Days to cover

How it works

Days to cover is a ratio, and FINRA defines it precisely for the file it publishes: the number of days of average share volume it would take to buy all the shares sold short during the reporting cycle. The formula is short interest divided by average daily share volume, rounded to hundredths.

The interesting part is the denominator, which FINRA also defines: total volume — or split-adjusted volume where there has been a split — divided by the number of trading days from the day after the previous settlement date through the current one. In other words, FINRA's average daily volume window is the reporting cycle itself, roughly ten trading days, and it moves with the calendar.

Nobody else is obliged to use that window, and most vendors do not. S3 Short Interest Data publishes days to cover over ten, thirty and ninety-day windows, which is the honest treatment: three numbers, each labelled. ORTEX carries days to cover beside its lending-derived estimate. ChartExchange computes it against average daily volume on top of the FINRA filing. Each is doing something defensible, and the results are not the same number.

Why two correct figures disagree

Take a stock with ten million shares reported short. Over the current FINRA reporting cycle it traded two million shares a day, so FINRA's file prints 5.00. Over the last ninety days it averaged five million a day, because the cycle just ended was unusually quiet, so a vendor using a ninety-day window prints 2.00. Same filing, same stock, same day, and a figure two and a half times apart. Neither is wrong. They are answers to different questions, and only one of them is labelled with its window.

Now compound that with the numerator. A reported short interest figure describes a settlement date already a week or more in the past by the time it is published — the whole of short interest's timing problem is inherited here, unchanged. A vendor's daily estimate is not that number at all: it is modelled from securities-lending inventory, so the ratio becomes shares on loan divided by a volume average. That is a third quantity with the same name on the label.

Three smaller mechanisms make the same point:

Split adjustment. FINRA adjusts the volume series for splits explicitly. A pipeline that joins raw historical volume to a short interest series produces a ratio that breaks at every split, in the same way and for the same reason that any unadjusted series does — see corporate action.

Window length changes what the ratio measures. A short window tracks current liquidity and swings violently around earnings and index events. A ninety-day window is stable and describes a market condition that may have ended weeks ago. Neither is a better default; the choice is the answer.

The denominator is not buying power. Every share traded is counted, including the short sales themselves and the market-maker turnover that never establishes a position. The ratio treats total volume as if it were capacity available to close shorts, which is an upper bound and usually a loose one.

Why it matters here

The practical rule is short. Never compare a days-to-cover figure between two products in this category, and never quote one without its window. If a vendor does not publish the window and the source of the numerator, the number is not comparable to anything, including its own history if the methodology changed.

If the figure matters, compute it. FINRA's short interest file is free and carries its own days to cover with a documented window; a volume series you already have plus that file gets you a ratio whose definition you know. ChartExchange sells the filing with an API from $19.95 a month if the download is not worth automating.

And for the question days to cover is usually a proxy for — how hard and expensive the short side of this stock has become — the borrow market answers more directly and moves daily rather than twice a month. iBorrowDesk publishes one broker's borrow fee and availability for nothing, and a fee moving from 2% to 40% has told you something no ratio over stale positions can.