How to Measure Post-Offering Price Performance Without Fooling Yourself

The short version

"How did the stock trade after its last few offerings?" is a reasonable historical question. It is also easy to answer badly. Small differences in how you pick the start date, which price you measure from, and which offerings you include can produce very different numbers, and a handful of observations can look like a pattern when it is noise.

This guide sets out a careful, descriptive method. It measures what happened after past offerings. It does not, and cannot, tell you what will happen after the next one. Every number below is hypothetical and exists only to show the arithmetic.

Step 1: Pick the event date precisely

An offering has several dates, and they are often different days:

Choose one definition and use it for every event. Announcement or pricing usually captures the market's first reaction. Then check the time of day: a deal announced at 8:00 PM ET on Monday was first tradable on Tuesday. If you measure from Monday's close, the reaction lands in day 1; if you measure from Tuesday's close, you have already skipped it.

Step 2: Define return windows

Measure the return from the reference price to the close a fixed number of trading days later. Common windows are 1, 7, 30 and 90 trading days (or calendar days, but be consistent).

Return over window = (Close on day N ÷ Reference close) − 1

Hypothetical Company X announced a registered direct offering after the close on a day it closed at $2.00. Its later closes:

All figures hypothetical.

Step 3: Adjust for splits and benchmark the market

Splits. If Company X did a reverse split inside the window, raw closes either side of it are on different bases. Use split-adjusted prices, or multiply pre-split prices by the ratio. A 1-for-10 split turns a $0.20 pre-split close into a $2.00 equivalent.

Benchmark. A stock that fell 10% in a month when its sector index fell 12% did better than its peers in that month. Subtract a benchmark return to get an abnormal (excess) return:

Abnormal return = Stock return − Benchmark return over the same window

For Company X's day-30 window, if a small-cap biotech index returned −8.0%, the abnormal return is −25.0% − (−8.0%) = −17.0%. Pick a benchmark that resembles the company (size and sector), and state which one you used.

Step 4: Aggregate carefully

With several offerings, report both the median and the mean, plus the count.

Hypothetical Company X day-30 returns across six offerings: −25%, −40%, −10%, +60%, −18%, −22%.

• Mean: (−25 − 40 − 10 + 60 − 18 − 22) ÷ 6 = −55 ÷ 6 = −9.2%
• Median: sort to −40, −25, −22, −18, −10, +60; the middle two are −22 and −18, so the median is −20.0%
• Count: 6

The one +60% observation pulls the mean far from the typical outcome. The median describes the middle event better; the mean tells you about the total including outliers. Reporting only one hides that difference.

The caveats that matter most

Sample size

A company that has done three offerings gives you three observations. They tell you what happened three times. Small samples are dominated by the specific circumstances of each deal: the news that accompanied it, market conditions that week, whether the offering included warrants. Always print the count beside any average.

Survivorship bias

If you study "companies that did offerings", and you only look at companies still trading today, you have excluded every company that was delisted (recorded by a Form 25), acquired or went dark afterwards. Those exclusions are not random. A study built only on survivors can describe a different world from the one that existed at the time of each offering.

Overlapping events

Companies that raise often may have a second offering inside the first one's 90-day window. The day-90 return of the first then includes the second deal's effect. Flag overlapping windows, or limit the analysis to windows with no other financing.

Selection and definition

An ATM offering has no single pricing date, so it does not fit an event-study frame the way a priced deal does. Mixing ATM program announcements with registered directs and underwritten offerings blends different events. Group by offering type, and say which types you included.

How to check a company on Signal8

• The Signal8 MCP server includes a post-offering performance tool that, for companies with an analyzed dilution history, returns each recorded offering's performance at 1, 7, 30 and 90 days along with the average, median and event count. An empty result means no offering performance was measured for that company, not that offerings had no effect.
• The developer API docs describe filing-level data you can use to build your own event windows.
• Offering filings appear on the live SEC filings feed, useful for confirming event dates and times.
• Company pages, for example AAPL's Dilution tab, list recent offerings where coverage exists; dilution analysis focuses on small caps.

FAQ

What date should I use to measure an offering's price impact?

Use the moment the information first became public, usually the announcement or pricing, and measure from the last close before it. Check the time of day: an after-hours announcement is first tradable the next session. Whatever definition you choose, apply it to every event so the results are comparable.

Why compare against a benchmark index?

Because the whole market or sector may have moved during the window. Subtracting a comparable index return isolates the part of the move that differed from peers. Without it, an offering made during a broad sell-off can look worse than it was, and one made during a rally can look better.

How many offerings do I need for a meaningful average?

There is no fixed threshold, but single-digit counts describe specific events, not a reliable tendency. Report the count, the median and the mean together, and check how much one extreme observation moves the mean. Treat any small-sample summary as history about those particular deals.

What is survivorship bias in offering studies?

It is the distortion that comes from studying only companies still listed today. Companies that were delisted, acquired or stopped reporting after their offerings drop out of the data, and they may have had different outcomes. Including them, using point-in-time data from the event date, gives a more complete historical picture.