In a securities-fraud case, damages are measured by isolating the share-price movement caused by a company’s misleading statement or its correction from everything else moving the stock that day — and the workhorse tool for that isolation is the event study, a regression-based statistical technique that strips out the market-wide and industry-wide portion of a return so the residual reflects only company-specific news. When investors claim they overpaid for a stock because management hid or distorted material facts, the financial question is not “how far did the price fall?” It is “how much of that fall is attributable to the truth coming out, as opposed to a bad day for the whole sector or the broader market?” The event study exists to answer that narrower question with discipline rather than impression.
I work as the financial-damages and economics expert in this kind of dispute, not the securities lawyer. Whether a statement was material, whether anyone acted with the required state of mind, and whether the legal elements of a 10b-5 claim are satisfied are questions for counsel and ultimately the court. My lane is the measurement: building a defensible model that quantifies how much of a price decline the evidence ties to the alleged misrepresentation, and separating that from the many other things that move a stock. This article walks through how that measurement works.
What an Event Study Actually Measures
A publicly traded stock moves every day for reasons that have nothing to do with the specific company. Interest-rate news, jobs reports, currency swings, a war scare, a sector-wide commodity shock, an analyst downgrade of a competitor — all of it pushes prices around. On any given day, a company’s observed return is a blend of those broad forces plus whatever is specific to that one firm.
An event study is the statistical procedure that pulls those two pieces apart. It estimates what the stock should have returned given how the overall market and the relevant industry performed, then compares that expectation against what the stock actually did. The gap between the two — the part the broad forces cannot explain — is the abnormal return, also called the residual or firm-specific return. That residual is the number that matters in a fraud case, because it represents the price reaction the market had to the company itself, net of the noise.
Put simply: if a stock dropped 9% on the day a company admitted it had been overstating revenue, but the entire market dropped 4% that same day on unrelated macroeconomic news, the event study’s job is to determine that something closer to 5% — not the headline 9% — is the firm-specific reaction worth analyzing. The whole exercise is about not crediting the defendant’s disclosure with movement the market would have produced anyway.
Why this matters for both sides
Event studies are not a plaintiff’s tool or a defendant’s tool. Both sides use them, and they do work at two distinct stages of a securities class action:
- At class certification, the question is whether the stock traded in a market efficient enough that a misstatement would have been absorbed into the price — the premise that lets a class of investors proceed together rather than each proving individual reliance. A defendant can now try to show, with its own study, that the alleged misstatement had no price impact at all.
- At the merits stage, the question shifts to causation and dollars: how much of the loss the corrective disclosure actually caused, and what per-share inflation existed across the period.
Because the same technique drives both stages, the choices an expert makes early tend to echo throughout the case. A model that looks clean at certification can be attacked at the merits stage if its assumptions do not hold up.
The Efficient-Market Premise Underneath It All
The event study rests on a specific idea about how stock prices work: in a reasonably efficient market, publicly available information gets reflected in the price quickly, so the price at any moment embeds what the market collectively knows. This is the analytical foundation for the “fraud-on-the-market” theory that allows securities classes to form. If the price reliably incorporates public information, then a public lie distorts the price for everyone who trades — and an investor does not have to prove he personally read and believed the false statement. He relied on the integrity of a price that the lie corrupted.
Efficiency here is a narrower claim than people assume. It does not mean the market is always right or that prices reflect some true intrinsic value. Bubbles happen. A stock can be wildly overpriced and still trade in an efficient market in the relevant sense — efficient meaning the price moves promptly when new public information arrives, not that the price is correct. The fraud-on-the-market premise depends on that responsiveness, not on accuracy.
Courts assess efficiency through a familiar set of practical factors: how many analysts follow the stock, how heavily it trades, whether there are active market makers, whether the issuer qualifies for streamlined securities registration, and — the factor that tends to carry the most weight — whether company-specific news produces measurable, prompt price reactions. That last one is where the event study earns its keep. Demonstrating that material disclosures historically moved the stock is direct evidence that the market for that security processes information the way the theory requires. Large-cap stocks on major exchanges usually get the benefit of a presumption; thinly traded shares, certain bonds, preferred stock, and derivatives require more careful showing because efficiency cannot be assumed.
The Market Model: How the Regression Works
The mechanical heart of an event study is a regression called the market model. It is a straightforward statistical relationship — not the more contested capital-asset-pricing model, despite some shared vocabulary — and it simply states that a stock’s return tends to move in a roughly linear relationship with the return on a broad index.
The expert estimates that relationship over an estimation window: a stretch of trading days before (or surrounding but not including) the event in question. Across those days, the regression produces two key figures:
- An intercept — roughly, what the stock tends to do when the index is flat.
- A slope — how sensitive the stock is to market moves. A slope above 1.0 means the stock amplifies the market; a slope below 1.0 means it dampens it.
Choosing the estimation window is itself a judgment call with real consequences. A longer window — often something on the order of 100 to 200 trading days — gives statistically tighter estimates because it uses more data. But too long a window risks capturing a period when the company’s risk profile was genuinely different (say, before it took on a large amount of debt), which makes the relationship unstable. A shorter window adapts faster to a changing company but is noisier. There is no universally correct answer, only a reasoned one. And the estimation window must never include the event dates themselves; folding the event into the baseline would quietly bury the very effect the study is trying to detect.
From the regression to the abnormal return
Once the relationship is estimated, the expert turns to the event window — the day (or sometimes a tight band of hours or days) when the disclosure hit the market. The model predicts the stock’s “normal” return for that window using the actual index return and the estimated intercept and slope. The abnormal return is the difference between what the stock actually did and that predicted normal return.
A subtle point trips up casual readers: the abnormal return can carry a different sign than the raw move. A stock can close down on a day when its firm-specific reaction was actually positive, simply because the broad market fell harder than the stock did. The raw number tells you almost nothing until the market component is removed. That is precisely the value the event study adds — and precisely why eyeballing a price chart is no substitute for it.
A Hypothetical Illustration of the Mechanics
The following figures are entirely hypothetical and invented to show how the arithmetic works. They are not drawn from any actual case, client, or company.
Imagine a fictional manufacturer whose stock falls on the day it issues a corrective disclosure walking back earlier statements about a key contract. Suppose the estimation-window regression produced an intercept near zero and a slope of 1.20, meaning the stock historically moved about 1.2% for every 1% move in the chosen market index. On the disclosure day, the market index fell 1.0%.
| Component | Value | How it’s derived |
|---|---|---|
| Slope (market sensitivity) | 1.20 | From the estimation-window regression |
| Index return on event day | -1.0% | Observed |
| Predicted “normal” return | -1.2% | 1.20 × (-1.0%), intercept ≈ 0 |
| Actual stock return on event day | -6.0% | Observed |
| Abnormal (firm-specific) return | -4.8% | -6.0% − (-1.2%) |
The stock’s headline drop was 6%, but the market explains only 1.2 points of it. The firm-specific reaction — the piece potentially tied to the disclosure — is about 4.8%. If the company’s pre-disclosure market value was, say, $500 million (again, hypothetical), that 4.8% translates to roughly $24 million of firm-specific value movement on that single day. Whether that movement is statistically significant, and whether all of it is attributable to the alleged fraud rather than something else that surfaced the same day, are the next two questions — and they are where most of the real analysis lives.
Testing Whether the Reaction Is Real: Statistical Significance
An abnormal return of -4.8% looks dramatic, but a number alone proves nothing. The expert has to ask whether a move that size is genuinely outside the range of the stock’s ordinary day-to-day wiggle, or whether it could plausibly be random noise. That is a test of statistical significance, and it compares the abnormal return against the stock’s own historical volatility — the typical scatter of its residuals during the estimation window.
Courts commonly look for significance at the 95% confidence level. But there is a technical trap here that I take seriously, because getting it wrong overstates the strength of a finding. When an event study examines a single firm on a single date — which is the usual securities-litigation situation — the appropriate statistical benchmark is not the familiar one drawn from large samples. A single-event analysis has very few degrees of freedom, which means the threshold the abnormal return must clear to count as significant is considerably higher than the textbook value many people reach for by reflex. An expert who borrows the standard large-sample cutoff makes ordinary moves look statistically meaningful when a stricter, correct standard would not. The hurdle in a single-firm study is genuinely demanding, and a rigorous model respects that.
There is also a defensible question of whether the test should be one-sided or two-sided — that is, whether the analyst is asking “was there any news effect, good or bad?” or the narrower “was there specifically bad news?” Reasonable experts disagree, the academic literature uses both, and courts have generally treated the choice as a question of fact rather than a clean right-or-wrong. What matters is that the choice is reasoned and disclosed, not buried.
Loss Causation and the Problem of Confounding Information
Demonstrating a statistically significant firm-specific decline on the disclosure date is necessary but not sufficient. The harder question — and the one where event studies most often succeed or fail — is loss causation: tying that decline to the corrective disclosure specifically, rather than to other company news that happened to land the same day.
Real disclosure days are messy. A company might announce a restatement at the same time it reports a disappointing quarter, loses a major customer, or sees an analyst slash its rating for unrelated reasons. Each of those is a confounding event, and each one inflates the abnormal return without being part of the alleged fraud. An expert who attributes the entire firm-specific decline to the misrepresentation, when part of it was driven by genuinely separate bad news, has overstated damages — and courts have not hesitated to throw out studies that fail this discipline. Identifying some confounding factors is not enough; the expert has to actually segregate the loss caused by the fraud from the loss caused by everything else on the table.
This is the same disaggregation discipline that underlies sound causation analysis in any economic-damages matter. In a lost-profits case I have to separate the harm caused by the defendant from harm caused by a recession, a new competitor, or the plaintiff’s own missteps. In a securities case the principle is identical — only the instrument changes. The expert who cannot draw that line cleanly cannot reliably support a damages number, in either setting.
There are concrete tools for the disaggregation. Adding an industry index alongside the broad-market index lets the model strip out not just market-wide noise but sector-wide moves — useful, for example, when a whole industry sells off on news that has nothing to do with the defendant. Tightening the event window — sometimes to intraday hours rather than a full close-to-close day — can isolate the at-issue announcement from a separate piece of news that hit earlier in the same session. The defense, predictably, probes the opposite direction: showing that a plaintiff’s study ignored a confounding factor is one of the most effective ways to undercut it. A study that survives those probes is one built knowing they are coming.
From Price Reaction to Per-Share Damages
Establishing a causal, firm-specific price decline gets the analysis to the door of damages but not through it. The recoverable loss in a fraud-on-the-market case generally follows an out-of-pocket measure: the difference between what an investor paid for the stock and what the stock was truly worth absent the fraud. That difference, on a per-share basis, is the inflation — the amount the misrepresentation artificially propped up the price.
Because the alleged distortion can persist over a stretch of time, the per-share inflation usually is not a single number but a schedule that runs across the class period — sometimes called the inflation ribbon. Inflation enters the price when the misleading information takes hold and bleeds out as the truth is revealed, often through a series of partial corrective disclosures rather than one clean reveal. The event-study residuals on each corrective date feed that schedule: each statistically significant, fraud-attributable firm-specific decline marks a point where inflation came out of the stock. An investor’s loss then depends on when he bought and sold relative to that ribbon — buying while the price was inflated and holding through the correction is what produces a recoverable loss.
The statutory damages framework also caps how much price decline is recoverable. Rather than letting losses run to wherever the stock eventually bottoms out, the recoverable decline is generally measured against an average trading price over a defined look-back window of roughly 90 days following the corrective disclosure. The logic is to prevent a plaintiff from claiming the full drop to a panic-driven trough that the market later recovered from — to anchor the loss to a more representative post-disclosure value rather than a single worst-case tick. Translating the price-reaction findings into per-share inflation, applying that cap, and aggregating across a class is detailed quantitative work, and it is squarely the forensic accountant’s and economic-damages expert’s role rather than the lawyer’s.
Robustness: Why a Single Model Is Never Enough
Any event study involves a chain of choices — which market index, whether to add an industry index, how long the estimation window runs, whether returns are computed in simple or continuously compounded form, how outliers are handled. Each choice is defensible, and each is also a place where an opposing expert can apply pressure. The most common rebuttal to an event study is not to deny the technique but to show that its conclusion flips when a reasonable alternative assumption is substituted.
The defense against that is robustness testing: running the model under several reasonable specifications and showing the firm-specific conclusion holds up across them. If the abnormal return on the corrective date stays significant whether you use a broad index or a sector index, a 120-day window or a 90-day one, simple returns or logarithmic ones, then the result is sturdy. If the conclusion only survives under one narrow set of choices, that is a warning sign — and far better to find it in my own workpapers than to have opposing counsel find it on cross-examination. A measured expert treats the base model as a starting point and then deliberately tries to break it.
This connects securities work to the broader economic-damages methodology I apply across cases. The discipline of reasonable certainty — the requirement that a damages number rest on sound method and reliable data rather than speculation — is the same standard whether the instrument is a publicly traded share or a closely held business. Event studies are simply the rigorous, market-data-driven way of meeting that standard when the loss shows up as a movement in a stock price. This kind of analysis is typically billed hourly, at approximately $400 per hour, the Florida market average.
FAQ
What is an event study in plain terms?
It is a statistical method that separates the part of a stock’s price move caused by company-specific news from the part caused by the overall market and the company’s industry. In a fraud case, it isolates how much of a price drop the evidence ties to a misleading statement or its correction, rather than to a bad day for stocks generally. The leftover, firm-specific piece is called the abnormal return, and it is the number the damages analysis is built on.
Does an event study prove securities fraud?
No. An event study measures price reactions and helps quantify causation and damages — it does not establish that a statement was false, material, or made with the required intent. Those are legal questions for counsel and the court. My work is the financial measurement: determining how much firm-specific price movement is attributable to the disclosure, and separating it from unrelated news.
Why can’t you just use the size of the stock’s price drop?
Because the raw drop blends company-specific news with whatever the broader market and the industry did that day. A stock can fall 8% on a day when 5 points of that came from a market-wide selloff and only 3 points reflected the company. Crediting the full drop to the disclosure overstates the harm. The event study exists to strip the market and industry portion out so the firm-specific reaction can be measured honestly.
What are confounding events and why do they matter so much?
Confounding events are other pieces of news that hit the same day as the corrective disclosure — a weak earnings report, a lost contract, an unrelated downgrade. They inflate the firm-specific price decline without being part of the alleged fraud. If an expert fails to separate the loss caused by the fraud from the loss caused by these other factors, the damages figure is unreliable, and courts have excluded studies on exactly that ground. Disciplined disaggregation is the core of credible loss causation.
How are per-share damages actually calculated?
Damages generally follow an out-of-pocket measure: the difference between the price paid and the stock’s value absent the fraud, expressed as per-share inflation that runs across the class period. The statistically significant, fraud-attributable price declines on corrective dates mark where inflation came out of the stock. A statutory cap then limits the recoverable decline relative to an average trading price over a look-back window of roughly 90 days after the disclosure, so losses are anchored to a representative post-disclosure value rather than a temporary low.
How do I reach you to discuss a securities-damages matter?
You can reach me directly at 954-282-9615 to talk through a potential engagement and whether an event-study damages analysis fits your case. I work as the financial-damages and economics expert alongside your securities counsel, who handles the legal elements; my role is the measurement — building a defensible model of price impact, loss causation, and per-share inflation, and stress-testing it before it ever reaches the other side.
About the Author
Joey Friedman is a CPA, Accredited in Business Valuation (ABV), and forensic accountant who holds a Master of Accounting and a Master of International Business and is a member of the AICPA and the Association of Certified Fraud Examiners. He also holds a Florida real estate license. Beyond those credentials, he has personally owned and operated more than a dozen of his own businesses across industries including marketing, printing, transportation, restaurants, hospitality and entertainment, and event planning — so he evaluates an event-study damages model with both a forensic accountant’s command of the financial mechanics and an operator’s grounded sense of what actually moves a company’s value.
{“@context”:”https://schema.org”,”@type”:”FAQPage”,”mainEntity”:[{“@type”:”Question”,”name”:”What is an event study in plain terms?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”It is a statistical method that separates the part of a stock’s price move caused by company-specific news from the part caused by the overall market and the company’s industry. In a fraud case, it isolates how much of a price drop the evidence ties to a misleading statement or its correction, rather than to a bad day for stocks generally. The leftover, firm-specific piece is called the abnormal return, and it is the number the damages analysis is built on.”}},{“@type”:”Question”,”name”:”Does an event study prove securities fraud?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”No. An event study measures price reactions and helps quantify causation and damages \u2014 it does not establish that a statement was false, material, or made with the required intent. Those are legal questions for counsel and the court. My work is the financial measurement: determining how much firm-specific price movement is attributable to the disclosure, and separating it from unrelated news.”}},{“@type”:”Question”,”name”:”Why can’t you just use the size of the stock’s price drop?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Because the raw drop blends company-specific news with whatever the broader market and the industry did that day. A stock can fall 8% on a day when 5 points of that came from a market-wide selloff and only 3 points reflected the company. Crediting the full drop to the disclosure overstates the harm. The event study exists to strip the market and industry portion out so the firm-specific reaction can be measured honestly.”}},{“@type”:”Question”,”name”:”What are confounding events and why do they matter so much?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Confounding events are other pieces of news that hit the same day as the corrective disclosure \u2014 a weak earnings report, a lost contract, an unrelated downgrade. They inflate the firm-specific price decline without being part of the alleged fraud. If an expert fails to separate the loss caused by the fraud from the loss caused by these other factors, the damages figure is unreliable, and courts have excluded studies on exactly that ground. Disciplined disaggregation is the core of credible loss causation.”}},{“@type”:”Question”,”name”:”How are per-share damages actually calculated?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Damages generally follow an out-of-pocket measure: the difference between the price paid and the stock’s value absent the fraud, expressed as per-share inflation that runs across the class period. The statistically significant, fraud-attributable price declines on corrective dates mark where inflation came out of the stock. A statutory cap then limits the recoverable decline relative to an average trading price over a look-back window of roughly 90 days after the disclosure, so losses are anchored to a representative post-disclosure value rather than a temporary low.”}},{“@type”:”Question”,”name”:”How do I reach you to discuss a securities-damages matter?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”You can reach me directly at 954-282-9615 to talk through a potential engagement and whether an event-study damages analysis fits your case. I work as the financial-damages and economics expert alongside your securities counsel, who handles the legal elements; my role is the measurement \u2014 building a defensible model of price impact, loss causation, and per-share inflation, and stress-testing it before it ever reaches the other side.”}}]}