“We’re a $600M company; nobody’s algorithm even sees us.”
Whether the algorithms see you is a measurable question. The dataset records the market reaction — or the absence of one — for every announcement at every market cap, over 20+ years. For a small cap, the honest number is worth more than a flattering one.
Thin news flow means fewer measured events. The readout says so.
Five scoreable announcements a year is normal for a sub-$2B issuer. The readout does not stretch five events into a trend line — it scores each one against 20+ years of its event class and states the sample size in a coverage note. Two of these five registered. Three did not. Both numbers are the diagnostic.
| Date | Event type | Headline | Reaction % | Volume multiple |
|---|---|---|---|---|
| Feb 19 | Earnings | FY results, guidance initiated | +5.1% | 3.4× |
| Mar 27 | 8-K | Credit facility amendment | 0.0% | 1.0× |
| May 07 | Earnings | Q1 results | 0.0% | 1.1× |
| Jun 16 | Contract | Government contract award | +4.2% | 2.6× |
| Jul 02 | Product | Product certification | 0.0% | 0.9× |
Illustrative, anonymized. Twelve months, five announcements: two registered, three recorded as no measurable reaction (0.0% at baseline volume). The coverage note states the sample size — five events is five events; the readout does not pad it.
A measured no-reaction is a board-level finding.
When your earnings pass at +0.2% on baseline volume while a size-matched peer’s move +2.8%, the problem is not the quarter — it is reach. That is the difference between “nobody sees us” as a feeling and as a measured, class-by-class comparison you can put in front of the board.
Illustrative, anonymized. Median measured reaction by event class: you vs a size-matched peer over the same period. The contract row shows the wire is not broken — the earnings and 8-K rows show where visibility is lost.
The small-cap readout, stated plainly.
Which of your announcements registered — reaction, volume multiple, and percentile against 20+ years of the event class — and which were recorded as no measurable reaction.
The same measurements for peers of your size, so a quiet tape reads as a visibility gap or a market-wide non-event, not a guess. Where the contract news lands and the earnings do not, the fix has a location.
Every readout carries a coverage note with the measured event count. Sparse data is reported as sparse — that is the product working, not failing.
Our float is thin. Can a reaction be measured at all?
Yes. Reaction and volume are measured against your own baseline — your average volume, your normal volatility — so the noise threshold scales to the name. A 2% move means something different at $600M than at $60B, and the measurement accounts for that.
We put out a handful of releases a year. Is that enough?
The readout states the sample size plainly: five events is five events. Each one is still scored against 20+ years of its event class, and the peer comparison adds context the small sample cannot. Sparse is workable; padded is not.
Who counts as a peer at our size?
Issuers in the same cap band and sector with comparable news flow, drawn from the 26M+ news records dataset. The peer set is shown to you on the call before anything is benchmarked against it, and you can adjust it.