Misconceptions about Tax-Aware Long-Short: Misleading Parallels with Direct Indexing, Part II

In Part I of this series, we examined how comparisons with direct indexing can lead to misconceptions about tax-aware long-short (TA LS) investing. We explained that the investment process is driven by the pursuit of pre-tax alpha, with loss realization occurring as a byproduct. In this second installment, we turn to two related misconceptions about the sources of TA LS tax benefits: first, that greater loss realization, rather than gain deferral, is the primary driver of tax benefits; and second, that pre-tax alpha is largely irrelevant to tax benefits.

For passive direct indexing, the main reason for trading is to harvest losses. Extrapolating this logic to TA LS leads to the false conclusion that TA LS strategies try to realize more losses than a comparable tax-agnostic strategy would. Furthermore, because TA LS typically realizes larger losses than direct indexing, another false conclusion is that TA LS is aggressively harvesting losses, kind of like "direct indexing on steroids."

However, our research shows that neither of these conclusions, both derived from extrapolating direct indexing logic, is true. Quite to the contrary, we find that most of the tax benefit of TA LS strategies, relative to their tax-agnostic counterparts, comes from the deferral of gains rather than the realization of larger losses. 1 1 Close For a short overview of this study, see Looking Under the Hood of Long/Short Tax-Aware Strategies (AQR Tax Matters, 2023).

In practice, a TA LS strategy trades continuously to express the alpha model's evolving views. In doing so, it realizes losses as a natural byproduct of typical active-strategy turnover — positions the alpha model no longer likes are sold, and some of those happen to have declined in value. This occurs at roughly the same pace as in a tax-agnostic strategy.

What the tax-aware optimization adds is a systematic effort to avoid liquidating gain positions – particularly short-term gain positions – unless the alpha model strongly favors doing so.

The net effect is that, compared with a tax-agnostic active strategy, the tax-aware strategy realizes roughly the same level of losses but far fewer gains. Realizing losses at a normal pace while deferring gains into the future produces a net capital loss.

The exhibit below makes this result concrete:

Three-Year Cumulative Gains and Losses Realized by the Strategies, as Percentage of Initially Invested Capital

Source: Krasner and Sosner (2024), "Loss Harvesting or Gain Deferral? A Surprising Source of Tax Benefits of Tax-Aware Long-Short Strategies." For illustrative purposes only. This exhibit shows total aggregate gains and losses realized during the first three years since inception as a percentage of initially invested capital. The boxes show the 25th, 50th, and 75th percentiles of aggregate gains and losses across 34 three-year-long vintages and the whiskers show their 10th and the 90th percentiles.

The "direct indexing plus shorting" framing feeds another misconception: that because TA LS strategies supposedly trade just to harvest losses, the level of pre-tax alpha is irrelevant to their tax benefits.

If anything, by extrapolating direct indexing logic, one might even expect alpha to hurt tax benefits. Indeed, it has been shown that, in direct indexing, higher returns mean fewer positions trading below cost, and, therefore, fewer losses to harvest. And here again, the extrapolation leads to the wrong conclusion.

In fact, our research finds the opposite: pre-tax alpha materially enhances the tax benefits of TA LS strategies. 2 2 Close For a summary of this study, see Why More Alpha Can Mean More Tax Benefits (AQR Tax Matters, 2026). Two effects drive this result:

• Faster portfolio (NAV) growth. Even if tax benefits were to shrink as a percentage of assets when returns rise, higher pre-tax alpha grows the asset base faster. Even if the benefits decrease in percentage terms, applying them to a larger asset base can still increase their dollar value.
• Position creation unique to long-short. When a long-short strategy generates alpha (which adds to the asset base), its leverage falls below target, so the manager adds new long and short positions to rebalance back to the leverage target. Those newly created positions start without embedded gains, so alpha behaves like a recurring capital contribution, without the investor adding any cash. This effect simply does not exist in long-only direct indexing.

Our research found the effect of pre-tax alpha on the magnitude of tax benefits to be substantial and increasing with both leverage and the investment horizon.

The misconceptions illustrated in this Part II show yet again that the "direct indexing plus shorting" mental model for TA LS investing most of the time leads to conclusions that are plain wrong.

TA LS strategies are built first and foremost to earn pre-tax alpha through active management, much like their long-standing tax-agnostic counterparts. This is fundamentally different from direct indexing, which is designed to track a passive index. Importantly, our point is not that TA LS is better or worse than direct indexing. We merely want to highlight that the two investment approaches differ so significantly that familiarity with one tells you very little about how the other works.