The Nasdaq 100 ETF lost its buy signal this week. Should I sell now? Amazon and Google had signals turn on and then off within three days.
2026-W43
Losing a signal means stopping buys, not selling
In our system the entry condition and the exit are separate lines. The entry condition is set by the relationship between short- and medium-term moving averages, while the exit is set by a stop line (Trail stop) a fixed percentage below the peak. For core ETFs and large single stocks, 70% of the peak is the stop line. The Trail buffer shows, in percentage points of the peak, how far the current price is from that stop line.
QQQ, last five trading days
| Date | Close | MA10 | MA50 | MA150 | Entry condition | Trail buffer | From peak |
|---|---|---|---|---|---|---|---|
| 10-16 | 709.80 | 742.88 | 726.33 | 694.46 | Met | 23.04% | -6.96% |
| 10-19 | 718.36 | 739.10 | 726.25 | 695.26 | Met | 24.17% | -5.83% |
| 10-20 | 717.63 | 734.89 | 726.20 | 696.04 | Met | 24.07% | -5.93% |
| 10-22 | 704.47 | 727.33 | 725.98 | 697.68 | Met | 22.35% | -7.65% |
| 10-23 | 704.40 | 722.64 | 725.44 | 698.51 | Not met | 22.34% | -7.66% |
The 10-21 row in the signal history repeated the prior day's close, so it was removed from the table.
The 10-23 loss resulted from the 10-day average (722.64) falling below the 50-day average (725.44). The gap between them is only 0.39%. The 50-day average is still above the 150-day average, so the medium-term trend is alive. The stop line is 70% of the $762.86 peak, or $534.00, which is 24.2% below the current price. The nearer line in practice is the 150-day average at $698.51, a 0.84% gap from the close. QQQ is not at a point to sell. It has a reason not to buy more.
The same loss, different weight: Google
| Date | Close | MA10 | MA50 | MA150 | Entry condition | Trail buffer | From peak |
|---|---|---|---|---|---|---|---|
| 10-16 | 336.91 | 345.25 | 341.36 | 345.51 | Not met | 13.40% | -16.60% |
| 10-19 | 340.16 | 344.88 | 341.10 | 345.75 | Met | 14.21% | -15.79% |
| 10-20 | 336.40 | 344.06 | 340.71 | 345.93 | Not met | 13.27% | -16.73% |
| 10-22 | 320.76 | 340.01 | 340.05 | 346.20 | Not met | 9.40% | -20.60% |
| 10-23 | 317.28 | 336.95 | 339.52 | 346.32 | Not met | 8.54% | -21.46% |
GOOG's 50-day average (339.52) is already below its 150-day average (346.32). The 10-19 entry, which came with the medium-term trend broken, was a signal created by a brief rebound. The stop line is 70% of the $403.96 peak, or $282.77, 10.9% below the current price.
Next week's earnings bear directly on this. At its Q2 results on July 23, Alphabet raised its 2026 capex outlook from $180-190 billion to $195-205 billion and posted its first quarterly free cash flow deficit, -$5.9 billion. The stock fell -7.1% that day (Let's Data Science, S&P Global Market Intelligence, 2026-07). If the same reaction occurs, GOOG would be at about $295, -27.0% from its peak, and its Trail buffer would shrink to about 3.0%, entering the danger zone below 5%. It would still not reach the stop line.
AMZN met the entry condition on 10-19 with a close of $262.45 (+4.98%) and lost it on 10-22 at $250.38. The 10-23 close of $248.00 is below its 150-day average of $250.77, and the Trail buffer is 16.35%. The stop line is $201.04.
A signal that turns on and off within three days is the cost of moving-average signals
Moving-average crossover signals are designed to catch big trends after generating several small false signals.
“The problem with these crossover signals is the large number of whipsaw trades, often four or five small losers before the trend fully develops.” — Mike Carr, Technical Analysis of Stocks & Commodities
False signals multiply in markets that swing without direction (Capital.com). This week was such a market. The Nasdaq volatility index (VXN) was at 28.30, with the FOMC and Big Tech earnings ahead. The 10-19 signals for AMZN and GOOG came from a one-day rebound just before earnings. For the same reason, signals that turn on right after next week's earnings could be just as shaky. With only a 0.39% gap between QQQ's two moving averages, one strong day would turn the signal back on.
One proposed way to cut false signals is to build a margin around the reference line (Marco Alves, Traders' Magazine, 2009). The two-trading-day hold condition in the action criteria below borrows this idea and is a personal rule, not a system rule.
The same earnings day, a different starting line: Microsoft
MSFT kept its entry condition this week, with a Trail buffer of 25.17%. In its July results, Azure revenue grew 43% and the commercial bookings backlog grew 84% to $678 billion, and it guided next quarter's capex to more than $50 billion (Tom Tunguz, 2026-07). Microsoft is being tested on the same day having explained its spending with its backlog, while Alphabet is being tested having shown higher spending and a cash-flow deficit.
Position actions
| Ticker | Current status | Action this week | Next criterion |
|---|---|---|---|
| QQQ, XLK | Entry condition lost, Trail 22.34% and 20.15%, medium-term trend intact | Hold, stop adding | Trim part on a close below QQQ's 150-day average of $698.51 |
| GOOG | Entry condition lost, medium-term trend broken, Trail 8.54% | Consider trimming before earnings | Trim further below about $303 (Trail 5%) |
| AMZN | Entry condition lost, below 150-day average, Trail 16.35% | Wait | Whether it recovers the 150-day average of $250.77 after 10-29 earnings |
| MSFT | Entry condition kept, Trail 25.17% | Hold | Whether the entry condition holds after earnings |
- If a re-entry signal turns on again in earnings week, check whether it holds on two trading-day closes before acting. That is why AMZN and GOOG's signals, which turned on with a one-day rebound, turned off in three days this week.
- The fact that seven exits were concentrated in AI-linked names does not mean liquidating all AI holdings. Names more than 20% from the stop line and names within 10% should be handled differently. The closest this week is GOOG.
- Since W41, holding off on candidates with a Trail buffer below 4% has been right for three weeks in a row. Apply the same criterion to next week's re-entries.
You can see detailed charts and the signals for other tickers on the /signals dashboard.