Insights

General Collateral vs Hard-to-Borrow Explained

Alphanume Team · May 6, 2026

The borrow-fee spectrum and why it gates which strategies are implementable.

Securities lending operates on a spectrum from cost-free ("general collateral" or GC) to punitive ("deep special" or "deep HTB"). The classification depends on the relative supply and demand for borrowable shares of a given security. For short-side researchers and traders, the classification is the primary gating factor on what is implementable — strategies that look profitable in backtest can be uneconomic once realistic borrow costs are applied.

The spectrum

Tier	Typical fee (annualized)	Implication
General collateral (GC)	0–1%	Borrow effectively costless; shorts can be carried indefinitely
Easy to borrow (ETB)	0–5%	Functionally same as GC at most brokers
Mild special / mild HTB	5–25%	Material cost but workable for high-conviction shorts
Special / HTB	25–100%	Cost approaches or exceeds expected alpha; short carry is the dominant economic factor
Deep special / deep HTB	100%+	Borrow cost dominates; only short-duration trades viable

The cutoffs are approximate — different brokers use different terminology and thresholds.

What makes a name GC

General-collateral names share several features:

Broad institutional ownership (large index funds, ETFs, mutual funds holding the security).
Active participation in securities-lending programs by those institutional holders.
Float large relative to typical short demand.
Lack of corporate actions or events that produce recall pressure.

The vast majority of the S&P 500 trades as GC under normal market conditions. Borrow costs for these names are minimal and have low variability.

What pushes a name into HTB

The transition from GC to HTB is gradual. Drivers:

Short interest building. As more shares are sold short, the pool of available borrow shrinks.
Lender pulls supply. Holders exit positions or stop lending.
Recall waves. Voting events, corporate actions, or fund outflows recall borrowed shares.
Operational frictions. Recently issued shares, restricted shares, and SPAC-merger-emergent shares may not be in active lending pools.
Market structure. Small floats, recent IPOs, and de-SPAC issuers structurally support less borrow.

Rate volatility

HTB rates are far more volatile than GC rates:

A name at 50% on Monday can be 200% by Friday or back to 25% the following week.
Earnings announcements, M&A news, and broad short-cover events can produce 5–10x intraday rate spikes.
Lender recall events can render a borrowable name unborrowable within minutes.

For strategy modeling, treating HTB rates as static is a significant error.

Strategy implications

The borrow tier gates which strategies are economically viable:

GC names: Long-horizon shorts, pairs trades, and statistical arbitrage are viable. Carry cost is negligible.
Mild HTB: Multi-week shorts viable if expected alpha exceeds the cost. Daily/weekly strategies face meaningful drag.
Deep HTB: Only short-duration event-driven trades viable. Multi-month shorts are uneconomic regardless of expected alpha.

Most dilution-event short setups occur in names that range from mild HTB to deep HTB. This is the structural reason that borrow-cost-adjusted returns are the only meaningful return measure for short-side dilution research.

The rebate question

In US prime brokerage, short sales generate cash collateral that is posted to the lender. The lender pays back a "rebate" — typically interest minus a spread. Net of the rebate:

For GC names, the rebate roughly offsets the cost; net rate can be slightly positive (cash earns short-rebate interest).
For HTB names, the rebate is below market interest rates; net rate is firmly positive (you pay).
For deep HTB names, the rebate may be essentially zero or negative; the net rate is the entire borrow cost.

The headline "borrow rate" quotes at most brokers are net rates. The underlying mechanics involve the rebate, but the customer sees a single quoted number.

Sourcing borrow rate data

For backtests, historical borrow rates are essential. Sources:

Broker locate feeds. Your broker's daily borrow availability and rates.
Third-party data providers. Provide cross-broker historical borrow rate data.
Public aggregations. Limited but available for some names.

For point-in-time backtest accuracy, the daily historical rate as known on each historical date is required — see what is point-in-time data.