Weight Shift vs. Heat — Zorro's Rotation Model Has Two Axes
Technology is rated "hot" and still slated to lose weight. Financials are rated "cool" and still gaining it. Here's why that's not a contradiction.
Open Zorro's sector rotation model this week and one pairing looks like a contradiction. Technology carries a "hot" heat rating — the model's label for overheated — and its weight is still projected to shrink from 31.5% of the S&P 500 today to 29.0% over the next three to six months, a shift of -2.5 points. Financials sit at the opposite pole: rated "cool," the model's word for undervalued, while its weight is projected to grow from 13.2% to 14.3%, a +1.1 point shift. One sector is expensive and shrinking. One is cheap and growing. Neither pairing is unusual, because the rotation model is actually answering two separate questions, not one.
Weight shift measures direction, not price
The "Rotating In" and "Rotating Out" chips on the rotation page come from a single number: current weight minus projected weight, flipped to show where the model expects the S&P 500's sector composition to sit three to six months out. It's a forecast of market-cap share, built the same way the index itself is built — bigger constituent gains and losses move a sector's slice of the pie, nothing about cheapness enters into that number directly. A sector can gain projected weight purely because its biggest names are expected to keep compounding faster than the rest of the index, with no judgment at all about whether that pace is sustainable or already priced in.
That's exactly the gap the model's own Rotation Rider strategy exploits on the buy side: it mechanically picks the two sectors with the most positive weight shift, full stop. Weight shift alone can't tell that strategy, or a reader, whether it's buying a sector cheap or paying up for one that's already extended.
Heat measures whether you'd be paying up
Heat is a different kind of output — a four-bucket read (overheated, running hot, fairly valued, undervalued) on whether a sector's current price looks stretched or discounted relative to its own fundamentals, not a weight forecast at all. Energy is the other "hot" sector on the board right now, up roughly 47% year-to-date against a model z-score north of 2 — a sector that's run far enough, far enough fast, to land in the same overheated bucket as Technology even though the two have almost nothing else in common. Consumer Discretionary sits at the other extreme: down roughly 5% on the year, rated "cool," with a projected weight gain of 0.8 points — the model's mean-reversion case, cheap and (modestly) rotating in.
Four sectors, four stories:
| Sector | Weight shift | YTD return | Heat → projected heat |
|---|---|---|---|
| Technology (XLK) | -2.5 pts | +36.0% | hot → warm |
| Energy (XLE) | -0.3 pts | +47.2% | hot → warm |
| Financials (XLF) | +1.1 pts | -0.2% | cool → neutral |
| Consumer Discretionary (XLY) | +0.8 pts | -5.1% | cool → neutral |
Both hot sectors are projected to lose a little weight and cool a notch — consistent with "extended, due for a pause." Both cool sectors are projected to gain weight and warm a notch — consistent with "oversold, due for a bounce." The two axes agree here, which is the ordinary case. They don't have to.
Where the two axes split
The split case is the more informative one, because it's where reading only one number would mislead you. A sector can be "rotating in" (gaining projected weight) while still rated hot, if its earnings momentum is strong enough to keep pulling index share even at a stretched valuation — the model would be saying "keep growing into the index, but don't assume the multiple holds." A sector can be "rotating out" while rated cool, if the model expects its weight to keep shrinking for structural reasons — client losses, secular decline — even though it already looks statistically cheap. Cheap and shrinking is a value trap; neither number alone catches it, but the two read together do.
This is the same two-variables-not-one problem Zorro's ranked Value Pick strip on the screener solves a different way, by scoring a stock on multiple factors and taking a percentile rank instead of trusting any single metric. A screener row that's cheap on P/E alone can still be a trap if margins are collapsing; a sector that's gaining weight alone can still be a trap if it's gaining weight at the top of its valuation range. Both tools exist because a single number is rarely the whole story, and both are built to force the second question rather than let the first one stand in for it.
Reading the two signals together
None of this is a timing signal or a forecast of what either sector does next — the model's own 3-to-6-month weight projection is itself just an estimate, and heat buckets update as prices move, sometimes within the same week. What the split tells you is which question a given sector is actually answering right now. "Hot and rotating out" is a sector the model thinks has run ahead of its fundamentals and is giving back index share because of it. "Cool and rotating in" is a sector the model thinks is cheap for a reason that's already fading. Reading weight shift without heat tells you direction with no sense of valuation; reading heat without weight shift tells you valuation with no sense of where capital is actually moving. The rotation page puts both numbers next to each other for exactly that reason — neither one is the model's answer by itself.