Stop Losses on Put Credit Spreads: What the Data Says

By Charlie Nguyen · Published Nov 4, 2026 · Model data Jan 3, 2011 to Oct 23, 2026 · 11 min read

A put credit spread stop loss is supposed to save the account when a short strike goes wrong. I tested the three rules traders argue about on SPX put credit spreads over 15 years of model data: no stop, a 2× credit stop, and a short-strike breach stop, plus a 3× credit stop and combinations with a 50% take-profit. The short version: stops cut the worst 1% and the max drawdown, they cost EV per day, and a breach stop is a worse deal than 2× for the risk it removes.

The three stop rules traders argue about

• No stop. Hold to expiration (or to a profit target). Max loss is the spread width minus the credit. Defined risk, undefined pain when it hits.
• 2× credit stop. Buy the spread back when the debit to close reaches twice the credit. Sold for $1.50? Stop at $3.00. You lose about one credit, plus fees, on a slow grind against you.
• Short-strike breach. Close when the underlying closes at or below the short put strike. The idea: get out once the trade is “in trouble,” before it becomes a max loss.

I also ran a 3× credit stop (looser) and each stop paired with a 50% take-profit, the rule tested in when to take profits on credit spreads.

What a stop is really buying

A stop does not create edge. Selling a put credit spread outside the expected move is the edge; a stop is insurance you pay for with lower EV. What you buy is:

• A smaller worst 1% of trades on gradual selloffs.
• A smaller max drawdown when you run one spread at a time.
• Fewer max losses as a share of trades.
• The bill: a lower win rate (you lock in losses that sometimes would have recovered), more commissions, and less EV per day.

How I tested it

Every trading day from Jan 3, 2011 to Oct 23, 2026, the model opened two SPX put credit spreads at the close:

• 45 DTE, 10 delta, $20 wide (31 trading days to expiration). Median model credit $0.93, short strike a median 8.5% below SPX.
• 7 DTE, 9 delta, $50 wide (5 trading days). Median model credit $1.73, short strike a median 3.1% below SPX.

Each spread was marked at every daily close and closed by one of the exit rules above. Stops and take-profits are checked at the close only; there is no intraday path. See POP vs POT for why the short strike gets touched more often than it finishes in the money.

Model backtest, not real fills. Option prices are Black-Scholes, with implied volatility from Cboe’s VIX9D (0.77 × VIX9D) for 7 DTE and VIX (0.73 × VIX) for 45 DTE, plus a put skew (σ rises 0.23–0.33 × per standard deviation below the forward) fitted to a real SPX put chain. Risk-free rate from the 13-week T-bill, dividend yield 1.8%. Fills at the model mid, checked at the daily close only, no slippage. Fees $2.60 per spread to open and again to close early, nothing at expiration (cash settled). P&L is per one spread ($100 multiplier). Results are hypothetical.

Results: 7 DTE spreads

What it shows. Holding with no stop made the most per day ($12.25) and won 96.2% of the time, but the worst 1% of trades lost about −$4,772 and the one-at-a-time max drawdown was −$20,900. A 2× credit stop cut that worst 1% to −$1,329 and the max drawdown to −$6,720, with max losses falling to 0.10% of trades from 1.36%. The price was EV: $9.60 per day and a win rate of 81.3% (the stop fired on 18.3% of trades).

A 3× credit stop sat in between: $10.12/day, worst 1% −$1,643. The breach stop kept more EV ($10.19/day) but barely helped the tail (worst 1% −$3,109, max drawdown −$19,470). By the time SPX closes through the short strike, the spread mark is often already deep in the hole.

Results: 45 DTE spreads

What it shows. The same pattern, louder on fees. Holding made $1.22/day; a 2× stop made $0.83/day and crushed the worst 1% from −$1,911 to −$354. The breach stop was the worst of the three for EV ($0.31/day) and still left a deep worst 1% (−$1,373). On a small $0.93 median credit, every extra commission matters, so a tight stop that fires often (28.4% of trades at 2×) also taxes the winners.

Why the worst trade barely moves

Look at the “Worst trade” column: it stays near a full max loss under almost every rule (−$4,879 vs −$4,841 on 7 DTE). That is the overnight gap. The model only sees the daily close, so a crash that opens the spread near max loss is booked before any stop can help. Stops truncate the slow losers; they do not erase gap risk. What changes is the worst 1% and how often you take a max loss, not the single worst print. Position size is still what keeps a gap from ending the account.

Pairing a stop with a 50% take-profit

Article #5 found that a 50% take-profit raised EV per day on short-dated SPX spreads. Pairing it with a 2× stop is the “belt and suspenders” setup:

• 7 DTE: 50% alone made $13.95/day with a 1.11% max-loss rate. Add a 2× stop and EV falls to $8.48/day, but max losses fall to 0.03% and the worst 1% to −$1,129.
• 45 DTE: the same combo nearly wiped EV ($0.19/day) because the credit is small and the stop fires often. Fees on a $0.46 winner hurt; fees on a stopped-out loser hurt more.

Use the combo when the credit is large enough that a second commission is a small slice of the target, and when cutting the max-loss rate matters more to you than maximizing EV per day.

Example: setting the stop on a real spread

Here is the SPX Desk’s 7DTE model spread from the Nov 4, 2026 close, with the 2× and 3× stop levels you would set next to a 50% take-profit. P50 is the chance of hitting that 50% target before expiration. It is an illustration, not a recommendation.

EXAMPLE, model figures. Cboe delayed SPXW quotes at the close, credit at the mids, SPX 7,688.62. POP = chance of any profit at expiration; POT = chance SPX touches the short strike first; P50 = chance the spread can be bought back for half the credit. 2× stop = buy back at $2.60; 3× stop = buy back at $3.90. Lognormal model at the short strike’s IV 16.6%, rate 4.03%, P50 from 40,000 simulated paths. See POP vs POT vs P50 and IV rank explained.

The advantage-play view

A good advantage player buys insurance when ruin is the enemy, not when it maximizes hourly win rate. Same here:

• EV per day still matters. A stop that cuts EV in half needs a clear reason: a bankroll that cannot absorb the no-stop drawdown.
• Pay for the tail you can actually cut. 2× credit helped the worst 1% a lot; breach helped it little. Buy the rule that moves the distribution, not the one that feels active.
• Gaps are a sizing problem. No daily stop deletes an overnight crash. Size so one max loss is a small slice of the account (more in bankroll management for +EV options).
• Don’t let fees eat a small credit. On 45 DTE $20-wide spreads, a tight stop fired often and the second commission erased most of the edge.
• Never blow up. The stop is a seatbelt. The speed limit is position size.

My stop-loss rules

1. Size first. Max loss on any one SPX put credit spread stays a small, fixed slice of the account. That is the real stop.
2. Short-dated SPX spreads: 50% take-profit at entry, 2× credit stop as a risk cap. The take-profit is for EV per day; the stop is for the worst 1%.
3. Skip the breach stop. It fires late, after the mark is already ugly, and barely helps the tail.
4. Small credits: prefer a wider stop or none. If a second commission is a big share of the credit, a tight stop taxes you twice.
5. Pick strikes outside the expected move with Open Lines (why), and sell more when IV rank is high (IV rank explained).
6. Trade SPX for cash settlement and Section 1256 tax treatment (SPX vs SPY options), so a stop never leaves you holding shares.

Sources and method. SPX, VIX, VIX9D and 13-week T-bill daily closes from Yahoo Finance (Jan 3, 2011 to Oct 23, 2026); VIX and VIX9D are calculated by Cboe. 3,945 45 DTE and 3,970 7 DTE overlapping daily entries (every trading day). “One at a time” runs a single spread, re-entering at the next close after each exit, and reports its total P&L and max drawdown per spread. Win = P&L above zero after fees. Max loss = a loss of 90% or more of the spread’s max risk. Stop / breach hit = share of trades closed by that rule before expiration. Model figures, not historical option quotes; results are hypothetical. Example: Cboe delayed SPXW quotes (Nov 4, 2026 close); SPX contract terms from Cboe SPX specifications. Code: raw/stoploss/backtest.py.

FAQ