free] [Beginner→Intermediate]

Do this, not just read: take one company's derivative-instruments footnote from a real annual report and reconstruct from it what exposure they had, which instrument they used, how they must have sized it, and what residual they were left with. If the footnote does not let you answer all four, write down which one is missing, that gap is itself the finding, and noticing it is exactly what an analyst is paid for.

Teach it back & journal

Feynman prompt. Explain to a colleague who understands stocks but has never traded an option why "we hedged the position, so we're covered" is almost always incomplete. Use one concrete structure (the collar from Worked example 1 is ideal) and walk them through the three numbers that actually describe it: the breakeven, the floor and the cap, each derived in front of them from max(K − S_T, 0) rather than recalled. Then go one level deeper: show that even a perfect index hedge on a real portfolio leaves beta error, basis and rounding behind, and put a rupee number on each from Worked example 6. Finally ask them the question that tests whether it landed: "if a hedge removes 52% of the variance, what fraction of the volatility does it remove?" If they say 52%, they have not got it, the answer is 30.6%, because volatility is the square root of variance, and the gap between those two numbers is where most of the false comfort in the hedging industry lives. If they can, unprompted, ask "what's left over, and what does it cost per year?" about the next hedge anyone proposes to them, you have installed the module.

Journal prompt. Pick one exposure you actually carry (a concentrated holding, a foreign-currency obligation, a floating-rate loan) and, without trading anything, write the hedge memo in your journal before you know how the quarter turns out: the exposure named as a risk factor; three structures with their sizing arithmetic; the residual you would still carry under each, in currency; the annualized cost; the alternative of owning less; and your decision with a pre-committed trigger. Leave it alone for a quarter, then answer three questions in writing. First: which residual estimate was closest to what happened, and which furthest, and was the gap in the arithmetic or in an assumption the arithmetic quietly made? (In every worked example here, the largest residual came from a parameter that was not in the formula.) Second: did the annualized cost of the hedge turn out higher or lower than the cost of simply holding a smaller position, and would you have believed that number in advance? Third, and the one that matters most: find one moment in the quarter when you wanted to change the hedge, and write down what you were feeling and what the pre-committed trigger said. The professional value of this module is that it replaces a feeling about risk with a number about risk; the journal is where you find out whether the replacement took. Then write one paragraph on the question this Ring keeps returning to: given that every hedge priced here costs real money and leaves a named residual, when is the honest answer simply to own less of the thing, and what would have to be true about an exposure for hedging to beat that?

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.