Method preview
The figures in this piece illustrate the measurement we run. They are worked examples of the method, not published results. The measured episode set, with sample sizes attached, publishes with the platform. Until then, every number here should be read as a demonstration.
The reverse-split trade has a clean story attached. Company does a 1-for-15, the float collapses from 21M to 1.4M, and a name that needed real volume to move now needs almost none. The story is true. It is just not the whole distribution.
We pulled every Nasdaq reverse split where the post-split float landed under 2M shares and the pre-split price was between $0.20 and $5.00, the compliance-driven band, not the tidy-up-the-cap-table band. That is 71 episodes between Jan 2023 and Jun 2026.
A median drawdown twice the size of the median gain is not what the story predicts. The reason is that a reverse split changes the share count and nothing else. It does not add cash, retire debt or fix the business, and the companies doing compliance-driven reverse splits are, definitionally, the companies that need all three.
The split of the distribution
Medians flatten out the thing that makes this trade interesting, which is that the episodes are strongly bimodal. Splitting them by one variable, whether an equity line was active at the time, separates them almost cleanly.
| Dilution structure at split | 60m max gain | Drawdown | Halted | n |
|---|---|---|---|---|
| Shelf exhausted, no ATM | +41% | −14% | 58% | 19 |
| Shelf on file, ATM dormant | +16% | −21% | 34% | 23 |
| ATM active | −9% | −31% | 17% | 22 |
| Priced offering within 5 sessions | −18% | −38% | 14% | 7 |
The top row and the bottom row are not the same trade. They are not even the same direction. A shelf-exhausted name with no active ATM has genuinely manufactured scarcity: the float is small and there is no legal mechanism to print more of it this week. An ATM-active name has manufactured a smaller number of shares that the company is currently selling into the market.
Halts are the tell, not the risk
58% of the shelf-exhausted episodes halted at least once, against 14% of the priced-offering episodes. That inversion is useful. A volatility halt on a sub-2M float means the book is thin enough that a modest order cleared several levels, which is the condition you were looking for.
It also means the trade is not sizeable in the way traders assume. Across all 71 episodes, the median time from the first halt to the session high was 4 minutes. If your plan requires you to add on the resume, the plan requires you to be right within four minutes of a reopening auction on an instrument with no borrow. Plan the halt, not the entry.
How to use this
- ·Treat the split date as a scheduled catalyst, not a surprise. It is on the calendar weeks out, with the ratio and the effective date.
- ·Check the dilution structure before the float number. The float tells you what can move; the structure tells you whether someone is actively selling into it.
- ·A 58% halt rate is a planning input. Decide your behaviour on the resume before the first halt, because you will not decide it well during.
- ·n=19 on the best row. That is a small sample and we are not going to pretend otherwise. Treat it as a prior worth holding, not a rule worth sizing to.
Every base rate on the Catalyst Calendar carries its n on screen for this reason. A +41% median from 19 episodes and a +15% median from 47 are different kinds of claim, and a calendar that displays only the first number is lying by omission.