Stop-losses on credit spreads did not work
We replayed every exit rule we could think of on put credit spreads. Holding to expiry beat all of them, because the big losses arrive as gaps, not as slow drifts a stop can catch.
The conclusion first
On real end-of-day bid and ask quotes from 2022 onward, net of fees, holding a 25-delta put credit spread to expiry earned more than taking profit at 25%, 50% or 75% of the credit, and far more than closing at 21 days to expiry. That held on four different rule sets and two underlyings, from 42 to 396 trades each.
Taking profit at half the credit gave up between 13% and 55% of the total earned, depending on the cell. Closing at 21 days to expiry was negative on both leveraged-ETF cells while holding was positive on both. The one exception was a 75% take-profit on the largest QQQ cell, which beat holding by a margin small enough to be noise.
Why the win rate goes up while the money goes down
Every early exit raised the win rate. Taking profit at a quarter of the credit pushed winners to about 97% of trades. That is the trap: a rule that banks small wins earlier looks safer on a scorecard and earns less, because it gives back the part of the credit that the quiet trades would have kept anyway.
Meanwhile the worst trade was the same under every rule except the 21-day exit. A daily mark cannot dodge a loss that happens overnight. When the underlying gaps through both strikes, the spread is at or near its maximum loss before any stop can be checked, so the stop fills near the bottom and then also clips the trades that would have recovered.
Gaps, not drift
A stop-loss assumes losses build gradually, so you can step out partway. On a defined-risk short put spread the dangerous moves are discontinuous: an earnings-like gap in an index, a macro shock, a weekend. The spread's maximum loss is reached in one print. That is also why a credit spread is the structure where the wing does the risk work: the wing caps the loss in advance, which no stop can promise.
The same reasoning ran through our same-day SPX study: the worst days were full-width losses that no entry or exit rule saw coming. See the lesson on 0DTE for those numbers.
What it changes in practice
If you sell put credit spreads, decide the size of the loss when you open the trade, through the width of the spread and how much of your account is at risk, rather than through an exit rule you hope to execute later. Our sizing table, in the lesson on drawdown odds, shows what fraction of equity at risk kept the odds of a 20% drawdown low.
What this study did not measure is capital recycling: an early close frees collateral sooner, so per year of capital a take-profit could look better if the freed cash is redeployed at the same edge right away. That is a separate question with its own test. On a per-trade basis, the answer was hold.
Historical replays are not a forecast. Options can lose more than the premium collected. This is education, not a recommendation to trade anything.