Securities Lending and the Short Rebate

Securities lending is the market that supplies shares to short sellers. The fee you pay to borrow (or the rebate you earn on cash collateral) is a first-class return driver for any strategy with shorts — pairs, stat arb, convert arb, and outright short factors. Ignore lend and your backtest is a fantasy. This article covers locate, rebate, HTB, and recalls.

The basic transaction

1. You want to short stock XYZ
2. Prime broker locates shares from a lender (long holder, ETF, custodian)
3. Shares are borrowed; you sell them in the market
4. You post cash (or other) collateral
5. You pay a borrow fee; collateral may earn a rebate

short_pnl ≈ −R_stock − borrow_fee + rebate_on_collateral − dividends_you_owe

See also short selling mechanics for corporate actions and buy-ins.

Easy-to-borrow vs hard-to-borrow

Status	Typical fee	Meaning
ETB (easy)	Near GC / few bps	Large supply
HTB (hard)	Hundreds–thousands of bps	Scarce supply
Threshold / no locate	Unshortable	Cannot initiate

HTB names are exactly where naive short-alpha looks best in price-only backtests — and where live P&L dies.

def short_net_return(stock_ret: float, borrow_fee_ann: float,
                     days: int = 1, div: float = 0.0) -> float:
    """Return to short seller over `days` (stock_ret = underlying return)."""
    fee = borrow_fee_ann * (days / 365)
    return -stock_ret - fee - div

Specials and the indicative fee

The indicative borrow rate is a quote, not a guaranteed fill. Specials spike around:

• Merger deals (merger arb)
• Heavily shorted meme / squeeze names
• Index deletions and ETF rebalance
• Low float IPOs

A signal that shorts high short-interest names without a fee model is often just harvesting a liquidity/lend premium you do not receive — or worse, paying it.

Recall risk

Lenders can recall shares. Your broker may buy in your short at a punishing price. Recall clusters when:

• Longs need shares for voting / corporate actions
• Lend supply migrates (ETF create/redeem)
• Crowded shorts unwind and supply shrinks further

Stress test shorts with forced cover at +10–20% gaps — standard stress testing for short books.

Rebate and financing

On the cash proceeds of the short:

rebate = benchmark_rate − spread

In easy names you may earn a rebate close to funding. In specials, rebate goes negative (you pay to hold cash against the short) — economically identical to a high borrow fee. Net financing must sit in the strategy ledger next to alpha.

Quant research implications

1. Point-in-time borrow fees — vendors revise; avoid lookahead
2. Universe filter — drop names with fee > threshold unless alpha clears it
3. Short capacity ≠ ADV capacity — limited by lend depth

(strategy capacity)

1. Crowding — high short interest + rising fees = squeeze fuel

(crowding)

def fee_adjusted_alpha(gross_alpha_ann: float, borrow_fee_ann: float) -> float:
    return gross_alpha_ann - borrow_fee_ann

If fee-adjusted alpha is near zero, the "short factor" was a lend trade in disguise.

Crypto analogue

Perps do not use stock loan; they use funding and margin. Spot short on CEX still needs borrowed inventory and can halt. Do not port equity rebate math to perps without replacing fee with funding + borrow.

Key takeaways

• Borrow fee and rebate are core short P&L — not footnotes
• HTB names inflate paper short alpha and kill live books
• Model locate failure and recall as hard risk events
• Filter or haircut signals by point-in-time indicative fees
• Short capacity is lend-constrained, not only ADV-constrained