Japan's Momentum Trade Is Cracking. The Value Rotation Is Real.

Jefferies has been telling clients to shift into the Japanese stocks the AI momentum trade left behind. That call came mid-July. Three weeks later, the Nikkei 225 plunged more than 2,000 yen in a single morning session as AI and semiconductor names got hammered. Kioxia, the memory chipmaker that rallied more than 600% since January, lost 16% in one day - and 44% for the month.

The competitor headline asks whether Japan value stocks are about to gain. The data says the rotation is already underway. The question is whether it's a mechanical position-shift that reverses as quickly as it started, or something structural enough to act on.

Let the numbers lead.

The momentum trade is unwinding, not pausing

Japan's equity market has been one of the most concentrated trades in the world. Goldman Sachs estimated that hedge funds' total Japan allocation sits at the 99th percentile of the past five years, measured as a share of global portfolios. Foreign buyers accumulated roughly ¥14.8 trillion in Japanese equities since April 2025's "Liberation Day". Retail margin-buying balances are 35% above their July 2024 levels and near a five-year high.

That matters because crowded positions don't just decline - they cascade. The 2024 summer selloff showed how: TOPIX fell 24% from its record in late July as a yen move triggered stop-losses, forced liquidations, and risk-parity fund reductions. Goldman's warning in July 2026 was that positioning is even more extreme now, with the AI-linked exporters, banks, and cyclical industrial stocks at the center of it.

Kioxia is the cautionary tale. A 600% rally built on high-bandwidth memory demand for AI data centers. Then July 17 hit: the stock plunged 16% in a single session as broader concerns about AI infrastructure spending and semiconductor valuations spilled across Tokyo. Advantest and SoftBank each lost around 9% the same day. This wasn't a Kioxia problem - it was a concentrated-position problem.

The factor question is simple: when momentum deteriorates this sharply and positions this crowded begin to unwind, where does the capital go?

TOPIX vs. Nikkei: the rotation has a measurable signal

The NT ratio - the Nikkei 225 divided by the TOPIX, which shows whether big-name momentum stocks are pulling ahead or trailing the broader market - fell from the 18x range in late June to the 17x range in early July. That's not a rounding error. It's the structural fingerprint of money flowing out of concentrated Nikkei-weighted names and into the broader, more value-oriented TOPIX universe.

On July 6, the Nikkei edged down while TOPIX hit an intraday record high. Advancing issues on the Prime Market reached 73% of the total. Fast Retailing hit a record high. Toyota traded firmly. Meanwhile, AI and semiconductor names like Tokyo Electron, SoftBank, Ibiden, and Taiyo Yuden posted losses. A T&D Asset Management strategist noted that "sectors that had been bought over the past three months are undergoing a correction, while stocks that had been sold are being bought back."

The NT ratio decline is the kind of structural signal this framework looks for first. It doesn't tell you which individual stocks to buy. But it tells you that the market's center of gravity is shifting - from the narrow set of high-momentum, high-contribution AI-linked names toward a broader base of companies that have been lagging.

Buybacks are the floor value investors are standing on

Jefferies briefed investors in mid-July that Japanese share buybacks are on pace for a record ¥20 trillion this fiscal year. Through May alone, announced buybacks reached ¥16.2 trillion - up 34% year-over-year and already nearing the total for all of 2025. Sony and Hitachi led the charge.

This is a structural tailwind for value-oriented names, not a narrative garnish. Buybacks directly reduce the share count, which lifts per-share earnings and earnings-per-share growth even when underlying revenue is flat. They also signal that management believes the stock is trading below intrinsic value - which tends to cluster in companies that haven't been the momentum darlings.

The unwinding of cross-shareholdings (long-held institutional cross-ownership that has historically propped up lagging names) combined with the stewardship code's push for capital efficiency is feeding this machine. Companies that were sitting on cash and dormant cross-shareholdings are now repurchasing shares. That creates a mechanical bid under stocks that the momentum trade bypassed.

The Jefferies call: quality, not just value

Jefferies didn't just say "buy cheap." The recommendation was specifically for high-quality stocks left behind by the AI momentum trade - what they described as "low-stress" names. That distinction matters.

There are cheap Japanese stocks that are cheap for good reasons. The framework doesn't grade them equally just because they have low P/E ratios. The barbell approach that works in uncertain regimes pairs quality growth with durable cash-flow businesses - not with value traps that are deteriorating on every factor.

The rotation from Nikkei-heavy to TOPIX-broad gives investors a comparison set to work with. Value in the TOPIX universe includes transportation equipment, domestic demand sectors, and companies benefiting from yen depreciation - names like Toyota that showed firmness while semiconductor stocks stumbled. These aren't passive holdings. They carry specific factor profiles: earnings exposure to the weak yen, governance-improved capital returns via buybacks, and momentum that's just beginning to build rather than peak.

What would break the thesis

The 2024 selloff taught one lesson: crowding cuts both ways. If the yen strengthens rapidly, exporter earnings take a hit and the value rotation could reverse as fast as it started. Goldman's FX team expects the yen to weaken further - USD/JPY at 162 in three months - but a geopolitical shock or faster-than-expected BOJ tightening could change the trajectory.

There's also the question of whether this rotation is mechanical or fundamental. If AI spending concerns ease and the semiconductor rally resumes, capital could snap back into the momentum trade, leaving value names behind again. The 44% monthly decline in Kioxia doesn't prove the AI thesis is broken - it proves that valuations based on peak-cycle expectations are vulnerable to sentiment shifts.

The framework doesn't predict these outcomes. It ranks stocks relative to peers and adjusts as the factor stack changes. If value factor grades are improving while momentum factor grades are deteriorating, the process tilts accordingly.

What to watch

The rotation is real, but it's early. The NT ratio at the 17x range is a directional signal, not a finish line. Buybacks on pace for ¥20 trillion are structural support, not a guarantee. The hedge fund positioning at the 99th percentile means this market can still move violently.

The process says: check the factor grades of the individual names before you buy into the rotation narrative. A value stock in the TOPIX universe that's also showing improving growth grades and strong profitability is a different portfolio role than one that's cheap because earnings are declining. One belongs in the growth sleeve with a valuation discount. The other belongs on a watchlist with a specific catalyst trigger.

Volatility is usually a signal that the market can't figure out what comes next. The response isn't more conviction - it's more structure. Japan's momentum trade is showing cracks. The value rotation has a measurable signal behind it. But every stock in that rotation still needs its own factor read before it earns a place in the portfolio.