Yesterday's Losers Lead as the Long Bond Hits 5.26%

• One-Day Momentum ran -0.67% at z=-2.4 — Thursday's losers were bought back and Thursday's winners sold, the day's most load-bearing style axis.
• Broadcom closed the session -6.15%, erasing $118.1 billion of market value on the VMware vCenter exploit.
• July retail sales census.gov fell -0.6% and the 30-year Treasury yield still rose 4.6 basis points to 5.257%.
• IGV fell -2.03%, the largest decline in the ETF book, against a +13.43% run over the prior 20 sessions.

August 14, 2026 · 16:22 ET

One-Day Momentum ran -0.67% at z=-2.4 — Thursday's most-punished names rose +2.27% as a group while Thursday's biggest gainers fell -2.11%. That spread carried 5.0% of the day's cross-sectional variance, a top-decile share for this factor against its own past year, and it did more work than anything on the calendar.

The Long Bond Rose and the Cross-Section Shrugged

July retail sales fell -0.6% against a +0.1% consensusBloombergCENSUS, and the long end went up anyway — the 30-year at 5.257%, up 4.6 basis points, the 10-year at 4.688%CNBC. Thursday's $25 billion auction had already cleared at 5.216%, the highest in 25 yearsBloomberg. None of it registered as a duration event in the cross-section: the Treasury (Duration) factor sits at z=+0.09 and Value at z=+0.65, both inert. Long-duration software took the sharpest single-day hit regardless, with IGV -2.03% interrupting a +13.43% 20-day run, while energy and the gold miners carried the other end on Iran-driven crude strengthBloomberg. The dollar index at 99.645, -0.32%, helpedCNBC.

Under a flat index — SPY -0.20% — the risk-seeking pockets were bid. Short Interest ran -0.33% at z=-2.0 with the barely-shorted bucket up +1.66%: the most-shorted names lagged again, the model's most persistent drag paying, and its 20-day z of -1.73 says that has been the pattern for weeks. Liquidity ran +0.55% at z=+2.0, driven entirely by its top bucket at +2.08% — the most heavily traded, most optionable names, where the crypto miners sat: WGMI +2.76% against BITO -0.76%, a bid in the equities and not the coin. The top 50 high-beta momentum names ran +2.36% against +0.05% for the rest of the universe, with the buying concentrated hard in the names that fit that profile most closely. This is the same complex that produced July's forced liquidation of Leopold Aschenbrenner's $45 billion Situational Awareness fund after a 40% drop in its AI-hardware bookWSJ.

Reversion at Two Speeds

The weekly lookback has been the one churning. Short-Term Momentum's completed 20-day return of -5.70% ranks in the bottom 0.2% of all 20-day windows since 2020, at z=-4.1, and today it added only -0.19%. One-Day Momentum's own three-month z is +1.30 — at the daily lookback the pattern since May has been continuation, and the factor's realized baseline at that horizon is weak enough that a single print is a diagnostic, not a regime. Today the churn reached down from the weekly clock to the daily one. The volatility configuration around it: implied index volatility at 14.28, down 31% from its July 29 high of 20.66, with implied single-stock volatility still at 36.43, the 75th percentile of its own history — the calm is in the index, not the names.

Where the Money Actually Moved

Broadcom closed the session -6.15%, erasing $118.1 billion, after reports that a critical VMware vCenter vulnerability is being actively exploited alongside institutional sellingBloomberg. Applied Materials fell -5.13% and $21.5 billion despite beating on revenue and guiding fourth-quarter sales above consensus, with the margin profile the sticking pointBloomberg. AMD went the other way, +6.49% and $51.2 billion added, almost none of it explained by the factor complex. The Semiconductors factor was positive through all of that, +0.58% at z=+0.96, because its top-exposure bucket at +2.37% is being carried by memory rather than large-cap logic — DRAM +0.70% on the session and +10.67% over five days, with SanDisk's pivot to long-term contracted supply the week's catalystMarketWatch and SK Hynix's $720 billion buildout the structural frame behind itCNBC.

The rest of the sweep was idiosyncratic in every direction. Reddit closed +12.55%, adding $4.0 billion, on its August 18 entry into the S&P 500 in place of AvalonBayCNBC. Nu Holdings rose +9.20% after its first billion-dollar quarterly profitCNBC. HeartFlow closed +35.79% on its Q2 report, then gave back -1.24% in late trading. Unusual Machines closed +25.21% as the White House announced 100% tariffs on imported drones, the same headline that took XAR +1.54% on 3.4× the typical session paceBloomberg. On the other side, Dillard's -8.77% and Viking Holdings -7.95% were the consumer casualties — but both moves are the stocks' own, not a discretionary factor being repriced.

That is the read worth carrying into next week: 81.0% of today's cross-sectional variance was stock-specific against a 66.3% one-year average, and the market factor accounted for 0.4% of it, an 11th-percentile session since 2020. Trailing-20-day realized pairwise correlation of 8.22%, the 3.8th percentile since 2020, has been sitting near these levels for weeks. A book hedged at the index level is not hedged at all in this configuration — the payoff is in what each name did on its own, and the only cross-sectional axis reliably taking money off one side and putting it on the other is the day-old one.

Economic Context

Advance Retail Sales, released at 8:30 AM ET, showed July sales down -0.6% to $763.6 billion against a +0.1% consensus, the sharpest monthly decline since May 2025, with core spending excluding autos and gas off -0.2%CENSUS. The preliminary Michigan Consumer data.sca.isr.umich.edu Survey, released at 10:00 AM ET, fell to 68.4 from 70.1 and below the 71.0 consensus, with year-ahead inflation expectations unchanged at 2.9%Bloomberg. The New York Fed Staff Nowcast, published at 12:45 PM ET, trimmed Q3 GDP bea.gov to 2.1% from 2.2%, citing the retail sales contraction as the driver.

Data compiled by FactorPulse AI; edited and verified by Jeff Klein. For informational purposes only. Does not constitute financial advice, an investment recommendation, or an offer to buy or sell any securities. Always consult a qualified financial professional before making investment decisions.

For more on factor construction methodology, see www.factorpulse.com/glossary.

← FactorPulse  ·  All Digests