Probability of Touch vs. Probability of Expiring ITM: Why They're Not the Same Number
Two different questions, one number often answers both — wrongly
A trader looks at a 30-delta call and says "there's roughly a 30% chance this finishes in the money." Fine — that's a reasonable read of what delta is approximating. But that number answers a narrower question than it sounds like it does: what's the probability the stock is beyond the strike at the single frozen instant of expiration? It says nothing about whether the stock wanders through that strike and back out again at some point before then — a path outcome that matters enormously if there's a stop-loss order, a margin threshold, or just a trader's nerve sitting between now and expiration.
Delta is an expiration-only snapshot
Delta (technically N(d1) in Black-Scholes) tracks closely with N(d2), the model's exact risk-neutral probability that an option expires in the money — close enough that traders use delta as a probability proxy every day, with the caveats covered in why delta isn't quite a real probability. Either way, both numbers describe the same kind of event: where the stock happens to be at one specific moment, expiration. Nothing in that calculation asks whether the stock crossed the strike an hour after you opened the trade and drifted back by the close. For a continuous price path, that's a completely different question with a completely different answer.
The reflection principle: why touching runs roughly double
Here's the piece of math underneath it. For a driftless random walk, every path that ends up beyond a strike at expiration necessarily touched that strike at some point along the way — prices move continuously, so there's no way to finish past a level without crossing it first. But there's a mirror-image set of paths that also touch the strike and then reverse, finishing back on the starting side. In the symmetric, no-drift case those two groups are roughly equal in size, which gives the classic result from Brownian motion known as the reflection principle: the probability of touching a barrier by time T is about twice the probability of finishing beyond it at time T. It's the same identity that makes a one-touch binary option price richer than a European digital struck at the same level and expiring on the same day — touching is simply the easier bar to clear.
Where the gap actually costs you
This isn't just a math curiosity. A short-premium seller who picks a strike at "20-delta, so an 80% chance this expires worthless" and sets a mental stop at "if it hits 40-delta, I'm out" is quietly relying on touch probability, not the expiration probability used to choose the strike in the first place — and touch probability is the bigger number. That strike gets tagged, and the stop trips, far more often than the original 20% framing implied. It's also why a short option can look like a loser mid-trade — triggering anxiety, an early roll, or an unplanned adjustment — on a path that would have finished comfortably OTM if left alone. None of that shows up in a "probability of profit" figure quoted only at expiration.
The same gap matters on the other side of assignment risk: an option can trade deep enough ITM intraday to make early exercise economically attractive to the holder of the opposite side, even on a name that finishes OTM by the close. The terminal probability said this was unlikely. The path probability, which is what assignment risk actually depends on, was never that small to begin with.
The rule of thumb, and where it breaks
"Touch probability ≈ 2 × delta-implied ITM probability" is a useful sanity check for a strike sitting reasonably close to the money with weeks, not months, on the clock — but it's an approximation built on a driftless, symmetric assumption that real markets don't actually satisfy. A stock's risk-neutral drift isn't zero, dividends push it further off, and volatility skew bends the picture more than either of those: since OTM puts carry richer implied vol than OTM calls, downside touch probabilities run fatter than the simple reflection-principle multiple would suggest, while upside touches run a bit thinner. Treat 2x as an intuition check on how much bigger touch risk is than terminal risk, not as a number to price a trade off of directly.
Checking it against what actually happened
The cleanest way to internalize the difference is to stop reasoning about it abstractly and look at real paths: pull a strike's delta at trade entry from the Greeks Matrix, then check the underlying's own historical candles for how often it actually traded through that level intraday versus how often it merely closed beyond it on expiration day. The two counts are rarely close, and seeing the gap on a name you actually trade does more than any formula to recalibrate how a stop or an adjustment rule should really be set.