Short Interest Decreases — Positions Being Covered

GO leads this ranking, followed by GEO and BFH. The ranking measures the change in reported short positions between two consecutive FINRA settlement files, so it reflects positions that were open on those two dates rather than trading that happened in between. The figures on this page come from the settlement dated Sep 15, 2026.

FINRA collects short positions twice a month, on the settlement dates around mid-month and month-end, and publishes each file about seven business days later, on dates it announces a year ahead. That gap is why this page states its settlement date and the age of the figures: a position can have changed substantially before the next file confirms it.

Short interest as a percentage of float is the shorted position measured against the shares actually available to trade, so it answers how crowded the short side is. Days to cover divides that position by average daily volume, answering how long ordinary trading would take to buy it back. Where no reliable float exists, the percentage is measured against shares outstanding instead and the row says so, because the same position reads lower on the wider denominator.

Less than it appears. A fall in the reported position is consistent with shorts buying back, which is the reading most people reach for, but it is equally consistent with a share count that changed underneath the figure: a reverse split rebases the position without anyone trading, and a symbol leaving the settlement file entirely can read as a collapse to zero. Even genuine covering says nothing about why, since a short may be closing at a profit, at a loss, or simply rebalancing. Check the raw share counts beside the percentage and whether a corporate action landed between the two settlement dates before treating a decrease as a signal.

A squeeze needs three things together, and no single column identifies one. First, a crowded position: a high percentage of float, not merely a large share count. Second, difficulty exiting: a high days-to-cover figure, meaning ordinary volume cannot absorb the buying that covering requires. Third, a catalyst that forces the timing, such as an earnings surprise, a regulatory decision or a financing announcement. Screening on any one of the three in isolation produces mostly false positives. Rows marked as split-affected belong in none of it: a split between the two settlement dates rebases the share count, so their change reflects arithmetic rather than anyone trading, and they are marked precisely so they can be set aside.

No. Short interest records positioning, not outcomes. Institutions short for hedging as well as for directional bets, so a large position may be one leg of a trade rather than a view on the company. Heavily shorted stocks have both fallen and risen sharply; the data tells you who is positioned how, and nothing about who is right.