Options Greeks

Why the Greeks exist

An option's price V depends on five things: the spot price S of the underlying, the strike K, the time to expiry T, the risk-free rate r, and the volatility σ. The strike is fixed at contract creation; the other four move continuously. A trader who has just sold one hundred calls cares less about the closed-form formula than about how that price will change if the market drifts a dollar, a day, a vol point, or a basis point. Each of those four sensitivities is a partial derivative — and the practice of organising option portfolios by their first- and second-order partials is what the trading floor calls "running the book in Greek-space."

Every Greek has both a definition and a hedging interpretation. Delta is a slope, but also the number of shares per option needed to neutralise a small spot move. Gamma is a curvature, but also the speed at which the hedge ratio itself decays into staleness. Theta is the time derivative, but also the daily premium a seller of options is "earning" for taking on the seller's risk. Vega is a volatility derivative, but also the exposure a vol trader is paid to carry. Internalising both views is the difference between knowing the formulas and being able to use them.

Delta — the slope

Delta is the first derivative of the option price with respect to the spot:

Δ = ∂V/∂S

For a European call, Δ ranges from 0 (deep out-of-the-money, the option is almost certainly going to expire worthless) to 1 (deep in-the-money, the option behaves like the stock itself). For a European put, the range is −1 to 0 by symmetry. Under Black-Scholes the closed-form is

Δ_call = N(d_1)
Δ_put  = N(d_1) − 1
d_1    = [ln(S/K) + (r + σ²/2) T] / (σ √T)

The hedging interpretation is the central one. A long call with Δ = 0.42 is equivalent, to first order, to holding 42 shares. To cancel that exposure, you short 42 shares against every 100 calls. The portfolio is now delta-neutral: a $1 wiggle in the stock moves the calls and the share position by equal and opposite amounts. This was exactly the construction Black, Scholes, and Merton used in 1973 to derive their pricing formula — they showed that if you can hold a delta-neutral portfolio continuously, the option's expected return must equal the risk-free rate, and the drift of the underlying drops out of the price.

A second reading of Δ is as a risk-neutral probability proxy: under the risk-neutral measure, N(d₂) is the probability of finishing in-the-money. N(d₁) is close but not equal — it weights that probability by the share you receive on exercise. For at-the-money options near expiry both are close to 0.5.

Gamma — the curvature

Gamma is the second derivative of price with respect to spot, equivalently the first derivative of Delta:

Γ = ∂²V/∂S² = ∂Δ/∂S

Under Black-Scholes,

Γ = φ(d_1) / (S σ √T)

where φ is the standard normal density. Three facts about Gamma drive how options books are managed.

• It peaks at-the-money. Gamma is largest when the spot is near the strike, where Delta is in the middle of its transition from 0 to 1 and most sensitive to small moves.
• It explodes near expiry. The 1/(σ√T) factor blows up as T → 0. A 30-day ATM option has a moderate Gamma; a 1-day ATM option's Gamma can be ten times larger, and a 0DTE ATM option's Gamma is enormous.
• It is symmetric. By put-call parity, calls and puts at the same strike have identical Gamma — the convexity is the same, only the directional exposure differs.

Hedging interpretation: Gamma is the rate at which your delta hedge goes stale. If you delta-hedge a long-Gamma position (e.g. you own options), spot moves work in your favour — as the stock rallies, Delta rises, so your initial short share-hedge becomes too small, and you "collect" the extra. Conversely, dealers who sell options end up short Gamma and must rebalance their hedge in the direction of the move (buy as the market rallies, sell as it falls), amplifying intraday volatility. This is the mechanism behind the modern "gamma squeeze" — a coordinated buying frenzy that forces market-makers to keep buying to stay neutral.

Theta — time decay

Theta is the partial with respect to calendar time:

Θ = ∂V/∂t

By convention, Theta is reported as the change in price per day, with the sign convention that Θ is negative for a long option (its value erodes as time passes) and positive for a short. The Black-Scholes Theta for a non-dividend European call is

Θ_call = − S σ φ(d_1) / (2 √T) − r K e^(−rT) N(d_2)

Two regimes matter. Far from expiry, Theta is small and roughly constant: an ATM call on a $100 stock at 30% vol with 90 days to go might bleed about $0.04 per day. Near expiry, Theta accelerates: the same option a week before expiry could bleed $0.10+ per day. The ratio Θ / V is largest near expiry, which is why option sellers crowd into short-dated premium and option buyers prefer longer-dated.

The hedging interpretation is that Theta is the seller's daily wage. A trader who is delta- and gamma-hedged but short Theta is in effect paid Θ per day for absorbing the residual second-order risks. In a perfectly executed Black-Scholes hedge, Theta and Gamma exactly offset on average: the Theta the seller collects compensates them for the Gamma exposure they must bleed off via repeated rebalancing. Real life never quite matches the model, and the gap between realised volatility and implied volatility decides who profits.

Vega — volatility sensitivity

Vega is the partial with respect to volatility:

ν = ∂V/∂σ = S φ(d_1) √T

Vega is not a Greek letter at all — the name is a market coinage that stuck. It is conventionally reported as the change in option price per one-percentage-point change in σ. Three patterns dominate.

• Vega peaks at-the-money. Like Gamma, Vega is largest when S ≈ K and falls off as the option moves deep in- or out-of-the-money.
• Vega grows with √T. Longer-dated options are vastly more sensitive to volatility — because the standard deviation of log-returns scales as σ√T, the option price is most exposed to σ when there is a lot of time for that σ to play out. A 1-year ATM call on $100 might have Vega ≈ $0.40; a 1-day ATM has Vega ≈ $0.02.
• Vega is positive for both calls and puts. Higher volatility increases the value of optionality regardless of direction.

For volatility traders, Vega is the headline exposure. A "long-vol" book is long Vega; it profits when realised or implied vol rises. A "short-vol" book is short Vega and profits when vol stays low or falls — the strategy that destroyed LTCM in 1998 and that periodically blows up funds in single bad sessions (notably XIV in February 2018 during "Volmageddon").

Rho — rate sensitivity

Rho is the partial with respect to the risk-free rate:

ρ_call = K T e^(−rT) N(d_2)
ρ_put  = − K T e^(−rT) N(−d_2)

Rho is the smallest of the headline Greeks for short-dated options on equities. A 30-day ATM call on a $100 stock might have Rho near $0.03 per percentage-point of r — meaningful for a 100,000-contract dealer book, negligible for a retail position. Rho matters most for long-dated options (LEAPS), for interest-rate-derivative products like caps and swaptions where r is the underlying, and for FX options where two rates (domestic and foreign) appear via Garman-Kohlhagen (1983).

Sign convention: long calls are long Rho (higher rates increase the forward, which helps the call); long puts are short Rho.

Worked example — Greeks of a 30-day ATM call

Consider a call on a $100 stock with K = $100, T = 30/365 = 0.0822 years, r = 4%, σ = 25%. Then

Quantity	Computation	Value
d1	[ln(1) + (0.04 + 0.03125)·0.0822] / (0.25·√0.0822)	0.0817
d2	d1 − σ√T = 0.0817 − 0.0717	0.0100
N(d1)	Φ(0.0817)	0.5325
N(d2)	Φ(0.0100)	0.5040
φ(d1)	(1/√2π)·e−d₁²/2	0.3976
Call price	100·0.5325 − 100·e−0.00329·0.5040	$2.84
Δ (call)	N(d1)	0.5325
Γ	φ(d1) / (S·σ·√T)	0.0555 per $1
Θ (per year)	−Sσφ(d1)/(2√T) − rKe−rTN(d2)	−19.36 → −$0.053 / day
ν (per σ-point)	S·φ(d1)·√T / 100	$0.114
ρ (per r-point)	K·T·e−rT·N(d2)/100	$0.041

Reading the table: each $1 the stock moves changes the call by about 53 cents. Each day that passes erodes the call by about 5 cents. A one-point rise in implied vol (say 25% → 26%) adds about 11 cents. A 100-bp rise in r adds about 4 cents. And Gamma tells you that today's Δ of 0.53 will rise by ~0.055 if the stock moves $1 — meaning your delta-hedge becomes stale that fast.

Delta-hedging in practice

The textbook prescription is continuous rebalancing. In practice, dealers rebalance at discrete intervals chosen to balance two costs: transaction costs (each trade pays the bid-ask spread) and gamma slippage (the longer you wait, the more your hedge drifts away from neutral). The optimal rebalance frequency rises with Gamma. A book that is short a thousand at-the-money 1-day options is rebalanced almost continuously during the trading session; a long-dated equity-index hedge might be touched only once a day.

A common mental model: imagine a dealer who has sold 1000 calls each with Δ = 0.5, Γ = 0.05. Initial delta-hedge: long 500 shares (the negative-of-the-short-calls' delta). If the stock rallies $1, the calls' total Delta rises by 1000·0.05 = 50, so the dealer is now under-hedged and must buy 50 more shares. If the stock then falls $1, Delta retreats, and the dealer sells those 50 shares — at a lower price. That is the gamma cost: the seller buys high and sells low to maintain neutrality. The Theta the seller earned during that time should, on average, exactly compensate for the gamma slippage — if realised volatility matches implied. When realised > implied, the seller loses money; when realised < implied, the seller wins.

Second-order Greeks

The five letters above are the first-order Greeks. Sophisticated trading books also carry second-order Greeks, which describe how the first-order Greeks themselves move.

Greek	Definition	What it measures	Where it matters
Vanna	∂Δ/∂σ = ∂ν/∂S	How Delta shifts as vol moves; or Vega as spot moves	FX risk-reversal traders, skew books
Charm	∂Δ/∂t	Delta decay — how the hedge ratio drifts as time passes	Month-end dealer rebalancing
Vomma (Volga)	∂²V/∂σ² = ∂ν/∂σ	Curvature of Vega in vol — convexity in volatility	Long-OTM "vol of vol" structures
Speed	∂Γ/∂S = ∂³V/∂S³	Rate of change of Gamma — third order in spot	Knock-in/out exotic option desks
Zomma	∂Γ/∂σ	How Gamma changes with vol	Cross-gamma hedging
Color	∂Γ/∂t	Gamma decay over time	Theta-Gamma scalping strategies

These show up explicitly on bank trading systems but rarely on retail brokerage platforms. The instinct that they encode — "the Greeks themselves are not static" — is what separates a model that works in the abstract from a model that works on a Friday afternoon when positions roll.

0DTE — zero days to expiry

The Chicago Board Options Exchange listed daily-expiry SPX options in 2022 and rapidly extended to QQQ and other indices. By 2024 these zero-day-to-expiry contracts ("0DTE") accounted for roughly 45 percent of all S&P 500 options volume and have become a significant intraday-flow driver in their own right. Two Greeks dominate their behaviour.

• Gamma is extreme. An ATM 0DTE has the largest Gamma of any listed equity-index option. Hours before expiry the Gamma profile is a near-spike at the strike. A $5 move in SPX through the strike can swing the option's Delta from 0.2 to 0.8 in seconds.
• Theta is extreme. The entire premium decays in one trading session. A 0DTE option that starts the day at $1.20 and ends out-of-the-money expires at zero by 4pm.

Market-makers who write 0DTE options end up short Gamma in size. To stay delta-neutral, they buy futures into rallies and sell into dips — a feedback loop that amplifies intraday moves. Several sharp afternoon moves in the S&P 500 since 2022 have been attributed by post-mortems to 0DTE-driven gamma squeezes; both the SEC and CFTC have flagged this as a structural concern worth watching. The current consensus is that 0DTE flow has changed intraday volatility microstructure but has not yet produced a systemic event.

The volatility surface

Black-Scholes assumes σ is constant. Market quotes do not. Plot implied σ as a function of strike (or moneyness) and time-to-expiry, and you trace out a two-dimensional surface that is anything but flat. Two features dominate.

• Skew (smile). For equity indices since 1987, out-of-the-money puts trade richer than equivalent calls — crash-protection demand pushes their implied vol higher. The graph of σ vs strike slopes down to the right, sometimes called a "smirk." For single names and FX, the curve is more symmetric ("smile").
• Term structure. Short-dated implied vol is usually below long-dated in calm markets ("contango") and can flip above ("backwardation") during stress.

Traders quote options in implied-vol points rather than dollars precisely because the surface is the actual market consensus on σ. The pricing model is just the unit converter. Local-volatility (Dupire 1994), stochastic-volatility (Heston 1993), and SABR (Hagan et al. 2002) models all attempt to extend Black-Scholes so its pricing matches every point on the surface simultaneously.

The risk reversal — a delta-neutral vol bet

A risk reversal is a structured trade that goes long one call and short one put, both out-of-the-money at the same delta (typically 25-delta). By construction the position is approximately delta-neutral and zero-cost, but it is long Vanna — it profits if call-side implied vol rises relative to put-side, or if the underlying drifts up while vol skew compresses. Risk reversals are quoted as a vol differential in FX markets ("25-delta RR 1y EURUSD: +0.4 vol") and are a clean instrument for taking views on skew without taking directional exposure to spot.

LTCM 1998 — short Vega writ large

Long-Term Capital Management was the highest-profile collapse of an options-trading fund in history. Two Nobel laureates (Myron Scholes and Robert Merton) sat on its board, and yet within four years of its 1994 launch the fund lost $4.6 billion and required a $3.6 billion Fed-organised bailout. The post-mortem is instructive in Greek terms.

A large piece of LTCM's book consisted of selling long-dated equity-index volatility — they were betting that implied σ at 5-year tenors was elevated relative to realised vol that would ultimately materialise, and that the trade would mean-revert. In Greek terms they were short Vega in size. Their pricing was probably correct, in the sense that the trade did eventually mean-revert. But the path to that mean reversion went through August 1998: Russia defaulted, global volatility spiked rather than decayed, and the short-vol book mark-to-market losses triggered margin calls. The fund's 25:1 leverage and concentrated short-Vega positioning meant that the losses cascaded faster than any orderly unwinding could keep up.

What killed LTCM was not the Greeks themselves — Black-Scholes worked exactly as advertised. What killed it was the combination of (a) concentrated short-Vega risk, (b) extreme leverage, and (c) the assumption that diverse positions would remain uncorrelated. When global liquidity dried up, every previously uncorrelated trade became one trade. The lesson burnt into every modern risk department: Vega exposures are netted across the firm and stress-tested against scenarios in which all correlations move to 1.

Real-world applications

• Market-making. Every option market-maker on every exchange runs a real-time Greek-space risk system that aggregates Δ, Γ, ν, Θ, ρ across thousands of strikes and expiries. Rebalancing decisions are driven by aggregate exposures, not individual positions.
• Volatility arbitrage. Long a delta-hedged option, you collect realised gamma if realised vol exceeds implied. Short the same, the reverse. Volatility-arbitrage funds (e.g. Capstone, LongTail Alpha) make this their core trade.
• Structured products. Bank-issued autocallables, accumulators, and dual-currency notes are decomposed by trading desks into baskets of vanilla options whose aggregate Greeks must be hedged.
• Risk reporting. Basel-style market-risk capital is computed against scenarios that shock spot, vol, and rate jointly — i.e. against the full Greek surface, not single-Greek sensitivities.
• Retail flow. Robinhood-style platforms display Delta and (sometimes) Theta to retail traders. The 0DTE option boom of 2022-2024 made "Theta-positive selling strategies" a meme-stock category in its own right.
• Insurance pricing. Variable annuities with embedded equity-linked guarantees are priced and hedged with the same Greek toolkit; a $3-trillion industry depends on dynamic Delta-Vega hedging by insurer treasuries.

Common pitfalls

• Confusing Delta with probability. Delta is close to but not equal to the risk-neutral probability of finishing in-the-money — that probability is N(d₂), not N(d₁). They diverge when the option is far from at-the-money or has long expiry.
• Treating Greeks as constants. Every Greek is itself a function of spot, time, vol, and rate. A book that is delta-neutral today is not delta-neutral tomorrow; that is what the second-order Greeks (Charm, Vanna) quantify.
• Ignoring Gamma in 0DTE. A position that looks small in dollar terms can carry enormous gamma exposure if it consists of short-dated at-the-money options. The 2024 SPX 0DTE volume reset what "small" means for a market-maker book.
• Assuming Vega scales linearly. Vega is itself dependent on σ via Vomma. In a vol spike, short-Vega positions lose more than a static Vega calculation suggests, because their Vega itself grows.
• Forgetting put-call parity. Greeks for puts and calls at the same strike are connected by P = C − S + Ke^−rT: Δ_put = Δ_call − 1, Γ_put = Γ_call, ν_put = ν_call. Mistakes here cost money instantly.
• Hedging only Delta. A delta-neutral book can still lose massively from a vol move (Vega) or a time pass (Theta-Gamma mismatch). Real desks hedge the Greek profile, not the single Greek.