27 Weeks of Public AI Stock Picks: What the Numbers Show
QSignals Research · Performance Review · November 1, 2026 Coverage: 27 weekly score snapshots, March 23 – October 19, 2026. Benchmark: Invesco QQQ (Nasdaq-100).
Key Takeaways
- Long-term scores led, with a heavy caveat. The 10 highest long-term scores returned +28.3% on average over six months versus +20.9% for QQQ. Micron accounts for most of that edge; excluding it, the top 10 averaged about +15.7% and trailed the index.
- Mid-term is the most dependable result. Across 24 snapshots the one-month top 10 returned +7.3% versus +3.8% for QQQ and +2.4% for the bottom 10, despite a sharp miss during the June chip selloff.
- Short-term is not working yet. The one-week top 10 rose in only 48% of cases and beat QQQ by just 0.2 points on average. We're re-testing those signals against the saved history.
- The backdrop shaped everything. An AI rally in April and May, a semiconductor-led selloff in June, and a Fed that has turned more hawkish since the summer explain most of the swings in this record.
Market Backdrop
The first half of the tracking window was one of the strongest stretches for AI-linked equities in years. Morningstar estimated that a broad global AI index gained roughly 45% across April and May. Memory was at the center of it: Micron had said its high-bandwidth memory (HBM) supply was sold out through 2026, and tight supply let memory makers raise prices into surging AI demand.
That momentum broke in June. Tech selling resumed in early June, then accelerated in the week of June 22, when semiconductor shares fell sharply in a single session and the Nasdaq-100 dropped around 3% on the day. By late June the Nasdaq was on pace for its worst month since March 2025. Commentary at the time pointed to higher-for-longer rate expectations, stretched valuations and growing doubts about the return on AI capital spending.
Markets recovered through the late summer and early autumn. A soft September jobs report cut the odds of an October rate hike and helped the Nasdaq-100 to a record high in October. On October 28 the Federal Reserve held its target range at 3.75%–4.00%, but two of the twelve voters dissented in favor of a hike, and markets continue to price further tightening. Rates are now the main macro risk for richly valued AI stocks.
Results Across Horizons
Show data
| Horizon | Snapshots | Top 10 | QQQ | Bottom 10 | Top-10 hit rate |
|---|---|---|---|---|---|
| 1 week | 27 | +1.0% | +0.9% | +0.2% | 47.8% |
| 1 month | 24 | +7.3% | +3.8% | +2.4% | 54.6% |
| 6 months | 6 | +28.3% | +20.9% | +2.8% | 71.7% |
The pattern across horizons is consistent with how the scores are built. Longer horizons lean more on business quality and trend persistence, which carried through the spring rally. One-week returns are dominated by news flow and market noise, and that's where the model has the least edge.
Long-Term (6 Months): Strong, but Concentrated
Show data
| Week of | Top 10 | QQQ | Bottom 10 | Top-10 hit rate |
|---|---|---|---|---|
| 03-23 | +38.8% | +25.4% | +2.5% | 80% |
| 03-30 | +63.7% | +30.8% | +33.8% | 80% |
| 04-06 | +40.0% | +28.8% | +17.3% | 80% |
| 04-13 | +19.0% | +21.9% | +13.1% | 80% |
| 04-20 | +11.6% | +11.0% | -27.4% | 70% |
| 04-27 | -3.2% | +7.5% | -22.4% | 40% |
Six snapshots are old enough to have full six-month results. The top 10 beat the bottom 10 in all six, by an average of 25.5 points, and beat QQQ in four of six.
The attribution matters more than the headline:
- Micron drove the result. It ranked first or second on long-term score in every one of these snapshots and returned between +73% and +206% over the following six months, riding the HBM cycle described above. Removing Micron, the top 10 averaged roughly +15.7%, about five points behind QQQ.
- The edge faded through April. The March 30 snapshot beat QQQ by 33 points; the April 27 snapshot trailed it by 11. Later snapshots captured less of the rally and more of the June drawdown.
- The sample is small and overlapping. Consecutive weekly snapshots share most of their holdings, so these six data points behave more like one or two independent periods.
The honest read: the long-term scores identified the year's most important AI winner early and kept it at the top. That is a real signal, but one stock is not a track record. The next few months, as summer snapshots reach the six-month mark, will be a fairer test.
Mid-Term (1 Month): Through a Full Cycle
Show data
| Week of | Top 10 | QQQ | Bottom 10 | Top-10 hit rate |
|---|---|---|---|---|
| 03-23 | +28.4% | +10.8% | +12.3% | 70% |
| 03-30 | +39.3% | +17.5% | +22.5% | 90% |
| 04-06 | +48.3% | +18.5% | +30.4% | 100% |
| 04-13 | +34.8% | +16.8% | +37.9% | 90% |
| 04-20 | +16.2% | +10.2% | -1.1% | 80% |
| 04-27 | +21.1% | +10.1% | +23.9% | 90% |
| 05-04 | +23.8% | +10.2% | +2.6% | 90% |
| 05-11 | +0.1% | -2.5% | -6.9% | 30% |
| 05-18 | +19.6% | +2.1% | +1.8% | 70% |
| 05-25 | +3.8% | -1.0% | -10.4% | 50% |
| 06-01 | -2.5% | -1.6% | -3.8% | 50% |
| 06-08 | -12.1% | -0.3% | -6.7% | 40% |
| 06-15 | -22.1% | -3.1% | -1.4% | 0% |
| 06-22 | -17.4% | -5.2% | -3.6% | 0% |
| 06-29 | -16.7% | -1.5% | +0.1% | 10% |
| 07-06 | -4.9% | -0.7% | -1.2% | 20% |
| 07-13 | -1.7% | -0.9% | -3.7% | 40% |
| 07-20 | +3.6% | +0.9% | +2.3% | 40% |
| 07-27 | +1.3% | +3.3% | +3.4% | 40% |
| 08-03 | +0.1% | +2.5% | +1.1% | 40% |
| 09-07 | +6.3% | +5.4% | +0.4% | 90% |
| 09-14 | +6.9% | +3.2% | -12.8% | 90% |
| 09-21 | +0.9% | -1.0% | -21.6% | 40% |
| 09-28 | -2.1% | -1.6% | -8.3% | 50% |
The one-month record covers a complete up-and-down cycle, which makes it the most informative of the three:
- March to May: the top 10 beat QQQ in all ten snapshots, often by 10 points or more, as AI hardware and infrastructure names led the rally.
- June: the worst stretch in the record. Snapshots from June 8 through June 29 lost between 12% and 22% over the following month, against declines of 0% to 5% for QQQ. The June 15 top 10 was concentrated in semiconductor and AI-infrastructure names (Credo, Micron, Arm, AMD and TSMC among them), exactly the group hit hardest in the June selloff. Momentum-led scores were most exposed when momentum reversed.
- September onward: steadier. The top 10 beat the bottom 10 in all four September snapshots, by as much as 22 points, while broadly tracking QQQ.
Even including June, the average one-month top 10 returned +7.3% against +3.8% for QQQ. The lesson from June is about concentration risk, which we come back to below.
Short-Term (1 Week): Not There Yet
Show data
| Week of | Top 10 minus bottom 10 |
|---|---|
| 03-23 | +3.6 pts |
| 03-30 | -6.2 pts |
| 04-06 | +4.4 pts |
| 04-13 | -2.5 pts |
| 04-20 | -5.9 pts |
| 04-27 | +0.1 pts |
| 05-04 | +20.1 pts |
| 05-11 | -0.4 pts |
| 05-18 | -1.8 pts |
| 05-25 | +0.8 pts |
| 06-01 | -4.3 pts |
| 06-08 | +6.5 pts |
| 06-15 | +11.0 pts |
| 06-22 | -0.2 pts |
| 06-29 | -1.9 pts |
| 07-06 | -4.2 pts |
| 07-13 | +6.6 pts |
| 07-20 | -3.1 pts |
| 07-27 | -6.7 pts |
| 08-03 | -7.1 pts |
| 09-07 | -4.8 pts |
| 09-14 | +4.1 pts |
| 09-21 | +4.6 pts |
| 09-28 | +0.6 pts |
| 10-05 | -2.7 pts |
| 10-12 | +5.3 pts |
| 10-19 | +5.7 pts |
The one-week top 10 beat the bottom 10 in 13 of 27 weeks, roughly a coin flip, and its average lead over QQQ was 0.2 points. Weekly results swing in both directions by double digits, which is typical for a universe heavy in small and mid-cap AI names.
We're treating this as the model's clearest weakness. The saved snapshots let us test which short-term inputs actually lined up with the next week's returns before changing anything, and any change will be visible in the same public record.
Market Regime: A Lagging Read
Show data
| Week of | Regime |
|---|---|
| 03-23 | bearish |
| 03-30 | bearish |
| 04-06 | bearish |
| 04-13 | bullish |
| 04-20 | neutral |
| 04-27 | neutral |
| 05-04 | neutral |
| 05-11 | neutral |
| 05-18 | neutral |
| 05-25 | neutral |
| 06-01 | neutral |
| 06-08 | bullish |
| 06-15 | bullish |
| 06-22 | bullish |
| 06-29 | bullish |
| 07-06 | bullish |
| 07-13 | bullish |
| 07-20 | bullish |
| 07-27 | neutral |
| 08-03 | bullish |
| 09-07 | bullish |
| 09-14 | bullish |
| 09-21 | bullish |
| 09-28 | bullish |
| 10-05 | bullish |
| 10-12 | bullish |
| 10-19 | neutral |
QSignals labels each week bullish, neutral or bearish from QQQ's trend and momentum, and adjusts mid-term scores accordingly. The record shows the expected weakness of a trend-based gauge: it was still reading bullish through mid- and late June, while semiconductors were selling off, and only moved to neutral in late July. Regime signals built on trailing price data confirm turns; they rarely call them. That is one reason the June mid-term miss was as large as it was.
Sector View
Show data
| Sector | Avg 1-month return |
|---|---|
| AI Tech | +6.7% |
| AI Growth | +6.7% |
| AI ETFs | +2.8% |
| Robotics | +1.4% |
| Energy | -1.2% |
| Space | -2.2% |
Across every stock and every snapshot, AI large caps and AI growth names returned about +6.7% per month on average, well ahead of robotics, energy and space. Energy names, which include the AI data center power trade, averaged slightly negative, and space names were the weakest group over the period. The model's best results came when its top scores sat in the leading AI hardware group, and its worst came when that same group sold off.
Risks and Limitations
- Concentration. Top-10 lists have leaned heavily toward semiconductors and AI infrastructure. That helped in the spring and hurt in June. A single-sector selloff can move the whole list together.
- Small samples. Six long-term and 24 mid-term snapshots are not enough to separate skill from a favorable market. Overlapping snapshots make the effective sample smaller still.
- Missing weeks. No snapshots were recorded for August 10 through August 31 because the weekly job didn't run. Those weeks are excluded, not filled in.
- What the numbers leave out. Returns are simple equal-weighted averages before trading costs, taxes and slippage. They are not the results of a portfolio anyone held.
Warning
A correction. Until this week, the public track record page showed only 19 of the 27 snapshots. A database query was capped at 1,000 rows, so the oldest weeks, including every week old enough to have six-month results, were silently dropped. The bug is fixed and the page now shows the full history. Every number in this review uses all of it.
What We're Watching
- The rate path. Two hawkish dissents in October and a market pricing further hikes leave long-duration AI growth stocks exposed if inflation data stays firm.
- AI capital spending. The June selloff showed how quickly sentiment can turn on doubts about the payoff from AI infrastructure spending. Hyperscaler capex guidance is the variable to watch.
- Memory pricing. Micron's run rested on tight HBM supply and rising prices. Any sign of easing supply would test the stock that has done the most for our long-term record.
- A wider universe. Coverage expanded from 55 to 78 stocks this week, adding Oracle, Arista, Vertiv, CoreWeave, ASML and Cloudflare among others. New names enter the record from their first Monday snapshot, which should reduce reliance on any single winner.
Methodology
Every Monday, QSignals scores each covered stock from 0 to 100 on three horizons and saves the scores with that day's price. Returns are filled in after one week, one month and six months. For each snapshot and horizon we compare the average return of the 10 highest-scored and 10 lowest-scored stocks with QQQ over the same window; hit rate is the share of top-10 picks with a positive return. Results above are simple averages across snapshots. Nothing is backtested: every score was recorded before the returns it is measured against. Scores appear on the public track record page with a one-week delay.
Sources
Market context is drawn from the following reporting. Track record figures come from QSignals' own score snapshots.
- Morningstar, After the Latest AI Stock Rally, Investors Face a Reality Check
- Fortune, Nasdaq selloff, June 9, 2026
- Trading Economics, US Tech Stocks Sell Off
- Barron's via Tiger Brokers, The 'Tech Wreck' Forces Wall Street to Ask How Much Upside Is Left
- The Motley Fool, Why Micron Stock Skyrocketed Last Year
- Benzinga, Nasdaq 100 Hits Record High As Fed Hike Bets Cool
- CNBC, Fed rate decision, October 28, 2026
- Seeking Alpha, Fed's Interest Rate Decision: October 28, 2026
Info
See every weekly snapshot and its returns at qsignals.ai/performance. This review is for information only and is not investment advice. Past performance does not guarantee future results. QSignals is not a registered investment adviser.