Style Premia Across Asset Classes

A stock-picker who has only ever traded equities tends to think value and momentum are equity phenomena: cheap stocks beat expensive ones, winners keep winning. They are not equity phenomena. Cheap currencies tend to outperform expensive ones. Commodities in backwardation tend to outperform commodities in contango. Bonds with steep curves tend to roll down profitably. The same handful of ideas — buy what's cheap, buy what's rising, buy what pays you to hold it — reappear in every asset class quants have looked at, which is the strongest evidence that they reflect something structural about how markets price risk, not a quirk of one market's data.

The same style, four different costumes

Take momentum. In equities it means ranking stocks by their trailing 12-month return and buying the top decile. In currencies it means buying the currencies that have appreciated most against the dollar over the past year. In commodities it means going long the futures that have trended up. In government bonds it means favoring the countries whose yields have been falling (prices rising) fastest. Four different instruments, four different data sources, one ranking rule.

Value works the same way: price-to-book for stocks, purchasing-power-parity deviation for currencies, the futures curve's slope relative to its own history for commodities, real yield relative to history for bonds. Carry — the return earned just for holding a position if nothing moves — is dividend yield minus financing cost for equities, the interest-rate differential for currencies, the futures roll for commodities, and term premium for bonds (see The FX Carry Trade and Building a Carry Basket and Its Crash Risk for the mechanics in two of these).

Why cross-asset blending actually helps

Two return streams that are each choppy on their own can combine into something far smoother if they don't move together. That's the whole case for this style of investing: momentum in equities and momentum in currencies are both real, both persistent-ish, and historically close to uncorrelated with each other, because a US equity momentum crash (say, a violent reversal after a recession bottom) has nothing mechanically to do with a currency momentum drawdown (say, a central bank surprise).

Worked example 1. Suppose equity momentum and FX carry each deliver a standalone Sharpe ratio of 0.5, and their return correlation over a long sample is $\rho = 0.1$ — nearly independent. An equal-risk-weighted blend of two strategies with the same individual Sharpe $S$ and correlation $\rho$ has a combined Sharpe of

$S_{\text{blend}} = S\sqrt{\frac{2}{1+\rho}}$

Plugging in $S = 0.5$ and $\rho = 0.1$: $S_{\text{blend}} = 0.5\sqrt{2/1.1} = 0.5 \times 1.35 \approx 0.67$. Blending two mediocre, nearly-independent streams produced a meaningfully better one — not because either strategy got smarter, but because their bad days rarely land on the same date.

Drag this scatter's correlation toward zero and watch the cloud lose its slope — that's what equity-momentum returns plotted against FX-carry returns actually look like month to month: no visible relationship, which is exactly the property a portfolio wants from its ingredients.

Worked example 2. In August–September 2008, equity value strategies and commodity carry strategies both drew down as the financial crisis hit, but currency momentum kept working for several more weeks because the yen-funded carry unwind (an FX-specific event) hadn't yet spread to commodity futures curves. A book running all four styles across all four asset classes lost less in that window than a book running the same styles in equities alone — the diversification showed up exactly when it was needed, though it did not prevent the eventual carry-unwind losses once the crisis reached FX in October.

What erodes it

Two things chip away at cross-asset style premia. First, crowding: the same handful of factors are now run by CTAs, risk-parity funds and multi-strategy platforms simultaneously, and when several of them delever at once (a "quant quake," as in August 2007 or briefly in 2018), the promised cross-asset independence breaks down precisely because everyone's book is correlated through the trade itself, not through fundamentals (see Factor Crowding). Second, transaction costs compound across markets — a momentum signal that is marginally profitable in liquid equity futures can be unprofitable in a less liquid currency or commodity, so "the same signal everywhere" understates how much implementation quality varies by asset class.