We Asked a Top Momentum Manager Why 52-Week Highs Are a Buy Signal, Not a Warning

ELI5/TLDR

Travis Prentice runs a momentum shop. He says the S&P’s calm surface hides massive churn underneath — semis up 30%, software down 23, value and momentum suddenly friends, quality having a bad year. His pitch: this is what an AI-driven regime shift looks like, and the passive-flow concentration that built up over the last 25 years is now sitting on a dynamite stack with AI as the lit fuse. The headline finding from his paper: stocks near their 52-week high are actually a better predictor of future returns than the standard 12-month momentum signal — and counterintuitively, they have better downside capture, not worse. The signal works because everyone’s instinct is to refuse to buy at a high.

The Full Story

The market underneath the market

The S&P looks fine. The dispersion underneath it is the actual story. Prentice opens with a stat that captures the regime: the SOXX (semiconductors) up 30% year-to-date, IGV (software ETF) down 23%. Two adjacent corners of “tech,” one of them on fire and the other being burnt. Quality and growth as factors are getting blowtorched even as the index sits flat.

What’s breaking is the shape of the last 25 years — globalization, software-as-perfect-business, and a small handful of names doing all the heavy lifting. What’s replacing it: a broadening trade where Russell 2000 outperforms large cap, value and momentum start agreeing with each other (rare), and capital starts spreading out into materials, utilities, energy, industrials. Prentice frames this as two seismic shifts running in parallel:

AI as a capital-intensive technology that needs hardware, power, real estate, infrastructure — not just code
Deglobalization and nearshoring pulling manufacturing back, which lights up sectors that have been ignored for a generation

“Most of my 30 years has been rampant globalization and this virtual software-driven market. We’re just seeing areas of the market that have been kind of underloved coming back.”

The passive risk that’s hiding in plain sight

Prentice is a Mike Green fan and his “Risks Hiding in Plain Sight” paper extends Green’s argument. The headline number: the S&P 500 entered the year 40% in seven stocks and 35% in tech. He’s careful to say passive isn’t broken — but “Wall Street takes really good ideas and grows them so big they become less good ideas over time.”

The subtle damage: investors have quietly redefined risk. Risk used to mean loss of capital or failing to earn a return. Now it means tracking error — how different am I from the S&P. That redefinition is what makes passive feel safe right at the moment when its concentration is most extreme.

The mechanic Green and Prentice both point to: market-cap weighting is a flow-driven decision, not a fundamentals-driven one. Bigger companies get more dollars regardless of what they’re worth. As long as flows keep going one direction, the math holds. The minute they reverse, the same disproportionate weight that created the outperformance creates the disproportionate selling pressure. You don’t need much of a rebalance from passive to active to do real damage to a top-heavy index.

The change agent that turns this from a theoretical risk into a real one, in his view, is AI — specifically AI making the previously-perfect software business model suddenly capital-intensive and competitively vulnerable.

Why momentum looks like a chameleon right now

The most useful frame in this conversation: momentum is not “growth stocks.” Those two factors are mildly positively correlated, but momentum’s best premium actually shows up in periods when it’s not loaded on growth. Right now Prentice’s strategy is loaded on materials (gold, silver, copper), utilities, and energy — the “boring” stuff. The Iran spike in March added to it.

“A momentum strategy is a chameleon. It’s just going to reflect what’s working and move away from what’s not.”

He also makes a quiet but important point about the Mag 7. People assume a momentum strategy owns all of them. In reality, “in the last year, only two of the seven have actually outperformed.” Momentum picks up on that. It’s a much narrower trade than the casual observer thinks.

The recency-bias finding

The standard momentum signal in academia is “12-minus-1” — the last twelve months of return, dropping the most recent month (which has reversal). Prentice’s research, which they nicknamed Back to the Future, found that over the last 20 years a more recency-biased look-back actually beats the 12-minus-1, especially in US large cap. His read on why: trends form and end faster now. He sees more violent overreactions at the end of trends — “blowoff tops” — and attributes it to zero-DTE options, FOMO, and the general behavioral chaos of modern markets.

The 52-week high — the real headline

This is the meatiest piece. The intuition first, then the numbers.

The intuition: humans hate buying at highs. It feels like the move is over, the stock is “expensive,” you’re getting in late. So on average, stocks at their 52-week high are systematically under-bought relative to where they should be. The behavioral aversion creates the alpha.

The research: Prentice’s team tested three different ways to measure proximity to the 52-week high:

Position relative to 52-week high — simple percentage off the high
High-to-price — how far above the 52-week low
Range 52-week high — where in the high-to-low range the stock is sitting

The range version had the best Sharpe across most universes. And here’s the surprise that even Prentice flags as counterintuitive: 52-week-high signals had better downside capture than standard momentum. This is the opposite of what most people would expect — buying high should mean getting clobbered when the trend breaks. The data says otherwise.

When they ran nested sorts (sorting first on momentum then on 52-week high, and vice versa), they found the 52-week high explains most of the momentum premium on its own. If you just took the top quintile of range 52-week high, you beat 12-1 momentum on a risk-adjusted basis. But Prentice still combines the two — pure 52-week-high has lower upside capture. Combining gets you most of the Sharpe with better balance.

“The adage of buy high, sell higher seems to be accurate. We’re all trained on buy low, sell high, but the opposite is actually better.”

The signal fails in the same conditions standard momentum fails: abrupt leadership rotations, sharp reversals after a downward trend (the V-bottom recovery where the new leaders are yesterday’s losers).

Quality is broken — temporarily

Prentice uses the Fama-French definition of quality: top quintile of operating profitability. By that measure, quality is having a rough stretch and is now negatively correlated with momentum, which is unusual.

His diagnosis: quality is a backward-looking metric. It looks at companies that have been historically profitable. In a regime shift like the one AI is driving, the companies with the highest historical profitability — software — are precisely the ones whose moats are getting questioned. The 1990s parallel: high-quality companies underperformed for years through the internet build-out, then quality came back. He expects the same pattern. Profitability will inflect, but maybe in hardware and physical infrastructure, not in the SaaS roll-up that defined the last decade.

The framing he keeps coming back to is rate of change: don’t ask “is this a good company or a bad company,” ask “is this getting better or getting worse.” The second derivative matters more than the level.

Implementation, not signal, is where the alpha leaks out

Two implementation details that get less attention than they should:

When you rebalance matters more than how. Prentice ran a 12-1 momentum strategy in 2025 and varied only the rebalance calendar by one month. The difference between the January/March schedule and the February/May schedule was one thousand basis points of return. Same signal, same universe, different month. If you happened to measure at the depths of the April tariff tantrum versus a month later, you got radically different portfolios. His response: rebalance daily and incrementally. Don’t wait for a calendar date. Move when the signal moves.

Combine fundamental momentum with price momentum. Pure price-chasing works empirically, but if the price move is backed by analyst revisions, earnings surprise, and post-earnings drift, the trend is more durable and the reversal risk is lower. They use both.

Key Takeaways

Index calm is masking factor chaos — semis +30%, software -23%, quality and growth deeply underperforming
The S&P entered the year 40% in seven stocks; passive flows, not fundamentals, drove most of that concentration
Investors have redefined risk as tracking error, which makes the most concentrated index in modern history feel like the safe option
Momentum is a chameleon, not a tech bet — currently loaded on materials, utilities, and energy
Recency-biased look-back periods (shorter than 12-1) have outperformed in US large cap over the last 20 years
Stocks near their 52-week high predict future returns better than standard 12-1 momentum, with better downside capture
The “range 52-week high” measure (where in the year’s range a stock sits) has the highest single-factor Sharpe ratio
The signal works precisely because investors instinctively refuse to buy at highs — the behavioral aversion is the alpha
Combine 52-week high with 12-1 momentum to balance Sharpe with upside capture
Rebalance timing alone created a 1000bps gap in 2025 — daily incremental rebalancing beats calendar dates
Pair price momentum with fundamental momentum (revisions, surprise, post-earnings drift) to reduce reversal risk
AI is the regime shift quality investors are getting caught by — software’s perfect-business model is the most exposed

Claude’s Take

This is a substantive, no-nonsense interview with a practitioner who actually publishes his research and has the receipts. The 52-week-high finding is the keeper — it’s not new (George and Hwang published it in 2004) but Prentice’s nested-sort work showing it absorbs most of the standard momentum premium is a genuinely useful update. The “better downside capture” angle is the part that should land, because it inverts the intuition completely. Most people would assume buying near highs means catching tops; the data says the opposite.

The passive-risk discussion is well-traveled territory if you’ve read Mike Green, but Prentice does add the tracking-error-as-redefined-risk frame, which is sharp. His honesty about not knowing how passive affects factor returns (“it depends on the day”) is refreshing in a field where everyone wants a clean answer.

Where the conversation gets thinner: the AI-as-change-agent thesis is more vibes than analysis. He’s right that AI shifts software’s competitive position, but the leap from “AI disrupts software margins” to “passive flows reverse and crater the index” needs more connective tissue than he provides. It might happen. The argument as stated is mostly assertion.

Score: 7. Strong on the 52-week-high research and the rebalance-timing experiment. Useful frame on the broadening trade. The macro speculation is the weakest part. Worth the listen for any factor-curious investor; the practical takeaways (combine signals, rebalance incrementally, watch the second derivative) are immediately actionable.