Insights

What Is Borrow-Cost-Adjusted Return?

Alphanume Team · April 21, 2026

Netting borrow fees against gross short alpha — the only credible return measure for short-side research.

Borrow-cost-adjusted return is the gross return on a short position less the borrow fees paid to maintain it. For short-side research, it is the only return metric that meaningfully captures the realized economics. A short strategy that produces 15% gross annualized return but pays 12% in borrow fees produces 3% net — not 15% — and any analysis that doesn't make this adjustment misrepresents the strategy.

The basic formula

For a short position held over period T:

Net return = Gross return − Borrow cost − Dividend liability

Where:

Gross return: The price-change return on the short position (negative if the stock rose, positive if it fell).
Borrow cost: Cumulative daily borrow accrual over the holding period.
Dividend liability: Sum of dividends paid by the company during the holding period (if any).

Why "gross return" misleads

For long strategies, gross return is approximately net return — costs are small (commissions, slippage) and don't materially change the conclusion. For short strategies, the cost wedge can be huge:

A 90-day short in a name with 30% annualized borrow rate costs 7.5% in borrow alone.
A 30-day short in a name with 100% annualized borrow rate costs 8.3%.
A 60-day short in a name with 200% annualized borrow rate costs 33%.

For deep HTB names, borrow cost frequently consumes the majority of expected alpha. Reporting gross returns without adjustment is misleading for any HTB-heavy strategy.

The data requirement

To compute borrow-cost-adjusted returns properly:

Daily historical borrow rate for each name in the strategy.
Daily mark-to-market position value (gross return naturally provides this).
Dividend events (declaration, ex-date, payable date, amount).
Daily accrual computation.

For the borrow rate, point-in-time historical data is required — not current rates. Many free data sources only provide current rates, which is inadequate for backtests of strategies in HTB names.

Worked example

Consider a 60-day short opened at $50 in a name with an average 50% annualized borrow rate and a 2% annual dividend yield:

Initial position: $50,000 (1,000 shares × $50).
Final stock price: $42 (16% decline).
Gross profit: $8,000.
Borrow cost: $50,000 × 50% × (60/360) = $4,167 (approximating using initial value; daily mark-to-market would refine this).
Dividend liability: $50,000 × 2% × (60/360) = $167 (one ex-dividend date in the window).
Net profit: $8,000 − $4,167 − $167 = $3,666.
Gross return: 16%. Net return: 7.3%.

The borrow cost consumed more than half the gross alpha.

The distribution effect

Borrow-cost adjustment changes not just the mean but the distribution of strategy returns:

Best trades (large declines) are reduced by the same percentage cost.
Worst trades (large rallies) are further degraded by the same borrow cost.
The hit rate is lower on a net basis than a gross basis.
The Sharpe ratio is materially worse on a net basis.

Strategy ranking can change with borrow-cost adjustment. A strategy that looks better gross may underperform an alternative on net basis if it operates in higher-borrow names.

The relationship to dividend liability

Borrow-cost-adjusted returns should also include dividend liability where relevant — see dividend liability on shorted stocks. For most dilution-event short candidates (non-dividend-paying small caps), dividend liability is minimal. For broader short strategies in dividend-paying names, it can be material.

How to model in backtests

For each position in the backtest:

Record entry date, entry price, position size.
Track mark-to-market daily.
Accrue daily borrow cost using the historical rate for that date and the current position value.
Apply dividend liability on ex-dividend dates.
Compute net return = gross return − cumulative borrow cost − cumulative dividend liability.

Reporting net return as the headline metric, with gross return as supplementary.

The capacity question

Beyond per-trade cost, borrow cost affects strategy capacity:

Large positions in HTB names increase the borrow demand and can push rates higher (the strategy's own activity worsens its costs).
Strategy size is gated by available borrow inventory; in deep HTB names, scaling up may be impossible regardless of expected alpha.
For capacity-conscious institutional implementations, focusing on the GC end of the borrow spectrum may produce more scalable strategies than alpha-maximizing in HTB names.