This week the semiconductor ETF was pushed to its stop line, and new buy signals turned on for health care, Eli Lilly and copper. Can I switch according to the signals, and was the midweek call to 'buy only half' correct?
2026-W42
Five days of two semiconductor names
These are values from the signal history for 10-12 through 10-16. The 'stop buffer' is the distance remaining until the price reaches the trailing stop line.
SMH (semiconductor ETF)
| Date | Close | 10-day MA | 50-day MA | 150-day MA | From peak | Stop buffer | Entry condition |
|---|---|---|---|---|---|---|---|
| 10-12 | 590.99 | 615.73 | 579.25 | 544.40 | -12.03% | 7.97% | Not met |
| 10-13 | 588.12 | 613.86 | 580.10 | 545.67 | -12.46% | 7.54% | Not met |
| 10-14 | 564.74 | 609.43 | 579.88 | 546.77 | -15.94% | 4.06% | Not met |
| 10-15 | 558.84 | 603.53 | 579.66 | 547.90 | -16.82% | 3.18% | Not met |
| 10-16 | 539.79 | 594.45 | 579.03 | 548.92 | -19.65% | 0.35% | Not met |
The 10-16 close fell below the 150-day MA for the first time.
MU (Micron)
| Date | Close | 10-day MA | 50-day MA | 150-day MA | From peak | Stop buffer | Entry condition |
|---|---|---|---|---|---|---|---|
| 10-12 | 989.04 | 1,055.39 | 979.51 | 814.26 | -21.18% | 8.82% | Met |
| 10-13 | 953.19 | 1,044.20 | 981.98 | 817.93 | -24.04% | 5.96% | Met |
| 10-14 | 914.13 | 1,028.95 | 982.26 | 821.11 | -27.14% | 2.86% | Met |
| 10-15 | 931.24 | 1,012.35 | 983.02 | 824.61 | -25.77% | 4.23% | Met |
| 10-16 | 878.84 | 992.77 | 982.97 | 827.63 | -29.95% | 0.05% | Met |
Even though MU closed below its 50-day MA from 10-13, its entry condition stayed met to the end. That is because the entry condition for the individual large-cap group is a slow signal that looks at moving average alignment. When direction turns sharply, as this week, the stop buffer warned first, and that warning was right.
The stop rule did its job this week
MIT's Kaminski and Lo showed that if prices move randomly a stop rule cuts expected returns, but if prices have trends a stop adds value. In monthly US stock data for 1950-2004, some stop rules added 0.5-1.0 percentage points per month during the stop periods, with long-term Treasuries as the refuge (Kaminski and Lo, Journal of Financial Markets, 2014). The authors linked this effect to periodic flight-to-safety phases.
This week was such a phase. The Philadelphia Semiconductor Index fell eight straight sessions, the longest streak since October 2008, and is down 16% since 10-06 (Investing.com, 24/7 Wall St., 2026-10-16, 10-17). While the decline continues in one direction, stops cut off losses.
There is a cost on the other side. When the streak of declines ends and a choppy market arrives, repeated re-entries after stops pile up costs. We found no study that calculates returns after streaks of declines in the semiconductor index in this search. So we do not use this decline as evidence that it is near a bottom.
Review of the 'buy only half' calls
Restating the weekly report's discussion review in numbers: the two half-size entry calls on MU ended at -7.80% and -5.63% afterward. The one call not to enter was right. The three half-size entry calls on XLV were +1.58% to +4.16% afterward.
Half-size entry only halves the size of the loss and does not change direction. The real variable that separated the calls was the stop buffer compared with that stock's daily range.
| Ticker | This week's average daily range | Stop buffer at the time of the call | Days it can withstand |
|---|---|---|---|
| MU | about 3.8% | 5.96%, 4.23% | about 1.6 days, 1.1 days |
| XLV | about 1.3% | 16.44-18.88% | about 12-14 days |
The average daily range is the mean of the absolute 5-session returns, and the number of days is my calculation of the buffer divided by that value. MU's buffer of 4-6% was a distance that disappears with a day or two of movement. Even the 7% buffer threshold the weekly report suggested is under 2 days for MU and over 5 days for XLV. The same percentage means different things for different stocks. A criterion worth considering is to not enter if the buffer divided by the daily range is under 3 days. It is not a criterion validated by backtest.
The new buy signals are of different quality
| Ticker | 10-16 close | 10-day MA | 50-day MA | 150-day MA | From peak | Stop buffer | Signal quality |
|---|---|---|---|---|---|---|---|
| XLV | 176.71 | 171.18 | 169.61 | 157.13 | -1.20% | 18.80% | Price above all moving averages. The cleanest |
| LLY | 1,257.03 | 1,190.47 | 1,180.17 | 1,092.07 | -2.76% | 27.24% | Newly met on 10-15. Largest buffer |
| CPER | 39.50 | 39.92 | 39.88 | 38.26 | -4.87% | 10.13% | Newly met on 10-15, but price below the 10-day and 50-day MAs |
CPER has one hidden link. A major pillar of the copper bull case is AI data center demand, and the first bearish factor analysts cite is a cut in big tech capex (Skillings Mining Review and others, 2026). Copper futures set a record of $6.95 per pound on September 22 and are now down about 6% at $6.53. In other words, CPER's new signal partly relies on the same AI capex that brought SMH down.
LLY is a major constituent of XLV. Buying both means buying the same exposure twice.
What to do specifically
| Ticker | Action | Condition |
|---|---|---|
| SMH | Hold no weight | The theme group enters only when the short-term trend indicator (MACD) crosses above its signal line. The 10-16 value is -3.21, below the signal line of 6.20. Do not chase rebounds before a re-entry signal |
| MU | Unwind the holding, no averaging down | A buffer of 0.05% is effectively the stop line. Review again after SK hynix earnings on 10-27 and the FOMC |
| XLV | Enter in tranches at half size | Price is about 4% above the 50-day MA, so fill on a pullback near the 10-day MA at 171 |
| LLY | Only one of it and XLV | LLY if you want single-stock risk, XLV if you want diversification |
| CPER | Watch, if entering then very small | When price regains the 10-day and 50-day MAs and big tech capex guidance holds |
The criterion for switching is not a single signal but the quality of the signal. The stop signal was right this week so follow it, and start small with new buy signals, beginning with stocks where price is above the moving averages and the buffer is many times the daily range.
For detailed charts and signals on other tickers, see the /signals dashboard.