Short answer from the research: high short interest and days to cover help you estimate “fuel” and crowding, but they do not reliably predict explosive upside by themselves. In most large-sample research, heavy shorting is usually a bearish signal because short sellers tend to be informed and are often early on negative fundamentals.

At the same time, squeezes do happen, but they are rare, short-lived, and catalyst-driven (price shocks, recalls, or coordinated buying pressure).

Where the edge starts to show up in backtests is when you treat short interest and days to cover as a conditional ingredient and then require additional confirmation: utilization/borrow tightness, major news catalysts (especially earnings), and real-time buying pressure (volume + momentum or social attention).

Related reading: Short Interest Strategies (Backtest)

A short squeeze is easiest to understand as a feedback loop: short sellers must buy shares to close, and when the price rises quickly, that “buy-to-cover” demand can add incremental pressure right when liquidity is worst. The U.S. Securities and Exchange Commission describes the core mechanism: a sharp price increase can trigger margin calls, forcing shorts to either post more collateral or exit; exits happen via buy orders, which can push price higher and cascade.

Two practical details matter for traders:

First, short selling typically occurs in a margin account, and brokers can require more margin when volatility is extreme. That is part of why squeezes are often violent: risk limits tighten at exactly the wrong time.

Second, short interest can exceed 100% of float because shares can be lent, sold, bought by a new holder, and then lent again, making the same underlying shares “count” multiple times in short interest statistics. The SEC explicitly noted this dynamic when discussing why GameStop’s short interest exceeded 100%.

One caution that shows up in regulatory and empirical reviews: in famous squeezes (including GameStop in 2021), short covering can contribute during some windows, but it may not be the primary driver of a multi-week run-up. SEC staff concluded that positive sentiment, not just buying-to-cover, sustained GameStop’s longer price appreciation.

“Short interest” is a position measure: the FINRA defines it as a snapshot of total open short positions on firms’ books and records as of a settlement date. In the US, firms report short interest twice per month, and FINRA/exchanges publish it on a schedule with a delay (publication can be the 7th business day after the settlement date). This matters because squeeze setups are fast-moving, and stale short interest is a real backtest trap.

The most common “how shorted is it?” metrics used in practice are:

The big conceptual point that shows up across academic work is that high short interest often reflects negative information. A classic large-sample result is that portfolios of heavily shorted stocks tend to underperform, especially in equal-weighted portfolios and in the subset of genuinely constrained stocks. Research also ties high shorting to anticipated negative earnings surprises and unfavorable news, consistent with informed short sellers.

That’s why “buy high short interest” as a standalone contrarian strategy usually disappoints in backtests: you’re usually buying something the market has strong, informed reasons to dislike.

Days to cover (DTC) is useful because it scales short interest by liquidity. FINRA’s glossary defines it as the number of days of average share volume required to buy all shares sold short, computed as short interest divided by average daily share volume.

Why traders like it is straightforward:

The deeper research nuance is important for your article thesis: days to cover can be informative, but mostly as a predictor of future returns’ direction via short sellers’ information, not automatically as a predictor of upside spikes. For example, a large international study finds the days-to-cover ratio and utilization are among the most robust short-sale measures for predicting future stock returns globally (in the direction consistent with informed shorting).

When you focus specifically on “explosive upside,” the evidence shifts. An empirical framework by State Street that targets retail-driven squeeze risk finds that (after controlling for other variables) days-to-cover is not a significant predictor of the next-month maximum “draw-up”. In their regression, utilization and a media-driven squeeze score show stronger explanatory power.

If you want a systematic “contrarian bullish squeeze” strategy that survives contact with data, the research pushes you toward a rule structure that looks like this:

Step one: treat high short interest and high days to cover as “squeeze fuel,” not as a buy signal by itself. Academic evidence says heavy shorting is typically bearish (negative future performance) because shorts are often informed.

Step two: require a credible catalyst that flips the narrative and forces risk limits. The squeeze itself is typically triggered by sharp price shocks and subsequent covering pressure, with the SEC describing the margin-call-driven feedback loop.

Step three: accept that squeezes are rare and short-lived. Large-sample evidence on short-squeeze prevalence finds that stock-day squeeze events are uncommon and often last only one day for a majority of cases.

A practical quantitative definition (consistent with what institutional “squeeze risk” models actually do) is to rank stocks by shorting pressure and borrow tightness, then only go long when a catalyst-driven price move appears.

Systematic Trading Rules: Entry, Exit, Risk Management, and Position Sizing
Institutional research models explicitly discuss using squeeze scores as an overlay to avoid being short the names most likely to squeeze, and they evaluate performance via mean return spreads and information ratios. That same architecture can be flipped for a long-biased squeeze capture: let “squeeze risk” pick the candidates, and let price/volume confirm the entry.

This is the point where “short interest + DTC” stops being a story and becomes an implementable screen.

Earnings are a natural trigger because the market often underreacts to earnings surprises over time (post-earnings announcement drift). In other words, the catalyst you want is not just “good news,” but good news that forces repricing.

Two pieces of evidence connect earnings and squeeze-style setups:

First, an open-access study on surprise in short interest finds that short interest report information is not incorporated instantly, and it documents a return drift and a risk-adjusted long–short spread on “surprise in short interest.” It also shows the predictive ability is materially stronger around earnings announcements. That supports the intuition that earnings are where short sellers and narrative shifts collide.

Second, IHS Markit’s short squeeze model work explicitly states that certain events, including earnings announcements, positive news events, and abnormal trading volume, increase the probability of a squeeze acting as a catalyst.

Why momentum and volume matter: a squeeze is, mechanically, an upside momentum event. The classic momentum literature finds meaningful abnormal returns in “buy winners, sell losers” strategies, while also emphasizing turnover and cost sensitivity. For squeeze trading, momentum is less about chasing and more about avoiding the biggest failure mode: buying high short interest names that never actually turn.

Where social attention fits: State Street’s framework combines securities lending pressure with digital and social signals and reports materially larger subsequent draw-ups for the highest-risk (“Alert”) regime. The SEC’s GameStop analysis also reinforces that in practice, sentiment and coordinated buying pressure can dominate over pure covering flow in sustaining a run.

Data realities drive the backtest design. If you backtest squeeze strategies with unrealistic timing, you will fool yourself.

A solid, defensible approach starts with these data constraints:

Short interest position data is published on a twice-monthly cycle and is a snapshot, not a continuous measure. FINRA explicitly warns that short interest is not the same as daily short sale volume, and it emphasizes the reporting cadence and publication process.

FINRA also provides the specific fields required to compute days to cover and explains its formula, as well as how average daily volume is computed over the reporting window.

If you use exchange-reported short interest in a backtest, you must align signals to when the market could have known them. The Deutsche Bank/DataExplorers research notes the importance of using timely data and explicitly discusses lagging timestamps (at least a couple business days, and they used a more conservative lag) to avoid look-ahead bias.

Universe selection: Institutional work on squeeze risk (State Street) describes a broad US universe, primarily the Russell 3000 plus some micro caps, when evaluating “draw-up” behavior. IHS Markit’s work uses a “highly shorted universe” and then ranks within it.

Data sources you can cite and actually use:

This section is where the headline question gets answered: high short interest and days to cover do not, on their own, backtest as a reliable “explosive upside” predictor. The evidence that does hold up is more conditional: squeeze risk rises with shorting pressure, but upside capture improves when you add real-time catalysts and tighter measures of borrow constraints.

In IHS Markit’s US research, a factor labeled “Short Interest Ratio (exchange data),” which they also call days to cover, shows 0.00% average D1–D10 return spread and a 0.000 information ratio before applying their short squeeze adjustment. After adjustment, the average spread improves to ~0.19% monthly with a modest information ratio.

That is a clean, backtest-based way to say: days to cover alone was not an edge in that sample; conditioning on squeeze risk helped.

State Street’s framework, focused on one-month maximum draw-ups, finds days-to-cover is not a significant predictor in their regression once other variables are included, while utilization and media signals are.

IHS Markit’s US short squeeze model reports:

The same report explicitly evaluates an overlay approach “using both mean return and information ratio,” and it shows short-squeeze adjustment improving information ratio for some short-interest-related factors.

For explosive-move magnitude rather than average-return compounding, State Street’s “Alert” regime shows an average maximum price increase (“draw up”) of about 33% over the following month for flagged names, compared to materially smaller averages outside the regime.

For CAGR, you can translate an average monthly spread into an implied annualized rate (this is a simplifying assumption, not a guarantee). For example, an average D1–D10 spread of ~1.12% per month implies roughly mid-teens annualized before costs if compounded, while 3.02% per month implies much higher, but the latter comes from a short out-of-sample window that is clearly regime-dependent.

For “Sharpe,” the IHS Markit work reports information ratio (IR) rather than Sharpe, which is standard for active spreads relative to a benchmark. Their days-to-cover factor example moves from IR 0.000 to roughly 0.082 after adjustment.

For “maximum drawdown,” the IHS Markit report provides performance charts and spread/IR statistics but does not publish a max drawdown in the excerpted tables shown here. That limitation is very common in vendor research and is one reason you should re-run these ideas on your own data with realistic slippage.

Even without full drawdown tables, the distributional story is well supported:

If you were screening this with a realistic mindset, you can implement it in layers that map directly to what the research says works.

Start with “fuel” and constraints (slow-moving, but necessary):

Upgrade to “pressure” (borrow tightness and utilization) if you can:

Then require a trigger (fast-moving, where the edge lives):

Risk factors you have to assume in implementation:

Comparison: short squeeze strategy vs traditional momentum trading strategies

A useful way to explain this to a smart friend: a “good” squeeze strategy tends to look like momentum with a constraint filter.

Classic momentum research finds economically meaningful abnormal returns in winner-minus-loser portfolios, but it also emphasizes high turnover and sensitivity to transaction costs. Squeeze strategies share that turnover problem (signals change quickly, and the trade window is short), and they add a second issue: execution gets harder exactly when you most need it, because volatility and spreads widen.

So, in today’s market, the practical takeaway is not “hunt high short interest and buy.” The research-supported framing is:

That combination is the through-line that shows up repeatedly in the strongest backtested and institutional frameworks.