Forecast for November 2026
How the model ranks the sectors for the month ahead
Each month a model ranks the eleven US market sectors by how it expects them to do against the S&P 500 over the coming month. Every forecast is kept and scored against what happened.
Status: not shown to work. Over 178 past months the model's ranking has not matched outcomes better than chance (rank correlation -0.002, where 0 is chance and 1 is perfect). Read the ranking below as an experiment in progress, not as a signal. See the track record.
The ranking
Made at the close of 30 October 2026, for the month that follows. The model forecasts each sector fund's return against the S&P 500 fund; the figure shown is that forecast against the average of the eleven, because only the order is scored.
| Rank | Sector | Fund | Against the average sector | |
|---|---|---|---|---|
| 1 | Energy | XLE | +0.72% | |
| 2 | Technology | XLK | +0.42% | |
| 3 | Communication services | XLC | +0.20% | |
| 4 | Industrials | XLI | +0.08% | |
| 5 | Consumer discretionary | XLY | -0.01% | |
| 6 | Financials | XLF | -0.02% | |
| 7 | Consumer staples | XLP | -0.10% | |
| 8 | Materials | XLB | -0.12% | |
| 9 | Real estate | XLRE | -0.24% | |
| 10 | Utilities | XLU | -0.27% | |
| 11 | Health care | XLV | -0.67% |
The forecasts are small because the model has found little to go on: the spread between its highest and lowest forecast is 1.39%, against a typical monthly gap between sectors of several percent.
How it has done
- Months scored
- 178, from December 2008 to September 2023
- Rank correlation
- -0.002 (interval -0.067 to +0.063). Zero is chance. About 0.05 or more can be told from zero with this much history.
- Momentum, for comparison
- +0.011. The model has to beat this simple rule to count as working.
- Top three minus bottom three
- -0.14% a month (interval -0.80% to +0.48%)
- Held back
- The 36 months from October 2023 are not scored until the model is frozen, then read once.
- Tries so far
- 2 model configurations have been tried. Intervals are widened to 97.5% to allow for that.
What goes in
321 inputs in 18 groups, each used only as it was known on the day: sector and market prices, economic releases as first published, what the companies in each sector reported, insider trading, short interest and more. See sources and method.