Call-IV Inflation on Hard-to-Borrow Stocks

_On HTB names, the forward price sits below spot because of the borrow cost. Through put-call parity, this depresses the synthetic forward and inflates call IV relative to put IV at equivalent delta. Sophisticated commenter on SPCE post (trading at 400%+ call IV) correctly identified that 'some of that 405 print is the cost of carry, not actual move expectation.' Selling calls on HTB names or buying puts vs calls at same delta may capture this premium._

Hypothesis

On hard-to-borrow stocks, call IV is systematically inflated relative to put IV via put-call parity borrow cost leakage, making call selling overpriced.

Method

Approach: Three-part: (1) theoretical model, (2) empirical options snapshot, (3) Monte Carlo simulation

_No historical options IV; simulation and snapshot only_

Analysis

The hypothesis rests on a well-understood consequence of put-call parity under stock-borrow constraints. When a stock carries a positive borrow rate — as all hard-to-borrow (HTB) names do by definition — the cost of carrying the short position is priced into the options surface. Because the synthetic short (long put, short call) must compete with the cash short, the higher borrow cost depresses the effective forward price, which in turn inflates implied volatility on the call side relative to an equivalent-delta put. The effect is not subtle: at a 20% annualised borrow rate, theory predicts approximately 7% excess call IV above put IV on a one-month tenor; at 50%, roughly 18%. For meme stocks at peak short-squeeze pressure, borrow rates have historically exceeded 100%.

The mechanism is recognised in academic options pricing literature and is consistent with empirical observations during high-profile short squeezes. Market-makers who write calls on HTB names must hedge by borrowing stock to delta-hedge; that borrowing cost enters their break-even and is passed through as wider call IV. A strategy that systematically sells calls on stocks with verifiably elevated borrow rates — while hedging or structuring to limit naked exposure — would in principle capture this mispricing each time the borrow-adjusted IV surface is distorted. The AMC data point from the original snapshot (implied borrow ≈ 215% annualised from the +29.9% call-put IV spread) is consistent with this behaviour being real and material.

Reproducing this as a systematic backtest is not feasible with freely available data. yfinance provides equity price and volume history but no per-strike option implied volatility, no borrow fee history, and no short-interest time series at daily or intraday resolution. Backfilling historical IV requires a paid options dataset — ORATS, OptionMetrics, or similar — which costs from several hundred to several thousand dollars per month. Borrow-rate history is even harder to obtain: Interactive Brokers publishes indicative borrow rates to their own clients, but no clean historical feed is freely accessible. Without both data sources, there is no way to identify qualifying HTB opportunities ex-ante, split an in-sample/out-of-sample window, fit a signal, or measure realised edge.

The p-value previously attached to this experiment was produced by a Monte Carlo bootstrap that was, at the time, miscalibrated — specifically, the null distribution did not correctly preserve autocorrelation structure in the simulated return series, causing it to assign artificially low p-values to strategies that had no genuine statistical power. That bug has since been corrected in the pipeline’s monte_carlo_pvalue helper. Because the earlier figure cannot be trusted, significance has been set to null here rather than carried forward as evidence of an edge that was never properly tested.

The theoretical case for call-IV inflation on HTB names is sound, and the AMC snapshot confirms the effect exists under the right conditions. However, those conditions are dynamic — HTB status fluctuates as short interest evolves — and identifying them reliably requires institutional-grade borrow-rate feeds that are outside the scope of this free-data pipeline. The experiment is therefore parked as INCONCLUSIVE rather than promoted to paper trading. It warrants revisiting if a borrow-rate or historical-IV data source becomes available.

Source

Originated from discussion on r/thetagang.

---

_Generated by labs-algo-trading. Automated research — not financial advice. Backtests overfit; treat verdicts as hypotheses._