A credit spread's exit rules are set at entry, not improvised later: close at a fixed fraction of the credit collected (commonly 50%), exit by around 21 days to expiration, and stop the loss at a multiple of the credit (commonly 2x). Mechanical exits exist because the final weeks hold the least remaining reward and the most gamma risk.
The second half of a credit decays slower and riskier than the first. Once half the premium is captured, the remaining reward shrinks while gamma grows into expiration, so the position holds all of its original risk for a diminishing payoff. Closing at 50% recycles capital into fresh, fully-paid setups and cuts the time each position spends exposed to a gap. Backtests of short premium disciplines have repeatedly found managed-early variants smoothing outcomes versus held-to-expiry ones, and the reasoning is visible without a backtest: reward decays toward zero, risk does not.
Gamma. Inside roughly three weeks to expiration, an out-of-the-money short strike stops behaving statistically and starts behaving like a coin flip on every sizable move. The time exit closes positions while adjustment is still possible at reasonable prices, regardless of profit or loss, and pairs with the 30-45 DTE entry convention to define the window where the structure keeps its intended character.
On the credit, not on a chart level: a common stop closes the spread when buying it back costs a fixed multiple (often 2x) of the credit collected. Anchoring on credit keeps the stop proportional to what the trade was paid, and because a defined-risk spread's absolute worst case is already capped at width minus credit, the stop's job is narrower than in undefined-risk selling: it converts a maximum loss into a smaller realized one and frees the capital.
Entry rules exist to keep bad trades out; every additional entry gate makes the book safer. An exit gate does the opposite: any condition that can delay an exit is a condition that can hold risk open. The standing principle in Options Scanner's rules engine is that risk reduction always fires: exits are evaluated continuously and no volatility screen, event window, or scoring layer may veto one. The asymmetry is the point.
The second half of the credit decays more slowly and has to be defended against gamma risk that grows as expiration nears. Closing at half captures the fastest part of the decay curve and recycles the buying power, and the practice is common enough that several published studies have measured it against hold-to-expiration baselines.
Multiples of the credit received, commonly between one and two times the credit, or a touch of the short strike. Both convert the vague sense that a trade is not working into a number that can be checked mechanically, which is the property an unattended monitor needs.
Expiration week concentrates gamma: a small move in the underlying swings the spread's value hard, and pin risk around the short strike adds assignment uncertainty. Time-based exits around 21 days to expiration exist to exchange the remaining slow decay for the removal of that regime.