If you have used momentum as part of your investment strategy over the past decade and found it disappointing, you are not imagining things. That is how Larry Swedroe opens his May 2026 analysis at Alpha Architect, and the sentence lands with the weight of an admission that the factor investing community has been slow to make in public.
The research at the centre of Swedroe's piece is Haim Mozes's study "Volatility Spikes and Momentum," published in the Spring 2026 issue of The Journal of Beta Investment Strategies. Mozes examined more than 30 years of market data — 1994 to 2024 — divided into three roughly equal decades, each bookmarked by a defining volatility event: the NASDAQ crash and September 11, the global financial crisis, and the COVID-19 pandemic. The central question Mozes pursued was whether the relationship between volatility spikes and momentum returns has changed over time, and whether that change can explain momentum's prolonged underperformance in the most recent decade.
The answer, across every metric Mozes tested, is yes.
Spikes Are More Frequent. They Reverse Faster.
In the most recent decade, 2014 to 2024, volatility spikes occurred far more frequently than in earlier periods. Using a moderate definition of a spike — the VIX reaching 1.5 times its three-month moving average — there were just seven spikes in the 2004–2013 period but 22 in the 2014–2024 period. That is a tripling of spike frequency in a single decade, and the implications for momentum are direct.
Mozes also found that when spikes did occur, they were resolved more quickly. Under that same 1.5x definition, spikes took an average of 54 trading days to reverse in earlier periods, compared to just 28 days in the most recent decade. The research suggests this faster recovery reflects greater market efficiency — markets are now quicker to process new information and reach a consensus, potentially driven by the rise of high-frequency trading, algorithmic strategies, and hedge fund activity.
For momentum investors, speed of recovery is not a feature — it is a problem. Momentum strategies depend on trends persisting. When fear spikes and then evaporates in under a month, the signal and the noise become indistinguishable.
The Stock Pattern: Down, Then Up — and Momentum Misses Both
The data show a consistent pattern in how stocks behave around volatility spikes: before the spike, the S&P 500 tends to fall meaningfully. In the 2014–2024 period, the average market return in the month before a moderate volatility spike was approximately -3.8%. For larger spikes, that prior-month decline was even steeper.
What follows the spike is the critical dynamic. Markets tend to recover — sometimes sharply — in the weeks after a VIX spike reverses. This down-then-up pattern creates a structural environment in which momentum strategies are positioned in exactly the wrong direction at exactly the wrong time. Momentum portfolios that shorted the prior losers (which are now the post-spike rebounders) and held the prior winners (which led the decline) take a double hit: losses on both legs.
A Broken Decade for Long/Short Momentum
The 2004–2013 period showed poor overall results, but that was largely due to a catastrophic -83.81% momentum return in 2009 alone — a classic momentum "crash" driven by the market's sharp recovery from the financial crisis. Excluding 2009, momentum would have averaged 7.17% annually during that decade. The 2014–2024 period, by contrast, saw consistently weak momentum across multiple years — exactly what the volatility spike hypothesis would predict given how frequent and fast-reversing spikes became.
This is an important distinction. The 2009 crash was a single, identifiable event — horrific in magnitude, but a one-year anomaly in an otherwise functional decade for momentum. The 2014–2024 period represents something structurally different: persistent, low-grade impairment spread across multiple years. Not a crash. A slow leak.
Why Is This Happening?
Mozes offers four structural explanations for the spike in volatility spike frequency, and Swedroe quotes them directly: "a) the increasingly rapid pace of technological change; b) the increasing polarization of the political process, which leads to more numerous disruptive legislative proposals and agency interpretations/rule making, c) the increasing ability of social media to amplify events, and d) the proliferation of highly leveraged investment vehicles and funds, which bring with them inevitable forced deleveraging episodes."
None of these forces are transient. Technological disruption is accelerating. Political polarization is structural. Social media's capacity to amplify fear is not diminishing. And leveraged vehicles — from volatility-targeting funds to options-heavy retail strategies — continue to proliferate. The conditions that produced the post-2014 momentum drought are not a weather system that will pass.
Consistency With Prior Research
Mozes's findings do not emerge in isolation. They are consistent with those of Andrew Berkin's 2021 study "When and Why Does Momentum Work — and Not Work?" — though Berkin's paper focused on single stock momentum while Mozes's focused on the overall market. The convergence of findings across different methodological frameworks strengthens the thesis: volatility spikes are not just correlated with momentum underperformance, they are causally implicated in it.
Swedroe's conclusion is unambiguous in its implications: "The structural environment that produced its strong historical returns may have changed in ways that are hard to reverse."
Key Takeaways for Advisors and Investors
Momentum's underperformance is not a factor drift — it has a mechanism.
The research provides a specific, testable explanation: more frequent VIX spikes mean more frequent momentum disruptions. Understanding the mechanism matters because it changes how advisors should frame client expectations around factor-based strategies.
Long/short momentum is the most exposed.
The short leg of a long/short momentum portfolio is particularly vulnerable during spike reversals, when yesterday's losers suddenly outperform. Long-only momentum implementations carry significantly less tail risk in a high-spike-frequency environment.
Volatility spike months should be treated as contaminated signals.
Be cautious about acting on momentum signals that coincide with a volatility spike — those signals may be noise rather than genuine trend information.
The post-spike recovery is real — but asymmetric.
Contrarian positioning after a VIX spike has historically been rewarded. Volatility spikes have historically been followed by meaningful market recoveries, though the exceptions — 2008, 2020 — are severe enough to demand humility.
Structural conditions are not self-correcting.
The four forces Mozes identifies — technology pace, political polarization, social media amplification, and leveraged vehicle proliferation — are all compounding. Advisors who build momentum allocations on historical return assumptions that predate 2014 are using a map drawn before the territory changed.
Factor diversification, not abandonment.
The research argues for recalibrating expectations and implementation, not for exiting momentum entirely. Combining momentum with value, quality, or low-volatility factors — which tend to perform better in the high-spike environments that hurt momentum — remains a coherent response.
The VIX was designed to measure fear. What Mozes's research reveals is that fear itself has changed — in frequency, in pattern, and in consequence for investors who built portfolios assuming that fear, when it came, would be rare and slow to leave.
Footnote:
1 Swedroe, Larry. "Why Momentum Investing Has Been Struggling—And What Volatility Has to Do With It." Alpha Architect, 8 May. 2026.
Copyright © AdvisorAnalyst
