White paper: Long short direct indexing

This white paper explores an investment strategy that combines factor-based investing, leveraged long short portfolio construction, and a tax-efficient direct-indexing process. The strategy is designed to track a benchmark index, such as the Russell 1000, while maintaining excess exposure to selected factors like growth. It periodically rebalances to harvest tax losses. This approach allows investors to express their views on factor exposures in a tax-aware manner, potentially generating pre-tax alpha and significantly improving after-tax returns.

Factor-based investing has become a cornerstone of systematic equity strategies, allowing investors to tilt portfolios toward characteristics such as value, size, momentum, quality, and growth [1, 2, 3]. However, many traditional implementations, including factor ETFs (exchange-traded funds), are typically long-only and invest in a narrower segment of the investable universe. Additionally, their structure generally lacks the flexibility to fine-tune factor exposures or efficiently harvest tax losses. By contrast, a direct indexing approach allows for broader market access, more precise control over factor tilts, and improved tax efficiency through ongoing, systematic loss harvesting.

Building on this insight, the strategy presented in this paper takes full advantage of direct indexing while incorporating a persistent tilt toward the selected factors. Unlike long-only tax loss harvesting, which becomes less effective over time due to gradual reduction in a portfolio’s ability to realize capital losses as individual positions accumulate unrealized gains—the long short structure enhances the ability to generate tax losses in various market conditions [3, 4].

Overview

Let’s examine a 140/40 growth-tilted portfolio that tracks the Russell 1000 index and break down what each term entails.

The growth factor, as defined by the Barra Global Total Market Equity Model for Long-Term Investors (GEMLT), captures characteristics of companies expected to grow their sales and earnings at an above-average rate. It incorporates multiple signals including analyst forecasted long-term earnings growth and historical sales and earnings growth.

A 140/40 long short portfolio is structured by taking a 140% long position in selected securities and a 40% short position in others. The proceeds from the 40% short sale fund the additional 40% long exposure, keeping the net market exposure close to 100%—consistent with the benchmark index. This structure provides greater flexibility to express factor tilts without increasing the investor’s overall market risk.

The long short design also enhances the potential for tax loss harvesting in a variety of market environments. When markets decline, the long side of the portfolio tends to accumulate losses that can be harvested. Conversely, in rising markets, short positions can generate realized losses as prices move against them—providing opportunities to offset capital gains elsewhere in the investor’s portfolio. This dual-sided mechanism greatly enhances the consistency and volume of tax loss harvesting compared to long-only strategies.

Methodology 

Our strategy builds upon the framework developed in our prior white papers, where rebalancing a direct-indexing portfolio is formulated as a constrained optimization problem. The core objective remains the same: to efficiently trade off tracking error minimization against tax-loss harvesting opportunities and transaction costs.

As in the earlier approach, we define a composite objective function incorporating a factor-based risk model, expected tax outcomes, and estimated trading costs. This function is minimized subject to additional constraints, which we will outline later, using a state-of-the-art numerical solver. The objective function is defined as follows:

   

Recall that the objective function consists of three key components:

The first component quantifies the expected tracking error, defined as the forecasted variance of the portfolio’s return relative to that of the benchmark. Minimizing this term ensures the portfolio remains closely aligned with the benchmark, maintaining similar performance over time.

The second component captures the net tax impact of rebalancing trades. Loss-generating trades contribute negatively to the objective function, thereby encouraging tax-loss harvesting, while gains are penalized to discourage tax liabilities. 

The third component models estimated trading costs, such as bid-ask spreads. Including this component discourages unnecessary turnover.

While the core optimization framework remains consistent with our previous long-only direct indexing approach—including constraints such as wash-sale avoidance—the key distinction in this factor-tilted long short strategy lies in an expanded constraint set that enables more sophisticated portfolio positioning. Specifically, we introduce new constraints to implement factor tilts, manage leverage, and control risk exposure, while still preserving tax efficiency and benchmark alignment.

To implement a factor tilt and control the portfolio’s exposure to each factor, we impose the following constraints for each factor:

   

To manage leverage, we constrain the gross exposure—the sum of the absolute portfolio weights—to a target level. In the case of a 140/40 portfolio, the short exposure is denoted by , where , and the long exposure is , resulting in a gross exposure of :

   

To ensure the rebalanced portfolio maintains a consistent level of market exposure, we constrain the portfolio beta relative to the benchmark index. Beta is a measure of a portfolio’s sensitivity to market movements; a beta of 1 implies that the portfolio is expected to move in tandem with the market, while values above or below 1 indicate amplified or muted sensitivity, respectively. To enforce this, we impose the following constraint on the portfolio’s total beta exposure:

   

Here, is the vector of asset-level betas

   

and is a small slack parameter that allows limited deviation. This ensures that the portfolio maintains a market exposure consistent with the benchmark.

Furthermore, to mitigate the risk associated with concentrated short positions, we impose concentration constraints that limit the size of individual holdings. Unlike long positions, where losses are capped at 100%, short positions can generate unbounded losses if not properly managed. To control this risk, we restrict the magnitude of each stock position:

   

These constraints help reduce exposure to idiosyncratic risk.

Simulations

To evaluate the performance of our strategies, we conducted a series of historical simulations using overlapping 10-year periods from April 1, 2005 to February 13, 2025. Each simulation was launched on a quarterly basis between April 1, 2005 and March 16, 2015, using market, constituent, and risk data available through February 13, 2025. 

In total, 41 simulations were performed. Each simulation begins with a one-time cash investment of $1,000,000 and proceeds with monthly rebalancing based on the respective strategy logic. The long-only (LO) strategy tracks the Russell 1000 via a direct indexing approach, while the long short (LS) strategy uses a 140/40 structure with a growth factor tilt. Both strategies are implemented using a consistent optimization framework and point-in-time forecasts from a factor-based risk model, ensuring no look-ahead bias.

All simulation results are deterministic—they use fixed inputs and a non-randomized optimization process. This ensures that multiple runs of the same simulation will produce identical results, eliminating the need for repeated trials or averaging.

Each strategy incorporates the following cost assumptions. The long-only portfolio is subject to an annual AUM fee of 0.22%, while the long short portfolio carries a higher management fee of 0.50%, reflecting its increased complexity and operational cost.

Transaction costs are applied at a flat rate of 2 basis point (0.02%) of traded notional (see our prior discussion on direct indexing transaction costs), charged symmetrically on both buys and sells. This means that an annual turnover rate of 529.34% will result in a 0.11% annual transaction cost. In addition, the long short portfolio incurs a financing cost: a 1.25% annual spread applied to the net debit balance, which is 40% of the portfolio value, resulting in a total annual charge of 0.5%.

Results

The following table summarizes simulation results for the long-only and long short strategies, with annualized return and cost components presented on both a pre-tax and post-tax basis. For post-tax calculations, we assume a 42.3% short-term tax rate and a 28.1% long-term tax rate¹. We assume that investors have sufficient long-term and short-term gains outside of the strategy, allowing losses to generate tax savings at the respective rates. Returns and fee rates are computed monthly and then annualized.

	Long-only	Long short	Difference
Post-tax return	1.17%	2.51%	1.34%
Tax alpha	1.25%	3.35%	2.10%
Pre-tax return	0.15%	0.29%	0.14%
Management fee	-0.22%	-0.50%	-0.28%
Transaction cost	-0.01%	-0.11%	-0.10%
Financing cost	0.00%	-0.50%	-0.50%
Loss harvested	3.06%	7.76%	4.70%
Tracking error	1.06%	1.66%	0.60%

The growth-tilted long short strategy outperforms the long-only direct indexing approach on a post-tax return basis, delivering 2.51% annually compared to 1.17% for the long-only portfolio—a difference of 1.34%. This outperformance is driven mostly by tax efficiency. The long short strategy realizes substantially more harvested losses (7.76% vs. 3.06%), resulting in a tax alpha of 3.35%. Despite its higher transaction costs (0.11% vs. effectively 0.01%) and a financing cost of 0.50% (which the long-only strategy does not incur), the long short strategy’s enhanced tax benefits more than compensate for these implementation frictions. 

The following figure shows the mean cumulative losses harvested—as a percentage of initial capital invested—by both the long-only and long short strategies over a 10-year period. 

The bars represent the average outcome across simulation runs, while the error bars capture the standard deviation, indicating the level of variability in loss harvesting performance from one simulation to another due to different market conditions. By year 10, the long short strategy on average has harvested over 100% of the originally invested capital in cumulative losses, more than three times the cumulative harvest of the long-only strategy for the Russell 1000 index, which levels off near 31%. The results illustrate the tax-efficiency advantage of the long short strategy, which consistently harvests more capital losses each year compared to its long-only counterpart.

The error bars (standard deviation) provide insight into how consistent these results are across different market conditions. While both strategies exhibit some variation, the long-only strategy shows relatively higher dispersion in later years. This is likely due to the declining availability of loss-harvesting opportunities over time, which makes its outcomes more dependent on the specific return path of each simulation. Conversely, the long short strategy maintains more stable performance, with variability increasing more gradually and remaining narrower relative to its mean.

Discussion

Through historical simulations, we demonstrate that a factor-tilted long short direct indexing strategy can enable investors to express factor-based views in a tax-efficient way. By allowing investors to take on short positions, the strategy not only enhances flexibility in portfolio construction but also significantly increases the availability of loss-harvesting opportunities.

Compared to a traditional long-only direct indexing approach, the long short strategy offers substantially greater tax efficiency. Across all simulation periods, it consistently generated higher levels of harvested losses and post-tax returns, despite involving extra complexity and incurring additional costs. This strong tax alpha can be especially valuable for investors in high marginal tax brackets seeking to reduce their near-term taxable liabilities.

That said, the benefits of the long short approach come with important trade-offs. The inclusion of short positions introduces additional sources of risk, including higher tracking error and greater sensitivity to market volatility. Moreover, the strategy entails higher implementation costs, such as financing costs on short positions and a larger operational footprint.  Nevertheless, our historical simulations show that the after-tax performance gains can more than offset the increased costs and risks, particularly over longer horizons. For investors with a strong preference for factor tilts and a high sensitivity to after-tax returns, the long short strategy offers a compelling extension of direct indexing—one that broadens both the investment and tax management toolkits.

Join the waitlist

Long short direct indexing will be available soon. Sign up on our waitlist for early access.

References

1. Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. The Journal of Finance, 47(2), 427–465.
2. Fama, E. F. 1993. Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics 33(1): 3–56.
3. Goldberg, L., Lewis, C., & Zorn, T. (2024). The Enhanced Value of Tax-Loss Harvesting in Long-Short Portfolios. Research Paper.
4. Sialm, H., & Sosner, S. (2018). Tax-Loss Harvesting and Portfolio Performance. Management Science, 64(2), 705–722.
5. Khang, K., Cummings, A., Paradise, T., & O’Connor, B. (2022). Personalized Indexing: A Portfolio Construction Plan, https://corporate.vanguard.com/content/dam/corp/research/pdf/personalized_indexing_a_portfolio_construction_plan.pdf

This white paper describes implementation and performance details of a long short direct indexing approach similar to that used on the Frec platform at the time of writing (June 2025) details may differ from the implementation used in the product now and in the future. This paper may be amended at any time to reflect new findings, improve readability, or correct inaccuracies.

This white paper is for information purposes only and is not intended as tax advice or a trade recommendation. Clients should consult with their personal tax advisers regarding the tax consequences of investing with Frec and engaging in these tax strategies, based on their particular circumstances. Clients and their personal tax advisors are responsible for how the transactions conducted in an account are reported to the IRS or any other taxing authority on the investor’s personal tax returns. Frec assumes no responsibility for tax consequences to any investor of any transaction. 

The effectiveness of Frec’s tax-loss harvesting strategy to reduce the tax liability of the client will depend on the client’s entire tax and investment profile, including purchases and dispositions in a client’s (or client’s spouse’s) accounts outside of Frec, the type of investments (e.g., taxable or nontaxable) or holding period (e.g., short-term or long-term. The performance of the new securities purchased through the tax-loss harvesting service may be better or or worse than the performance of the securities that are sold for tax-loss harvesting purposes.