The trend in net new highs, one day later

First, credit where it’s due — this is a genuinely good instinct. Net new highs (new 52-week highs minus new lows) is one of the honest ways to take the market’s pulse: it asks how many stocks are actually participating, not just where the cap-weighted index closed. Smoothing it into a trend and asking whether that trend is rising or falling is exactly the right kind of question. And the gap the note describes is real — I found it too. In fact, measured the way the claim is measured, my numbers are if anything more dramatic. So the observation isn’t wrong. I just wanted to chase down where the gap actually comes from.

Here’s the whole thing in one sentence: the return gap is coincident, not predictive. On days the net-new-highs trend is rising, the market is rising — that’s the same fact said twice, because a rising trend in participation and a rising tape are the same afternoon. The test that matters is whether the reading tells you anything about tomorrow. So I lagged the signal by a single day — read it at the close, act the next morning — and looked again.

01The gap is real — and it’s the market and the signal moving together

Start with the claim on its own terms. I built the trend the natural way: a moving average of net new highs (I tried 10-, 20- and 50-day windows — same story throughout), “rising” when today’s average is above yesterday’s. Then I split every trading day since 1993 by that state and annualized the returns. On rising-trend days the market compounds at roughly +59% a year; on falling-trend days it loses about 39%. That’s not three-to-one — it’s a sign flip, an even starker version of what the note describes.

But look at what’s being measured: the return on the same day as the reading. Net new highs expand precisely on the days stocks rise, because a stock making a new high and a stock going up are the same event. Sorting days into “up” and “down” buckets and then discovering the up bucket went up isn’t a forecast — it’s a mirror. The honest question is whether today’s reading pays you tomorrow. So I shifted the signal forward one day and kept everything else identical.

02Lagged a day, the return edge is gone

This is the crux. Once the signal is something you could act on — read at the close, position taken the next morning — rising and falling trends are followed by almost identical returns. Extend the horizon and it only gets flatter: a month, a quarter, a year out, a rising net-new-highs trend and a falling one lead to the same place, and both sit right on top of the everyday baseline. The direction of the trend, once you can only know it after the fact, carries no information about where prices go next.

03What does survive: a little more turbulence

Not everything washes out, and this is the part of the note worth keeping. The volatility half of the claim partly holds up even after lagging: when the net-new-highs trend is falling, the market’s forward volatility really is higher — about 21% annualized versus 16% when the trend is rising. That’s roughly 1.3×, more modest than the “more than twice” in the original, but it’s a real, persistent effect. Falling participation goes with choppier tape.

The catch is what kind of signal that is. It tells you about risk, not direction — useful for how much to size, not for whether to be in or out. And in my own testing, volatility-scaling overlays like that have been hard to turn into an edge: they mostly help by getting out of the way, and tie the simpler version at best. So it’s a true observation with a narrow, honest use: a turbulence gauge, not a timing switch.

04So what do you actually do?

Keep watching net new highs — as a read on how healthy today’s advance is, it’s excellent, and a rolling-over trend is a fair reason to expect a choppier tape and to size a touch more carefully. What the history won’t support is treating the trend’s direction as a return forecast: the eye-popping same-day gap is the signal and the market moving in lockstep, and it doesn’t carry into the days you could actually trade. If you want a participation-based rule that has protected capital, the plain trend of price itself — the index against its own 200-day line — has done that job in test after test, while breadth gates layered on top tend only to subtract. Read net new highs as a thermometer, not a stopwatch.

None of this makes the original point wrong, and that’s the good part. It reached for the right variable — participation, not just price — and framed it as a trend, which is how you should think about breadth. Chasing that same variable one step further just adds a coda: the dramatic return gap is coincident, it doesn’t survive the one-day lag that makes a signal real, and the piece that does survive is a modest volatility tell, not a timing edge. Net new highs are genuinely worth watching right now. They’re describing the room you’re in — not the door out of it.