How each of the 9 stock groups (and 19 subgroups) behaved in each of the 7 market regimes. Groups are equal-weighted baskets of their members; the benchmark is the equal-weighted average of all 2,000 stocks, so every number here is relative to the typical stock that day.
Pre-profit growth and cash burners at one end, low-beta value and yield & real assets at the other. The regimes decide which end wins, and the two ends almost always move in opposite directions.
When yields rise fast on a steep curve, pre-profit growth trails the typical stock by about 6% a month and cash burners by 5%, while low-beta value leads by 3.5% and yield stocks and banks by about 2%. It holds after removing beta (t ≈ 2 to 5).
In growth-led rebounds pre-profit growth leads by about 5% a month; in the AI melt-up by 4.7%, with high-beta hypergrowth at +7.4%. Yield & real assets lag by about 3% in both. Beta explains part of it, not all.
+0.8% a month over all days and positive in 5 of 7 regimes; it lags clearly only in reflation. Some of that is hindsight: groups are labelled on September 2026 fundamentals, so today's quality growers are partly yesterday's winners.
Late-cycle calm is the quiet mirror of the melt-up: yield & real assets +1.7% a month and low-beta value +1.1%, while pre-profit and cash burners lag by about 3%.
Crash / panic is 45 days in two spells (COVID). Quality growth held up and banks fell hardest, but no significance test is possible and the returns are extreme. Treat it as a single episode.
Each cell is the group's average over the regime's days. Bold values with a dot are more than two standard errors from zero, with errors clustered by regime spell because days inside one spell are not independent.
Same day: the return on the day carrying the regime label, which uses that day's close (descriptive). Next day: the return on the following day, when the regime is already known. Most large effects keep their sign in the next-day view because regimes persist, but they shrink.
Groups ranked within each regime by excess over the typical stock, % per month.
Log excess return of a group over the equal-weight universe, summed day by day, with each day shaded by its regime. Rising stretches are where the group earned its lead.