Theta decay, measured on 19 years of real SPY quotes: ATM extrinsic dies ~2× faster than theo
None of this is new or novel, I'm posting this moreso for newer traders to get a feel for some of the more nuanced relationships between theoretical pricing and markets.
* Data: end-of-day SPY option chains, Jan 2007 → Jul 2026, marked at bid/ask \*\*mid\*\*.
* For every SPY expiration, on the first trading day with 43–46 DTE I "select" three structures at that day's chain: the ATM straddle (strike nearest spot), the 25Δ strangle, and the 10Δ strangle (strikes nearest those deltas). Entry requires live bids on every leg.
* I then track those \*\*exact fixed strikes\*\* every day to expiration and compute \*\*extrinsic (time) value\*\* = mid − intrinsic at that day's spot.
* Each path is normalized by its entry extrinsic; the curves below are the \*\*median across all 796 qualifying expirations\*\*, with interquartile bands on the chart.
* 796 expirations = every monthly plus every weekly that had quotes \~45 days out.
\*\*The ATM curve is nothing like the textbook.\*\*
By 21 DTE the median real ATM straddle has lost 71% of its extrinsic. The spot-pinned BSM curve (what every theta-decay illustration actually plots) says it should have lost 31%. The real curve decays at roughly \*\*double\*\* the textbook rate, the whole way down.
\*\*The ranking is inverted.\*\*
Theory says a fixed OTM strike decays \*faster\* than ATM. In the data, the wings \*\*outlive\*\* the body: at 21 DTE the 25Δ strangle still holds 57% of its extrinsic while the ATM straddle holds 29%.
https://preview.redd.it/4x78uoh7yhbh1.png?width=2782&format=png&auto=webp&s=2de2aadf4f8e1bc341e4909bc5fc0c28aad6b70a
https://preview.redd.it/m4caroh8yhbh1.png?width=2086&format=png&auto=webp&s=55343c4b043d19acf99aef21a92f724903d0fdb8
https://preview.redd.it/8ukw6a0ciibh1.png?width=2385&format=png&auto=webp&s=572dc5f40b2ce3ed064a1ae941a6eed6cc28d936
Why this happens (no mystery, no "edge" claim):
* The textbook curve assumes spot stays glued to your strike. It doesn't. Spot wanders — which \*drains\* extrinsic from the ATM strike it left behind and \*feeds\* extrinsic to whichever wing it approaches.
* It's a gamma/path effect, priced in and well understood — but it means the decay chart everyone reasons from describes a contract that stops existing the moment the market moves. If you buy an ATM option and the market trends, your extrinsic doesn't follow the hockey stick; it follows the yellow line.
What it's NOT saying: extrinsic disappearing is not seller P&L.
* A short straddle "collects" that extrinsic but pays out intrinsic when spot moves — this chart is about where time value goes, not about who profits.
\*\*Limitations:\*\*
\- Calendar-day DTE → you can see a 7-day weekend sawtooth in the curves.
\- Medians of normalized paths; means are noisier but same shape.
\- Two data regimes (true EOD 2007–2014, last-hourly-snapshot 2015–2026) — I split the sample by era and the shape/ranking is identical in both halves.
\- Extrinsic floored at 0 (deep-ITM mids occasionally print under intrinsic).
\- SPY only. Single-name behavior (earnings, harder-to-borrow) will differ.
\- Deep-ITM puts near ex-div carry early-exercise richness — second order at these strikes.