Dilution tracker benchmark
The dilution tracker benchmark tests each tracker against the SEC filing itself. In round 1 Signal8 Pro (DilutionScan and the Signal8 dilution tab) matched the filing on 10 of 10 figures, tied with AskEdgar Pro; DilutionWatch's free view matched 8. Raw CSV is on the round 1 page, with the plan tested for each tool.
The next round adds the remaining paid plans and hourly latency checks, under the protocol below. Its results will be published here as a table and CSV, every vendor's figures included.
The protocol
- Sample. 40 dilution events filed during one month: new ATM supplements, S-3 and F-3 shelves, priced offerings with warrants, and 8-K Item 3.02 unregistered sales, drawn at random from EDGAR, stratified by form.
- Ground truth. For each event, two people read the filing and record the instrument, the share count or dollar amount and the exercise or conversion price. Disagreements are resolved against the filing text.
- Latency. For each tool, the time from SEC acceptance to the event being visible in the tool, checked hourly for 48 hours.
- Instrument recall. Of the ground-truth instruments, how many each tool shows, and how many it shows with the right figures.
- Unmeasured handling. Whether a tool shows an instrument it has not read as zero or as unknown.
- Price. The monthly price for one user on the day of the test.