A risk signal fired and I sat out per the rules, but that's exactly the stock that ended up rallying the most that week. Is it still worth following these stop-loss rules?
2026-W38
The rules went 3-for-5 this week
Of the five scoreable signal discussions this week, three hit and two missed. The two misses sting. On September 14, Micron was flagged to avoid because its stop-loss buffer of 3.64% fell 0.36pp short of the 4% threshold — and it then rallied +9.93%. The same day, the semiconductor ETF was also skipped for a 0.60% stop-loss buffer, missing a +5.82% move. Following the rules cost the two biggest rebounds of the week.
But this avoidance isn't a malfunction — it's the design working as intended.
The day the entry condition was on happened to be the worst day
| Date | SMH close | 10-day MA | 50-day MA | Entry condition | Stop-loss buffer | Off high |
|---|---|---|---|---|---|---|
| 09-14 | 541.50 | 559.03 | 568.09 | Met | 0.60% | -19.40% |
| 09-15 | 542.11 | 557.57 | 566.84 | Not met | 0.69% | -19.31% |
| 09-16 | 545.56 | 557.61 | 566.13 | Not met | 1.21% | -18.79% |
| 09-17 | 560.61 | 558.62 | 565.48 | Not met | 3.45% | -16.55% |
| 09-18 | 573.00 | 560.66 | 564.78 | Met | 5.29% | -14.71% |
Reading this table top to bottom reveals the week's real structure. The entry condition was on during the -4.75% plunge day (09-14), stayed off through the entire three-day rally, and switched back on only after the rally had already finished, on Friday's close. The trend-following rule's inherent lag played out in the least favorable way possible within a single week.
Why the rule shouldn't be changed anyway
This isn't unique to this strategy — it's a mathematical property of trend-following rules in general. A market's best days and worst days tend to cluster in the same crisis window. Trend rules stay fully invested going into a quiet uptrend, get caught by the crash, switch to defense only after prices have already fallen, and often stay defensive even as the sharpest rebounds fire off near the bottom (Summitward, 2026). Avoiding the bad days and missing the good days are, by nature, the same trade.
What matters is the net total. The same analysis found that the gains from avoiding the worst days outweighed the losses from missing the best days, so removing both together beat simple buy-and-hold by 0.47pp annually (Summitward, 2026). That's why the industry calls whipsaw losses "the cost of doing business to avoid bear-market devastation," and why there's a recurring warning that tweaking a model to reduce recently experienced whipsaw tends to produce worse outcomes in other regimes (Proactive Advisor Magazine).
This week's two misses aren't evidence to fix the rule — they're evidence the rule is functioning. Had the September 14 plunge continued for three more days, the same verdicts would have prevented losses. This time, the rebound simply came instead.
What should change isn't the rule — it's position sizing
What actually generated returns this week wasn't loosening the rule, but splitting size. On September 17, once the stop-loss buffer recovered to 7.90%, Micron was taken at half size for +3.92%; the same day the semiconductor ETF was taken at half size for +2.21%. Using three sizing tiers — 0, 0.5, and 1 — instead of a binary in/out decision is what actually reduced the cost of avoidance.
Applying that same principle next week looks like this.
| Ticker | Weekend stop-loss buffer | Judgment ahead of the event |
|---|---|---|
| SMH | 5.29% | Below the 8% safe zone. Keep at half size, don't add until it recovers |
| AIPO | 1.95% | Even meeting the condition, this is stop-out distance from entry. Hold off |
| BOTZ | 4.22% | Negative in a week semiconductors rose. Exclude until relative theme weakness is confirmed |
| CPER, DBB | 12.43%, 12.12% | The only two names with enough buffer to absorb the event. Drivers independent of rates |
Two confirmed catalysts — the September 24 US-China summit and the September 30 Micron earnings — sit just two days apart. A 5.29% stop-loss buffer is a distance that could be wiped out by a single day's move, and event outcomes can't be known in advance. Trimming size ahead of an event and breaking the rule to enter anyway are two entirely different actions.
What should investors do
The most common failure mode is wanting to lower the threshold right after an experience like this week's "the rule cost us a missed trade." Lowering the threshold from 4% to 3% would have caught Micron this week, but in a scenario where the September 14 decline had continued, that same lower threshold would have meant entering right at the stop-out line. Rather than changing the threshold value, it's better to add just one refinement: create a tier where a stop-loss buffer that narrowly misses the threshold gets a 0.25-0.5 size entry instead of zero. September 17's half-size verdict already demonstrated how effective that approach can be.
Detailed charts and signals for other tickers are available on the /signals dashboard.