The 12 signals · 2 min · Updated 27 Sept 2026
A company can look cheap per share and still cost you through dilution. How we track the number of shares, and why a reverse split must not look like a buyback.
Your stake is your shares divided by all the shares. When a company sells new shares, converts debt into stock or pays its staff in shares, the pie is cut into more slices and yours shrinks, even if the business does well. A company that buys back its own stock does the opposite.
From the filings. Every 10-K and 10-Q states on its cover page how many shares were outstanding on a recent date. We read that figure from every filing, which gives a point roughly every three months. When a company has two classes of shares and the cover gives them separately, we use the total from the balance sheet instead.
A one-for-thirty reverse split divides the share count by thirty overnight. Read naively, that is a 97 % buyback. Our series adjusts every point for splits, and it applies a split only when the jump in the actual share count around its date confirms it. If a declared split does not match what the count did, the earlier points are dropped rather than adjusted wrongly.
A real case, from the filings: FuelCell Energy did a one-for-thirty reverse split in November 2024. Unadjusted, its share count fell by 95 % in a year. Adjusted, it went from about 15 million shares at its fiscal 2023 year-end to about 80 million in August 2026: more than five times as many.
The share count trend is one of the twelve signals of the Leak Score. It counts in favour when the adjusted count falls more than 2 % in a year, it is neutral up to a 1 % rise, and it counts against above that.
The dilution section of the dossier, for Premium members, shows the adjusted share count and its trend, insider buying and selling from Form 4 filings, the public float and whether the baby shelf rule applies.