Defendants Robinhood Markets, Inc., Robinhood Financial LLC and Robinhood Securities, LLC's Memorandum of Points and Authorities in Support of Their Motion to Exclude the Opinions of Dr. Adam Werner on Market Efficiency — In re January 2021 Short Squeeze Trading Litigation (Dkt. 566, S.D. Fla.)
Summary
Defendants Robinhood Markets, Inc., Robinhood Financial LLC and Robinhood Securities, LLC's Memorandum of Points and Authorities in Support of Their Motion to Exclude the Opinions of Dr. Adam Werner on Market Efficiency in In re: January 2021 Short Squeeze Trading Litigation, Case 1:21-md-02989-CMA (Case No. 21-MD-2989-ALTONAGA/DAMIAN, Document 566, 28 pages), in the Southern District of Florida, entered June 7, 2023. Robinhood argues that Dr. Werner's opening-report efficiency opinions are unhelpful because they cover the full year preceding the Class Period, and that his news/no news test is unreliable because it sweeps in stock-price movements with no connection to value-relevant news. The memorandum also attacks the rebuttal-report opinions and requests a hearing; counsel certify conferring with Plaintiffs' counsel on June 2, 2023 without agreement.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Defendants Robinhood Markets, Inc., Robinhood Financial LLC and Robinhood Securities, LLC's Memorandum of Points and… (Dkt. 566).”
Original at archive.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/2023-06-07-alcott-co-v-raphael/
, SHA-256 0fa519b2affe0c15847d0130725f33bce3b507cd4c1e2258f000877b23af6c1d.Full text
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 1 of 28
UNITED STATES DISTRICT COURT
SOUTHERN DISTRICT OF FLORIDA
CASE NO. 21-MD-2989-ALTONAGA/DAMIAN
In re:
JANUARY 2021 SHORT SQUEEZE
TRADING LITIGATION
_____________________________________/
This Document Relates to the Federal Securities Tranche
DEFENDANTS ROBINHOOD MARKETS, INC., ROBINHOOD FINANCIAL LLC AND
ROBINHOOD SECURITIES, LLC’S MEMORANDUM OF POINTS AND
AUTHORITIES IN SUPPORT OF THEIR MOTION TO EXCLUDE THE OPINIONS
OF DR. ADAM WERNER ON MARKET EFFICIENCY
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 2 of 28
TABLE OF CONTENTS
PRELIMINARY STATEMENT .....................................................................................................1
BACKGROUND .............................................................................................................................3
LEGAL STANDARDS ...................................................................................................................5
ARGUMENT ...................................................................................................................................6
I. Dr. Werner’s Opinions Concerning Efficiency in His Opening Report Are
Unhelpful and Unreliable. ........................................................................................6
A. Dr. Werner’s Analysis of the Year Preceding the Class Period Is
Unhelpful. ....................................................................................................7
B. Dr. Werner’s Analysis Is Unhelpful and Unreliable Because It
Sweeps in Stock-Price Movements With No Connection to Value-
Relevant News. ..........................................................................................10
C. Dr. Werner Deliberately Departed from His Methodology in Prior
Cases, Apparently To Achieve a Desired Result. ......................................16
II. Dr. Werner’s Opinions Concerning Efficiency in His Rebuttal Report Are
Also Unhelpful and Unreliable. .............................................................................18
CONCLUSION ..............................................................................................................................20
REQUEST FOR HEARING ..........................................................................................................21
LOCAL RULE 7.1(A)(3) CERTIFICATION ...............................................................................21
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 3 of 28
TABLE OF AUTHORITIES
Cases Page(s)
Alcott Co. v. Raphael,
275 F.2d 551 (5th Cir. 1960) ...................................................................................................15
In re Alibaba Group Holding Ltd. Sec. Litig.,
No. 15-mdl-2631 (S.D.N.Y.) ...................................................................................................14
In re Allstate Corp. Sec. Litig.,
966 F.3d 595 (7th Cir. 2020) ...................................................................................................12
ATSI Commc’ns, Inc. v. Shaar Fund, Ltd.,
493 F.3d 87 (2d Cir. 2007).........................................................................................................7
Basic, Inc. v. Levinson,
485 U.S. 224 (1988) .....................................................................................................4, 7, 8, 11
Bell v. Ascendant Sols., Inc.,
No. 01-CV-166, 2004 WL 1490009 (N.D. Tex. July 1, 2004) ................................................16
Brokop v. Farmland Partners Inc.,
No. 18-cv-2104, 2021 WL 4916240 (D. Colo. July 23, 2021) ................................................10
Broussard v. Maples,
535 F. App’x 825 (11th Cir. 2013) ..........................................................................................15
Cammer v. Bloom,
711 F. Supp. 1264 (D.N.J. 1989) ...............................................................................................4
City of Tuscaloosa v. Harcros Chems., Inc.,
158 F.3d 548 (11th Cir. 1998) ...............................................................................................8, 9
Comcast Corp. v. Behrend,
569 U.S. 27 (2013) .....................................................................................................................8
Cordoves v. Miami-Dade Cnty.,
104 F. Supp. 3d 1350 (S.D. Fla. 2015) ........................................................................16, 17, 18
Daubert v. Merrell Dow Pharms., Inc.,
509 U.S. 579 (1993) ...............................................................................................................5, 6
FindWhat Investor Grp. v. FindWhat.com,
658 F.3d 1282 (11th Cir. 2011) ...............................................................................4, 13, 16, 19
ii
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 4 of 28
FTC v. Simple Health Plans LLC,
No. 18-CV-62593, 2021 WL 810262 (S.D. Fla. Mar. 3, 2021) ...............................................15
Geyer v. NCL (Bahamas) Ltd.,
203 F. Supp. 3d 1212 (S.D. Fla. 2016) ............................................................................6, 7, 10
Halliburton Co. v. Erica P. John Fund, Inc.,
573 U.S. 258 (2014) ...........................................................................................................10, 12
Korsing v. United States,
No. 16-cv-22190, 2017 WL 7794276 (S.D. Fla. Aug. 24, 2017) ............................................11
Krogman v. Sterritt,
202 F.R.D. 467 (N.D. Tex. 2001) ..............................................................................................4
Lee-Bolton v. Koppers Inc.,
319 F.R.D. 346 (N.D. Fla. 2017) ...............................................................................................7
Local 703, I.B. of T. Grocery & Food Emps. Welfare Fund v. Regions Fin. Corp.,
762 F.3d 1248 (11th Cir. 2014) .................................................................................................4
McCorvey v. Baxter Healthcare Corp.,
298 F.3d 1253 (2002) ...........................................................................................................6, 11
In re Montage Tech. Grp. Ltd. Sec. Litig.,
No. 14-CV-722, 2016 WL 1598666 (N.D. Cal. Apr. 21, 2016) ..............................................16
Nat’l Sur. Corp. v. Ga. Power Co.,
No. 2:17-CV-68, 2019 WL 4394403 (N.D. Ga. Sept. 12, 2019) .............................................15
In re Northfield Lab’ys, Inc. Sec. Litig.,
267 F.R.D. 536 (N.D. Ill. 2010) .........................................................................................10, 16
Prosper v. Martin,
989 F.3d 1242 (11th Cir. 2021), cert. denied, 142 S. Ct. 435 (2021) ........................................8
Ramirez v. E.I. DuPont de Nemours & Co.,
No. 8:09-CV-321, 2010 WL 2921619 (M.D. Fla. July 23, 2010) ...........................................21
Rink v. Cheminova, Inc.,
400 F.3d 1286 (11th Cir. 2005) ...........................................................................................6, 16
Schoen v. State Farm Fire & Cas. Co.,
No. 21-CV-264, 2022 WL 16579767 (S.D. Ala. Nov. 1, 2022) ..............................................15
Tershakovec v. Ford Motor Co.,
No. 17-CV-21087, 2021 WL 2592390 (S.D. Fla. May 12, 2021) ...........................................15
iii
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 5 of 28
United States v. Hansen,
262 F.3d 1217 (11th Cir. 2001) ...............................................................................................21
United States v. Reddy,
534 F. App’x 866 (11th Cir. 2013) ........................................................................................8, 9
In re Welding Fume Prods. Liab. Litig.,
No. 1:03-CV-17000, 2005 WL 1868046 (N.D. Ohio Aug. 8, 2005) .................................12, 19
In re Zantac (Ranitidine) Prods. Liab. Litig.,
No. 20-MD-2924, 2022 WL 17480906 (S.D. Fla. Dec. 6, 2022) ................................15, 16, 18
Statutes & Rules
Fed. R. Evid. 702 ...............................................................................................................1, 5, 6, 12
Local Rule 7.1(a)(3) .......................................................................................................................21
Local Rule 7.1(b)(2).......................................................................................................................21
Other Authorities
Fischel, Use of Modern Finance Theory in Securities Fraud Cases Involving
Actively Traded Securities,
38 Bus. Law. 1 (1982)..............................................................................................................11
iv
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 6 of 28
PRELIMINARY STATEMENT
Plaintiffs rely on two reports from their expert, Adam Werner, in their motion for
class certification. In the opening report, Dr. Werner concludes that the markets for certain
Affected Stocks were efficient for the year leading up to the Class Period, and in the rebuttal, he
reaches the same conclusion for the weeks leading up to the Class Period. But in reaching those
conclusions, Dr. Werner creates entirely new methodologies out of whole cloth that are
inconsistent with the academic literature and his own approach as an expert in prior cases. He
also analyzes the wrong legal standard. Dr. Werner’s opinions regarding market efficiency are
thus unreliable and unhelpful, and should be excluded under FRE 702 and Daubert.
To begin, Dr. Werner provides in his opening report a series of opinions
concerning market efficiency that should be excluded. First, Dr. Werner’s efficiency opinions in
his opening report are unhelpful because they concern the wrong time period, namely the full
year preceding the Class Period. Plaintiffs contend they were injured when they sold certain
stocks during the Class Period. Reliance on the integrity of the market is an element of their
claims. That reliance can be proved on a class-wide basis only if Plaintiffs are entitled to a
presumption of market efficiency at the time they sold their shares (i.e., during the Class Period).
In other words, to certify a class, the evidence must show the markets were efficient during the
Class Period itself. But Dr. Werner never analyzed efficiency during the Class Period. Instead,
he looked only at the entire preceding year. Because even Dr. Werner concedes it would be
inappropriate to assume efficiency during the Class Period based on his conclusions about the
entire preceding year, he has provided no opinions that are helpful in analyzing the relevant
questions for the purpose of establishing reliance. Dr. Werner’s suggested class-wide damages
framework is doomed by the same error. He relies on the premise that the January 27, 2021
closing price (i.e., the closing price the day before the Class Period begins) was the “fair price”
because he says the markets for the Affected Securities were efficient for the year before that
date. But again, an analysis of the full year is incapable of determining whether the closing price
on the last day of that year was the result of an efficient market. (See Part I.A.)
Second, the test for efficiency in Dr. Werner’s opening report—which he calls a
“news/no news” test—is unreliable even aside from his time-frame problem. The purported
purpose of this test is to study whether the stocks at issue behave differently on days when there
is news about the company as compared to days when there is not. The idea is that a difference
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 7 of 28
between “news” and “non-news” days suggests that the news caused the stock-price movements,
which is relevant because a stock is generally deemed efficient if it reacts to new, value-relevant
information. But Dr. Werner’s “news/no news” test improperly sweeps in articles about the
price movements themselves and associated trading volume. This test is entirely circular and
illogical. Dr. Werner uses this overly broad definition of “news” because he misunderstands the
legal standard, which he wrongly believes requires a type of efficiency inconsistent with the
definition used by financial economists. This methodological error is particularly acute here
given the widespread news coverage of price movements for “meme stocks.” (See Part I.B.)
Third, Dr. Werner’s efficiency opinions are also unreliable because, in the middle
of his work on this case, he stopped using the definition of “news” that he has used in every prior
case in favor of his new, results-oriented approach untethered from the academic literature. One
of Robinhood’s experts recreated Dr. Werner’s “news/no news” test using the methodologies
Dr. Werner applied in prior cases, and came to very different results, by which most of the
Affected Stocks did not trade in efficient markets—even for the year before the Class Period.
Dr. Werner’s creation of a new methodology therefore biased his result, providing yet another
reason to conclude that his opinions are unreliable. (See Part I.C.)
In his rebuttal report, Dr. Werner presents a different test—which he calls a
“binomial test”—to study market efficiency for the week before the Class Period (January 21 to
27, 2021). This test is unhelpful and unreliable because it too suffers from fatal methodological
errors. Here, Dr. Werner considers stock-price movements for the stocks at issue only on days
when there was potentially value-relevant news. Because he observes significant stock price
movement in the Affected Stocks on many of those “news days,” he concludes that the markets
must be operating efficiently (i.e., reacting to that news). But he ignores the days when there
was no such news. Without a comparison of the “news” days with the “non-news” days,
Dr. Werner’s binomial test cannot yield reliable conclusions because it is entirely possible that
the prices for the Affected Stocks were moving on all days regardless of whether there was
potentially value-relevant news. That would undermine—rather than support—market
efficiency. And that is exactly what happened here. When one of Robinhood’s experts looked at
both the “news” and “non-news” days, he found no difference—meaning that the binomial test
cannot reliably support a conclusion here of market efficiency. (See Part II.)
Dr. Werner’s market efficiency opinions should be excluded for all these reasons.
2
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 8 of 28
BACKGROUND1
In January 2021, retail investors banded together via online forums to drive up the
price of certain stocks, including by attempting to “squeeze” short investors. (See Ex. 5,
Grenadier Report ¶¶ 24, 39-41.) This spike in retail investor trading activity resulted in extreme
market volatility and dramatic run-ups of the price of dozens of stocks. (Id. ¶ 38.) Nine of those
stocks—AMC, BBBY, BB, EXPR, GME, KOSS, NOK, TR and TRVG—are the subject of this
litigation; Plaintiffs call them the “Affected Stocks.” (See ECF No. 527, Am. Compl. ¶ 1.)
The stock-price movements in the Affected Stocks in January 2021 were
extraordinary. For example, GME’s stock price rose a remarkable 788% in the five trading days
immediately before the Class Period (i.e., through January 27, 2021). (Ex. 5, Grenadier Report
¶ 76.) KOSS’s stock price rose 1,591% over the same five-day period. (Id.) Eight of the nine
stocks at issue (all but TRVG) experienced their single highest five-day return in the previous
twenty-five years. (Id. ¶ 78.) Moreover, between January 21 and January 27, 2021, the total
daily trading volume for the Affected Stocks increased almost fifteen-fold. (See id. ¶¶ 92-96.)
As a result of this unprecedented volatility, Robinhood’s NSCC deposit requirements increased
substantially. (Id. ¶ 54 & fig. 1.) Robinhood thus began imposing limits on trading on its
platform, including increased margin requirements and, ultimately, a position close only
(“PCO”) on certain stocks on January 28, 2021, followed by share purchase limits in following
days. Alleging that these restrictions caused the prices of the Affected Stocks to decline,
Plaintiffs brought suit. The Court allowed Plaintiffs’ Section 9(a)(2) and 10(b) market
manipulation claims to proceed, ruling that Plaintiffs who sold while those restrictions were in
place (i.e., the Class Period) had stated a claim. (ECF No. 503, MTD Order at 15, 33, 47-48.)
Class discovery proceeded, and on February 16, 2023, the parties exchanged
expert reports. Robinhood proffered reports from Steven Grenadier, a professor of financial
economics at Stanford Graduate School of Business who chaired the finance department at
Stanford, and Daniel R. Fischel, an emeritus professor of law and business at University of
Chicago Law and Business Schools and former dean of the Law School. Plaintiffs proffered the
testimony of Dr. Adam Werner. Dr. Werner recently retired as a lecturer in economics at
1
A detailed factual and procedural history is laid out in Robinhood’s Opposition to
Plaintiffs’ Motion for Class Certification (see Opp. at 3-12), but the facts most relevant to this
motion are summarized herein. All defined terms used herein have the same meaning set forth in
Robinhood’s Opposition to Plaintiffs’ Motion for Class Certification.
3
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 9 of 28
California Polytechnic State University, San Luis Obispo. (Ex. 1, Werner Dep. Tr. 8:4-5.)
Dr. Werner has not held the position of Professor at any university and has never published an
article in a peer-reviewed journal of economics or finance. (Id. at 19:10-20:14.) Dr. Werner is
an experienced expert witness for plaintiffs in securities cases and has been hired by the Rosen
Law Firm as an expert in over ten cases. (Id. at 12:18-22.) He has never testified on behalf of a
defendant in a securities class action, nor that a market was inefficient. (Id. at 14:17-18:2.)
In his opening report, Dr. Werner analyzed market efficiency for the year before
the Class Period. (Ex. 8, Werner Report ¶ 1.) To do so, he examined the five Cammer factors
and the three Krogman factors. (See id. ¶¶ 31, 80 (citing Cammer v. Bloom, 711 F. Supp. 1264
(D.N.J. 1989); Krogman v. Sterritt, 202 F.R.D. 467 (N.D. Tex. 2001)).)2 The only one of these
factors, however, that is a “direct test” of market efficiency is the fifth Cammer factor, which
examines “price reaction to new information.” (Id. ¶ 31; Ex. 1, Werner Dep. Tr. 138:5-142:3.)
The other factors are structural and simply measure whether conditions are in place that could
support market efficiency, and without which market efficiency is generally absent. (Ex. 8,
Werner Report ¶ 23; Ex. 5, Grenadier Report ¶ 68.) While a number of the Affected Stocks meet
structural conditions for efficiency (such as trading volume and analyst coverage), and may trade
efficiently in normal times, that does not mean the Affected Stocks traded efficiently during the
“meme stock” events immediately preceding and during the Class Period. (Ex. 5, Grenadier
Report ¶¶ 63-72; Ex. 6, Grenadier Rebuttal ¶¶ 29-33.) Efficiency at a particular time can be
determined by a direct test of price reaction to new information (Cammer factor 5).
Dr. Werner acknowledges that Cammer factor 5 is “more informative than the
other factors in determining market efficiency.” (Ex. 1, Werner Dep. Tr. 142:4-143:17.) In
purporting to analyze that factor, Dr. Werner conducted what he calls a “news/no news” test.
(Ex. 8, Werner Report ¶ 68.) He admits he conducted this test in a manner he has never done
before. (Ex. 1, Werner Dep. Tr. 173:16-175:19.) As a first step, he ran a regression to find out
2
The Eleventh Circuit has declined to require district courts to use the Cammer factors or
any other set of factors. See Local 703, I.B. of T. Grocery & Food Emps. Welfare Fund v.
Regions Fin. Corp., 762 F.3d 1248, 1254-56 (11th Cir. 2014). The court has explained that
under the efficient market hypothesis, “in an open and developed securities market, the price of a
company’s stock is determined by the available material information regarding the company and
its business.” FindWhat Investor Grp. v. FindWhat.com, 658 F.3d 1282, 1310 (11th Cir. 2011)
(quoting Basic, Inc. v. Levinson, 485 U.S. 224, 241 (1988)).
4
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 10 of 28
“how the Affected Companies’ stock prices typically behaved in relation to the overall stock
market and industry-specific factors.” (Ex. 8, Werner Report ¶ 56.) Based on that regression,
Dr. Werner identified days on which the residual price movements for each stock were
statistically significantly different from what would be expected for that company after
controlling for market and industry factors. (Id. ¶ 65.) Dr. Werner then looked at the number of
“news releases about a subject company that can be identified through the news aggregating
database, Factiva.” (Id. ¶ 68.) Dr. Werner dubbed the 10% of days with the most such articles
“high information flow days”; the remainder, “low information flow days.” (Id. ¶¶ 70-71.)
Dr. Werner then compared the proportion of days with statistically significant returns in the
“high information flow” group to the proportion of such days in the “low information flow”
group. (Id. ¶ 75.) Based on his unreliable analysis, he concluded that the market was efficient
for seven of the nine Affected Stocks (all but KOSS and TR). (Ex. 1, Werner Dep. Tr. 46:5-16.)
On March 28, 2023, the parties exchanged rebuttal reports. In his rebuttal,
Dr. Werner analyzed efficiency for the three weeks before the Class Period. (Ex. 9, Werner
Rebuttal ¶ 17.) He again examined the Cammer/Krogman factors. (See id.) In analyzing
Cammer factor 5, he now borrowed, for his definition of news, the potentially value-relevant
news identified by Defendants’ experts, and performed a “binomial test” that considered only the
days with news in the final week before the Class Period. (Id. ¶ 48-49.) He ignored the days
without news. (Id. ¶ 49-50.) Dr. Werner concluded on the basis of this test that six of the seven
remaining Affected Stocks were efficient during the three weeks before the Class Period.
Dr. Werner did not examine efficiency during the Class Period in either of his
reports. (Ex. 1, Werner Dep. Tr. 31:11-17.) This was not a decision Dr. Werner made; he
testified that the Rosen Law Firm instructed him to do this. (Id. at 29:23-31:23.) The two tests
at issue on this Daubert motion—the “news/no news” test from Dr. Werner’s opening report, and
the binomial test from the rebuttal report—are the only direct tests of efficiency (i.e., tests under
Cammer factor 5) that Dr. Werner conducted for this case.
On April 28, 2023, Plaintiffs moved for class certification (ECF No. 559), relying
on Dr. Werner’s opinion that the Affected Stocks traded in efficient markets. (Id. at 17-21.)
LEGAL STANDARDS
Federal Rule of Evidence 702, as explained by Daubert v. Merrell Dow
Pharmaceuticals, Inc., 509 U.S. 579 (1993), and its progeny, requires courts to “act as
5
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 11 of 28
‘gatekeepers’ which admit expert testimony only if it is both reliable and relevant. District
courts are charged with this gatekeeping function to ensure that speculative, unreliable expert
testimony does not reach the jury under the mantle of reliability that accompanies the appellation
‘expert testimony.’” Rink v. Cheminova, Inc., 400 F.3d 1286, 1291 (11th Cir. 2005) (citing
Daubert, 509 U.S. at 589; McCorvey v. Baxter Healthcare Corp., 298 F.3d 1253 (2002)). The
Eleventh Circuit has developed three criteria for assessing the demands of Daubert:
“(1) the expert is qualified to testify competently regarding the
matters he intends to address; (2) the methodology by which the
expert reaches his conclusions is sufficiently reliable as determined
by the sort of inquiry mandated in Daubert; and (3) the testimony
assists the trier of fact, through the application of scientific,
technical, or specialized expertise, to understand the evidence or to
determine a fact in issue. The Eleventh Circuit refers to these
requirements as the ‘qualifications,’ ‘reliability,’ and ‘helpfulness’
prongs, respectively.”
Geyer v. NCL (Bahamas) Ltd., 203 F. Supp. 3d 1212, 1215 (S.D. Fla. 2016) (citations omitted).
“The party offering the expert has the burden of satisfying each of these three elements by a
preponderance of the evidence.” Rink, 400 F.3d at 1292.
ARGUMENT
Under FRE 702 and Daubert, the Court should exclude Dr. Werner’s opinions on
market efficiency, including the “news/no news” test in his opening report (see infra Part I) and
the binomial test in his rebuttal (see infra Part II).
I. DR. WERNER’S OPINIONS CONCERNING EFFICIENCY IN HIS OPENING
REPORT ARE UNHELPFUL AND UNRELIABLE.
Dr. Werner’s news/no news test from his opening report is inadmissible for three
reasons. First, Dr. Werner does not study efficiency during the Class Period. Instead, he studies
only the full year before the Class Period. But, as Plaintiffs themselves note, efficiency in the
year before the Class Period is not at issue in this case and Dr. Werner concedes that his
conclusion that markets were efficient during the year preceding the Class Period cannot
establish efficiency during the Class Period. Moreover, even looking at the year he studied, the
law clearly prohibits commingling of data that can mask the relevant issues. Here, there were
extreme, unprecedented trading frenzies in the Affected Stocks in the weeks immediately
preceding the Class Period. Analyzing a full year of data—even if relevant—clearly masks this
atypical activity most proximate to the relevant time period and constitutes just the sort of
6
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 12 of 28
misleading commingling that courts have found unreliable under Daubert. (See infra Part I.A.)
Second, based on a fundamental misunderstanding of the meaning of efficiency,
Dr. Werner’s definition of “news” sweeps in news that has no bearing on the value of the stock.
That is fundamentally at odds with the academic literature and the approach Dr. Werner himself
has applied in other cases. (See infra Part I.B.) Third, until this case—indeed, until partway
through his work on this case—Dr. Werner had never used this methodology before, and the
methodologies he used in prior cases yield vastly different results. The inference is unmistakable
that Dr. Werner deliberately chose the methodology he used in this case to arrive at the
conclusion of efficiency that Plaintiffs need for class certification. That unscientific and results-
oriented process renders his methodology unreliable and inadmissible. (See infra Part I.C.)
A. Dr. Werner’s Analysis of the Year Preceding the Class Period Is Unhelpful.
Most of Dr. Werner’s opening report is dedicated to analyzing efficiency of the
markets for the Affected Stocks in the year preceding the Class Period. (Ex. 8, Werner Report
¶¶ 21-90.) But, as Dr. Werner admits, market efficiency is not a static condition and the fact that
a market may have been efficient during an entire preceding year does nothing to establish
efficiency within the Class Period. Indeed, this case is the perfect example of how market
conditions can and do change, with the unprecedented flurry of trading activity in the weeks
immediately preceding the Class Period leading to stock prices so extreme that a number of the
relevant companies had to issue statements cautioning that the skyrocketing prices could not be
explained by any news relevant to the company’s value. Dr. Werner’s conclusion therefore does
not “assist[] the trier of fact . . . to determine a fact in issue.” Geyer, 203 F. Supp. 3d at 1215.
Dr. Werner’s opinion as to efficiency for the year preceding the Class Period is therefore
unhelpful and inadmissible. See Lee-Bolton v. Koppers Inc., 319 F.R.D. 346, 376 (N.D. Fla.
2017) (excluding expert’s testimony that “d[id] not provide information about” the pertinent
factual questions and was therefore “unhelpful to the class certification issues in th[e] case”).
1. Werner’s Analysis of Efficiency for the Year Preceding the Class
Period Does Not Tend To Prove Reliance Within the Class Period.
Plaintiffs assert market manipulation claims on a class-wide basis. (ECF No. 503,
MTD Order at 1.) Reliance on the integrity of the market is an element of those claims. (Id. at
38 (citing ATSI Commc’ns, Inc. v. Shaar Fund, Ltd., 493 F.3d 87, 101 (2d Cir. 2007)).) The
Supreme Court held in Basic Inc. v. Levinson, 485 U.S. 224 (1988), that securities plaintiffs may
establish reliance on a class-wide basis by showing that the market was efficient at the time they
7
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 13 of 28
transacted—in other words, during the Class Period. Id. at 247.
Dr. Werner does not even attempt to answer that question. Instead, he analyzes
efficiency outside the Class Period. (Ex. 8, Werner Report ¶ 1 (analyzing efficiency for the year
before the Class Period); Ex. 9, Werner Rebuttal ¶ 17 (analyzing efficiency for the three weeks
before the Class Period).) But as Plaintiffs concede, reliance outside the Class Period is not an
element of Plaintiffs’ claims. (ECF No. 559, Mot. to Certify at 4, 15.) And not even Dr. Werner
contends that the factfinder should infer from his analysis of the year before the Class Period that
the markets for the Affected Stocks were efficient during the Class Period. To the contrary,
Dr. Werner expressly opines that his analysis of the time before the Class Period does not apply
to the Class Period. (Ex. 8; 9, Werner Report ¶ 1 n.1; Werner Rebuttal ¶ 17).
Because Dr. Werner’s analysis as to efficiency answers the wrong question by
addressing the wrong time period, his conclusion as to efficiency is unhelpful and must be
excluded. See Prosper v. Martin, 989 F.3d 1242, 1247-50 (11th Cir. 2021) (affirming exclusion
of expert’s testimony on a factual issue that “was not relevant”), cert. denied, 142 S. Ct. 435
(2021); United States v. Reddy, 534 F. App’x 866, 874-75 (11th Cir. 2013) (excluding portion of
expert opinion that analyzed period outside indictment); City of Tuscaloosa v. Harcros Chems.,
Inc., 158 F.3d 548 (11th Cir. 1998) (excluding portion of opinion based on inapplicable data set).
2. Werner’s Analysis of Efficiency for the Full Year Prior to the Class
Period Also Does Not Help the Fact-Finder Determine Damages.
Under Comcast Corp. v. Behrend, 569 U.S. 27 (2013), Plaintiffs also must proffer
a methodology to calculate damages on a class-wide basis. This is distinct from the reliance
element. Here too, Plaintiffs rely on Dr. Werner’s opinions regarding market efficiency, which
derive from his “news/no news” test. (Ex. 9, Werner Rebuttal ¶ 115; Ex. 8, Werner Report
¶¶ 92-99.) Dr. Werner’s proposed framework for a damages methodology would use the closing
price on January 27, 2021, as the “fair price” in the but-for world (i.e., the price at which each
stock would have traded but for Robinhood’s alleged misconduct). But as further explained in
Robinhood’s Opposition, for the January 27, 2021 closing price to be even a starting point to
determine the but-for price, that price would have to reflect the price in an efficient market. (See
Opp. at 27-28.) Just as his “news/no news” test cannot establish market efficiency during the
Class Period for the reasons set forth in the preceding section, the test also cannot demonstrate
that the closing price on January 27, 2021, reflected the price in an efficient market.
As an initial matter, Dr. Werner specifically excluded January 27, 2021, from his
8
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 14 of 28
regression model due to the “potentially abnormal returns” on that day. (Ex. 8, Werner Report
¶ 63.) He thus concedes that the returns on January 27, 2021, are so potentially aberrational that
they must be excluded from his model, yet argues an analysis of 365 other days somehow
confirms the efficiency of the price on January 27, 2021. That makes no sense.
Even setting this admission aside, Dr. Werner’s analysis is unreliable because it
commingles a relevant subset of data with a larger, more general data set. Multiple Eleventh
Circuit cases have held that this requires exclusion of the general analysis. In City of Tuscaloosa
v. Harcros Chemicals, Inc., the court excluded the opinion of an expert who “commingled”
Florida and Alabama data where the allegations related to activity only in Alabama. 158 F.3d at
566-67. Reasoning that “[i]ncluding [such] data . . . skews any cumulative measurements,” the
court found the analysis unreliable. Id. Likewise, in United States v. Reddy, the expert’s period
of study covered both the indictment period and several months thereafter. 534 F. App’x at 872,
874. Noting that “any portion [of an expert’s opinion] that does not meet Daubert standards is
not permitted,” the court held that the portion of the expert’s analysis that “fell outside of the
indictment period” was properly excluded. Id. at 871, 874-75.
Such improper “commingling” is precisely what Dr. Werner has done here. Even
divorced from the specific facts of this case, it is obvious that an analysis of efficiency across the
366 days ending January 27, 2021, cannot itself reliably determine whether the markets for these
stocks were efficient on that single day, because it improperly commingles a general data set
with a narrower one. By analogy, a study showing that demand for Miami Heat jerseys over a
full year was no higher than demand for jerseys of other NBA teams would be unhelpful in
determining whether demand for Miami Heat jerseys was higher during the NBA finals. Both
could easily be true. Indeed, as Dr. Werner conceded, efficiency is not a static condition of
markets, and may change over time. (See, e.g., Ex. 2, FXCM Dep. Tr. 67:22-68:3 (agreeing that
it is “possible for a market to be efficient at one point in time but not another”); id. at 141:19-
143:5 (discussing Werner’s decision to conduct distinct analyses of different sub-periods because
“the stock traded differently before and after” a given date).)
That is particularly obvious here. As the Court has already detailed in prior
rulings, the period immediately before the Class Period is starkly different from the preceding
months. (See, e.g., ECF No. 503, MTD Order at 5-8.) Trading volume in the Affected Stock
skyrocketed in January 2021. (Compare Ex. 8, Werner Report ¶¶ 32-34 with Ex. 9, Werner
9
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 15 of 28
Rebuttal ¶¶ 19-21.) So did prices. (Ex. 5, Grenadier Report ¶ 80 (noting that a portfolio
comprised of the nine Affected Stocks held for the week before the Class Period outperformed
one million randomly selected analogous portfolios simulated from the last 25 years).)
Widespread news commentary at the time—including commentary cited by Dr. Werner—
expressly discussed the “unprecedented” behavior of the market. (See, e.g., Ex. 24.)
By failing to separate out the unprecedented time immediately before the Class
Period from the entire preceding year, Dr. Werner commingles data sets with noticeably different
characteristics, from which different conclusions about efficiency could naturally follow. The
Court cannot conclude that a market was efficient where the evidence indicates that there is such
a substantial variation within the period analyzed. See, e.g., In re Northfield Lab’ys, Inc. Sec.
Litig., 267 F.R.D. 536, 548-49 (N.D. Ill. 2010) (refusing to certify a class where the evidence
showed that the market was efficient for only part of the period analyzed); Brokop v. Farmland
Partners Inc., No. 18-cv-2104, 2021 WL 4916240, at *17-18 (D. Colo. July 23, 2021), report
and recommendation adopted in relevant part, 2021 WL 4913970, at *16 (D. Colo. Sept. 30,
2021) (refusing to certify a class for the period during which efficiency was not shown).
Dr. Werner’s analysis of efficiency therefore does not “assist[] the trier of fact . . . to determine a
fact in issue,” Geyer, 203 F. Supp. 3d at 1215, and the Court should exclude it.
B. Dr. Werner’s Analysis Is Unhelpful and Unreliable Because It Sweeps in
Stock-Price Movements With No Connection to Value-Relevant News.
Even aside from the time frame problem discussed above, Dr. Werner’s “news/no
news” test fails to reliably establish market efficiency because Dr. Werner defines “news” so
broadly as to include articles (such as ones about the “meme stock” price movements
themselves) with no connection to the value of the company.
1. Dr. Werner Did Not Apply the Academic Understanding of
Efficiency.
At deposition, Dr. Werner testified on multiple occasions that his report does not
apply the academic definition of efficiency widely adopted by his peers and colleagues. (See,
e.g., Ex. 1, Werner Dep. Tr. at 66:8-69:5, 78:3-80:14, 82:4-21.) Instead, Dr. Werner tries his
hand at interpreting Supreme Court precedent—even though he has no legal training
whatsoever—and concludes (erroneously) that the Supreme Court articulated a new, more
“plaintiff friendly” standard for efficiency in Halliburton Co. v. Erica P. John Fund, Inc., 573
U.S. 258, 272 (2014) (Halliburton II). That is simply not true, and Dr. Werner has deviated from
10
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 16 of 28
the area of his expertise and is offering an unhelpful and unreliable opinion.
In his opening report, Dr. Werner explains that an efficient market is one in which
“all publicly available information is incorporated in the security price and any new information
that impacts the economic outlook of the firm gets quickly reflected in its security price.” (Ex. 8,
Werner Report ¶ 21.) That definition of an efficient market is accurate and comports with the
understanding of academics in the field. For example, Prof. Andrei Shleifer at Harvard
University has observed that, in an efficient market, “[w]hen investors learn something about
fundamental values of securities, they quickly respond to the new information,” and “prices
should not react to changes in demand or supply of a security that are not accompanied by news
about its fundamental value,” i.e., “the net present value of its future cash flows.” (Ex. 12, Dep.
Ex. 179, Inefficient Markets 2, 5.) Dr. Werner acknowledges that Prof. Shleifer is a “leading
authority on corporate finance,” and Prof. Shleifer was the first authority to whom Dr. Werner
looked when asked to define a financial concept. (Ex. 1, Werner Dep. Tr. 265:13-267:2.)
But rather than apply this well-accepted academic definition of efficiency,
Dr. Werner claims that the Supreme Court in Halliburton II changed the rules for securities
litigation and adopted a definition of efficient markets that simply requires that movements in
stock price be explained by any news about the Company, regardless of whether that news is
value relevant. (Id. at 65:5-68:6, 165:6-167:5.) As a result of that mistaken premise (encouraged
by his discussion with plaintiffs’ lawyers), Dr. Werner has eschewed academic literature in favor
of what he—as an economist, and not a lawyer—believes that case law requires of him. That
plainly does not pass muster under Daubert. Daubert is intended to root out junk science, not to
encourage it. See McCorvey, 298 F.3d at 1256 (noting that Daubert serves “to ensure that
speculative, unreliable expert testimony does not reach” the finder of fact); Korsing v. United
States, No. 16-cv-22190, 2017 WL 7794276, at *9 (S.D. Fla. Aug. 24, 2017) (rejecting an
expert’s testimony based on Daubert’s concern with “junk science”).
In fact, far from rejecting the definition of efficiency in the academic literature,
Halliburton II reaffirmed reliance on that definition. In Basic, the Court explained that “[r]ecent
empirical studies have tended to confirm Congress’ premise that the market price of shares
traded on well-developed markets reflects all publicly available information,” citing an article by
Prof. Fischel. Basic, 485 U.S. at 246 & n.24 (citing Fischel, Use of Modern Finance Theory in
Securities Fraud Cases Involving Actively Traded Securities, 38 Bus. Law. 1 (1982)). More than
11
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 17 of 28
thirty years later, the Halliburton II Court reviewed the “academic debates” about market
efficiency that the parties and amici curiae brought to the Court’s attention. The Court
acknowledged the occasional deviations from efficiency that are documented by academics in
the field but adhered to the value-relevant news based definition of market efficiency grounded
in the literature. Halliburton II, 573 U.S. at 270-72. As the Seventh Circuit explained, “[t]he
efficient capital markets hypothesis has been criticized, but in Halliburton II, the Supreme Court
rejected arguments to overrule Basic. Whatever the empirical merits of critiques of the efficient
market hypothesis, as a matter of law it remains the foundation for fraud-on-the-market claims.”
In re Allstate Corp. Sec. Litig., 966 F.3d 595, 605 n.2 (7th Cir. 2020) (internal citations omitted).
In short, Halliburton II did no more than affirm Basic’s embrace of the financial
principle from the academic literature that, in an efficient market, a company’s stock price
should move in response to news about the value of that company and should remain relatively
stable in the absence of such news. See Halliburton II, 573 U.S. at 269-73; (Pl. Ex. 34, Fischel,
Efficient Capital Markets 907 (“[C]ourts have expressly based their acceptance of the [fraud-on-
the-market] theory not on the legislative history of the securities laws but rather on the academic
support for the efficient markets hypothesis.”).) Dr. Werner’s arm-chair legal analysis that
Halliburton II sets forth a standard of efficiency under which price movements in response to
news that has no bearing on the value of the company, or indeed no news at all, has no basis in
law or academia. (Ex. 1, Werner Dep. Tr. 65:19-68:6, 84:12-85:9, 160:14-161:11.) In applying
a definition of market efficiency that is inconsistent with the academic understanding,
Dr. Werner has thus rendered an opinion that is not scientific opinion testimony admissible under
Daubert and FRE 702. See, e.g., In re Welding Fume Prods. Liab. Litig., No. 1:03-CV-17000,
2005 WL 1868046, at *7 (N.D. Ohio Aug. 8, 2005) (excluding an expert’s opinions that were
“improperly premised on” an incorrect legal standard).
2. Dr. Werner’s News/No News Test Improperly Includes News That Is
Unrelated to the Economic Outlook of the Firm.
Dr. Werner’s erroneous understanding of Halliburton II led him to conduct a
news/no news test that includes “news” unrelated to the economic outlook of the firm. Indeed, it
includes news that is literally only about the stock price movement. A test based on that
defective definition of news can only lead to an unreliable and irrelevant conclusion.
Consistent with the academic definition of efficiency that he himself articulated in
his opening report, Dr. Werner should have limited the “news” he studied to value-relevant news
12
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 18 of 28
(i.e., news about the company’s future cash flows). That is precisely what he did in previous
cases, including in cases that came years after the Halliburton II decision that Dr. Werner
suddenly now determined—halfway through his analysis in this case—somehow changed the
ground rules. (See, e.g., Ex. 11, NIO Rebuttal Report ¶¶ 16, 31, 52 (filtering out non-value-
relevant information before performing the news/no news test); Ex. 26, NIO Dep. Tr. 83:14-21;
see also Ex. 10, Innocoll Report ¶ 84 (stating importance of restricting test to news “that has
potential to change investors’ expectations of future cashflows”).)
But as Dr. Werner testified, he did the opposite here. (Ex. 1, Werner Dep. Tr.
154:8-155:15.) Specifically, Dr. Werner defined high-information-flow days as the 10% of days
with the most news articles about a given company—i.e., every company, by definition, will
have 10% of days count as “high information,” regardless of how many or what type of news
articles those are. (Ex. 8, Werner Report ¶¶ 68, 70.) By including all articles mentioning a
company, Dr. Werner sweeps in news that has nothing to do with the company’s future cash
flows. For example, Dr. Werner’s data set includes articles that mention only the fact that there
was a bubble (see, e.g., Ex. 15), only the existence of an actual or attempted short squeeze (see,
e.g., Ex. 27), or only a change in the stock price (see, e.g., Ex. 17). The data set even includes
articles that expressly noted the lack of value-relevant news. (See, e.g., Ex. 18.) These kinds of
articles should not change an investor’s estimate of future cash flows. (Ex. 6, Grenadier Rebuttal
¶ 12.) As such, the mere correlation of stock-price movements and news about those price
movements proves nothing about efficiency, and indeed price movements in response to such
information would be evidence of inefficiency, not efficiency. See FindWhat, 658 F.3d at 1310
(“A corollary of the efficient market hypothesis is that disclosure of confirmatory information—
or information already known by the market—will not cause a change in stock price.”).
Dr. Werner’s failure to filter out these articles is also refuted by directly relevant
authorities on which he relies. In particular, Dr. Werner cites a 2004 paper by Ferrillo, Dunbar
and Tabak (Ex. 8, Werner Report ¶¶ 73, 76), and testified that he relied heavily on Dr. Tabak’s
approach in compiling his “news/no news” test in this case. (Ex. 1, Werner Dep. Tr. 224:23-
225:13 (“So here I defer to Dr. Tabak.”).) In the paper cited by Dr. Werner, Dr. Tabak and his
co-authors first proposed and defended the concept of the news/no news test. (Ex. 14, Ferrillo et
al. 119-20.) But in doing so, the authors expressly stated that “one would still have to be careful
to . . . remove any stories that exist because they report on a price movement.” (Id.) By his own
13
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 19 of 28
admission, Dr. Werner did not undertake this critical step. (Ex. 1, Werner Dep. Tr. 227:17-22.)
At deposition, Dr. Werner attempted to justify that failure by claiming that
Dr. Tabak no longer requires removal of news that simply reports on price movements. (Ex. 1,
Werner Dep. Tr. 225:7-13.) That is false. The only support Dr. Werner provided for this
proposition was an expert report Dr. Tabak submitted for the Rosen Law Firm in In re Alibaba
Group Holding Ltd. Securities Litigation, No. 15-mdl-2631 (S.D.N.Y.). In that report, Dr. Tabak
conducted a news/no news test based specifically on earnings announcements. (Ex. 7, Alibaba
Report ¶ 39.) This test is methodologically quite different from Dr. Werner’s use of a database
with all news articles, as earnings announcements are intimately related to investor expectations
of a company’s future cash flows. As a complement to his earnings-announcement test,
Dr. Tabak also conducted news-mention-based searches, but specifically excluded from those
searches “recurring pricing and market data.” (Id. Ex. 8a n.7.) In addition, consistent with the
methodology he outlined in Ferrillo et al., Dr. Tabak “reviewed the stories . . . and remov[ed]
any that were solely reporting on [the stock’s] price movements or volume (i.e., instances where
the news was responding to the [stock] price rather than the [stock] price potentially responding
to news) . . . or were deemed to be non-material because they reported data irrelevant to the
value of the Company.” (Id. ¶ 39.) This is a far cry from Dr. Werner’s approach.
Dr. Werner’s failure to filter out stories about stock-price movements, as directed
by Dr. Tabak and his co-authors (and consistent with the remainder of the academic literature), is
especially problematic given the unique facts of this case. The unexplained stock-price
movements at issue were themselves one of the most important news stories of the year. (See,
e.g., Ex. 23.) As a result, stories reporting only on those price movements and/or the existence
of an actual or attempted short squeeze comprise a meaningful portion of the stories in
Dr. Werner’s data set, especially in January 2021. On these unusual facts, his observation that
more news is associated with more stock price movement likely reflects the fact that, as
Dr. Tabak put it, “the news was responding to the [stock] price rather than the [stock] price
potentially responding to news.” (Ex. 7, Alibaba Report ¶ 39.) That reverse direction of
causation—for which Dr. Werner did not test or adjust his results (Ex. 1, Werner Dep. Tr. 64:11-
18)—skews his results and renders any conclusion of efficiency unreliable.
In sum, Dr. Werner’s inability to point to any academics who have applied his
methodology evidences the fundamental deficiencies in that methodology and requires
14
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 20 of 28
exclusion. See FTC v. Simple Health Plans LLC, No. 18-CV-62593, 2021 WL 810262, at *9
(S.D. Fla. Mar. 3, 2021) (excluding expert report that “relie[d] upon a methodology not
previously used in his field”); In re Zantac (Ranitidine) Prods. Liab. Litig., No. 20-MD-2924,
2022 WL 17480906, at *27 & n.26 (S.D. Fla. Dec. 6, 2022) (same). Experts do not submit
reports consistent with Dr. Werner’s understanding of efficiency because that understanding is
unsupported by either law or economics. Because his flawed definition of “news” is designed to
test his flawed interpretation of efficiency, Dr. Werner’s news/no news test should be excluded.
See, e.g., Broussard v. Maples, 535 F. App’x 825, 827 (11th Cir. 2013) (excluding analysis based
on invalid proxy for variable of interest); Tershakovec v. Ford Motor Co., No. 17-CV-21087,
2021 WL 2592390, at *5 (S.D. Fla. May 12, 2021) (same), report and recommendation adopted
in relevant part, 2021 WL 3578011 (S.D. Fla. Aug. 13, 2021).
3. Dr. Werner Failed To Examine the Substance of the News Articles on
Which He Relies.
Dr. Werner’s news/no news test is also unreliable because he did not even bother
to review the news articles underlying his test. (See Ex. 1, Werner Dep. Tr. 232:18-233:15.)
First, because he did not bother to review the articles, Dr. Werner’s data set
includes articles that have nothing to do with the Affected Companies. These include articles
about rapper Princess Nokia (see, e.g., Ex. 19); the restaurant chain Panda Express (see, e.g.,
Ex. 22); the crime blotter in Blackberry Township, Pennsylvania (see, e.g., Ex. 20); and,
inexplicably, seven articles about NFL quarterback Aaron Rodgers’ “journey of self-
actualization.” (See, e.g., Ex. 16 (reporting that Rodgers made a phone call to reporters “from
the parking lot of a suburban Los Angeles Bed Bath & Beyond”); Ex. 1, Werner Dep. Tr. 314:8-
316:10; see also Ex. 4, Fischel Rebuttal ¶ 22 n.33 (listing other examples of non-value-relevant
news in Werner’s data set).) These articles cannot show efficiency of Nokia, Express,
BlackBerry, or any of the six other Affected Stocks, and illustrate the pollution that pervades
Dr. Werner’s data set. An expert’s opinions are no better than the data on which they rely, and
the conclusions that Dr. Werner draws from his low-quality data are equally low-quality and thus
unreliable. See Alcott Co. v. Raphael, 275 F.2d 551, 557 (5th Cir. 1960); Schoen v. State Farm
Fire & Cas. Co., No. 21-CV-264, 2022 WL 16579767, at *8 (S.D. Ala. Nov. 1, 2022) (excluding
expert’s conclusion in light of low quality of data on which it was based); Nat’l Sur. Corp. v. Ga.
Power Co., No. 2:17-CV-68, 2019 WL 4394403, at *5 (N.D. Ga. Sept. 12, 2019) (same).
Second, Dr. Werner also did not bother to consider whether information in those
15
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 21 of 28
articles was unexpected. If information contained in a news article was not new, a response to
that information would not indicate efficiency. See FindWhat, 658 F.3d at 1310 (“[I]nformation
already known by the market[] will not cause a change in the stock price.”). In fact, a delayed
reaction to publicly available news is indicative of inefficiency. (See In re Montage Tech. Grp.
Ltd. Sec. Litig., No. 14-CV-722, 2016 WL 1598666, at *10 (N.D. Cal. Apr. 21, 2016) (reaction
to old information “would actually be evidence of market inefficiency); see also Ex. 21
(reporting on stale news).) This is yet another reason to conclude that Dr. Werner’s analysis is
unreliable and inadmissible. See Montage, 2016 WL 1598666, at *10 (excluding testimony of
expert who concluded market was inefficient without reading news he cited).
C. Dr. Werner Deliberately Departed from His Methodology in Prior Cases,
Apparently To Achieve a Desired Result.
Dr. Werner also failed to follow the same methodology he followed in prior cases.
The only reasonable conclusion is that he did so because applying his standard methodology
would have led to the conclusion that most of the Affected Stocks did not trade in efficient
markets, even over the full-year period he erroneously studied (see supra Part I.A).
“In evaluating the reliability of an expert’s method, . . . a district court may
properly consider whether the expert’s methodology has been contrived to reach a particular
result.” Rink, 400 F.3d at 1293 n.7. “[C]herry-picking of data demonstrates unreliability and
may justify the exclusion of an expert’s testimony.” Zantac, 2022 WL 17480906, at *57. An
expert’s contradiction of his own past practice is particularly salient evidence of such a contrived
methodology. See Cordoves v. Miami-Dade Cnty., 104 F. Supp. 3d 1350, 1364 (S.D. Fla. 2015)
(such a contradiction evidenced a lack of “intellectual rigor”); Zantac, 2022 WL 17480906, at
*132 (an expert’s inconsistency is “evidence of results-driven reasoning, a factor in favor of
exclusion,” especially where all of the expert’s methodological choices “favor[] the same
outcome”). As a result, courts regularly exclude expert testimony where the expert’s
methodology was consciously or artificially chosen to reach their preferred conclusion. See, e.g.,
Zantac, 2022 WL 17480906, at *119, *264 (excluding opinion of an expert whose methodology
was designed to reach a desired conclusion); Northfield, 267 F.R.D. at 548 (same); Bell v.
Ascendant Sols., Inc., No. 01-CV-166, 2004 WL 1490009, at *3 (N.D. Tex. July 1, 2004) (same).
That is exactly what happened here. Prior to submitting his opening report in this
case, Dr. Werner had never used the definition of “news” he used here. (Ex. 1, Werner Dep. Tr.
173:16-174:8.) In past cases, Dr. Werner restricted himself to news of major events at the
16
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 22 of 28
company. When Dr. Werner first embarked on his work in this case, he actually began by using
the approach he had in all of his prior cases. (Id. at 174:6-19, 229:2-235:2.) Partway through his
work on this case, he changed tack to adopt a novel methodology that was not limited to value-
relevant news, but also included every news article that hit on search terms—including ones
(particularly common for “meme stocks” in January 2021) reporting on unexplained stock-price
movements, and even ones about unrelated topics as described above.
Had Dr. Werner used this previous methodology, his results would have looked
very different. To demonstrate the results-oriented nature of Dr. Werner’s departure from his
prior methodology, Prof. Grenadier recreated Dr. Werner’s news/no news test over the year
preceding the Class Period using the methodology (and definition of “news days”) that
Dr. Werner has used in prior cases. (Ex. 6, Grenadier Rebuttal ¶¶ 52-54 & Ex. 1.)3
In five prior cases, Dr. Werner defined “news days” as days on which a company
released earnings or guidance. (Ex. 6, Grenadier Rebuttal Ex. 1A.) If Dr. Werner
had used that methodology here, he would have found a cause-and-effect
relationship for only three of the Affected Stocks. (Id.)
In seven prior cases, Dr. Werner defined “news days” as days on which a
company filed a Form 8-K or 6-K (a “current” report of material events). (Id.
Ex. 1B.) Under that methodology, a cause-and-effect relationship existed for only
two of the Affected Stocks. (Id.)
In two prior cases, Dr. Werner defined “news days” as days on which a company
issued a press release. (Id. Ex. 1C.) Under that methodology, a cause-and-effect
relationship existed for only one of the Affected Stocks. (Id.)
And with the combination of any of the three types of major events Dr. Werner
used in his prior cases, his news/no news test would have found a cause-and-
effect relationship for only two of the Affected Stocks (AMC and BBBY) in the
year preceding the Class Period. (Ex. 6, Grenadier Rebuttal ¶ 53 n.96.)
Critically, Dr. Werner does not dispute these results. (Ex. 1, Werner Dep. Tr. 200:17-22.)
Dr. Werner should not be permitted to change his methodology without any basis
in academic research or legal precedent to arrive at the result his clients want. See Cordoves,
104 F. Supp. 3d at 1364 (excluding opinion where expert’s methodology “contradicted . . . his
own past practice”). Such “cherry-picking” is the hallmark of an unreliable methodology and
3
Dr. Werner may have additional reports prior to this case in which he conducted a news/no
news test. Prof. Grenadier used all of Dr. Werner’s publicly available reports that Prof.
Grenadier was able to locate. (See Ex. 6, Grenadier Rebuttal ¶ 52 (listing the fourteen cases).)
17
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 23 of 28
requires exclusion of Dr. Werner’s opinion. See id.; Zantac, 2022 WL 17480906, at *57, *132.
II. DR. WERNER’S OPINIONS CONCERNING EFFICIENCY IN HIS REBUTTAL
REPORT ARE ALSO UNHELPFUL AND UNRELIABLE.
In his rebuttal report, after reviewing the analyses by Prof. Grenadier and Prof.
Fischel of efficiency during and immediately preceding the Class Period, Dr. Werner conducted
a new test, which he refers to as a “binomial test.” (Ex. 9, Werner Rebuttal ¶¶ 45-52.) This test
should also be excluded under Daubert because it does not actually test whether stock prices
move in response to news, and thus provides no information on whether a market is efficient.
To conduct his so-called “binomial test,” Dr. Werner defined a “news day” as any
day that Prof. Grenadier or Prof. Fischel identified in their opening reports as having potentially
value-relevant news (i.e., news about the specific company that was potentially relevant to its
stock price, such as a rating downgrade) in the week before the Class Period (from January 21 to
January 27, 2021). (Id. ¶ 46.) Dr. Werner then used a basic statistical test to determine the
likelihood that a given number of one-in-twenty events would occur during the week-long period
he was measuring. (Id. ¶ 49.) In this case, those one-in-twenty events are “news days” with
statistically significant stock-price movements. (Id.) The binomial test shows that, for six of the
Affected Stocks, the probability of randomly selecting as many days with statistically significant
price movements as were present among the “news days” was less than 5% (the threshold for
statistical significance). (Id. ¶ 50.) Dr. Werner concludes (wrongly) that “we can reject the null
hypothesis that the companies’ security prices behave no differently on days with company-
specific news than on days without company-specific news.” (Id. ¶ 51.)
The problem is that Dr. Werner did not look at the days without company-specific
news. As he admitted at deposition, the binomial test does not compare “news” days to “non-
news” days. (Ex. 1, Werner Dep. Tr. 334:14-21.) It merely asks whether it was statistically
likely for stock prices to move to that extent that frequently on “news” days during the five-
trading-day period from January 21 to 27, 2021. (Ex. 9, Werner Rebuttal ¶ 49.) But this tells us
nothing about “news” days vs. “non-news” days if the period at issue (January 21 to 27, 2021)
was one in which the stocks became much more volatile and prices were moving significantly
without regard to the presence of news. Dr. Werner’s conclusion that “security prices behave []
differently on days with company-specific news than on days without company-specific news”
simply does not follow from his binomial test, because the binomial test does not even consider
“non-news” days. (See Ex. 9, Werner Rebuttal ¶ 51.)
18
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 24 of 28
This illustrates a second key mistake that derives from Dr. Werner’s mistaken lay
attempt to interpret Halliburton II. Dr. Werner testified at deposition that he took this approach
here because he believes that Halliburton II rejects the notion—widely adopted by academics
and Dr. Werner himself in other cases, including those after Halliburton II—that prices should
remain stable in the absence of new information. (Ex. 1, Werner Dep. Tr. 65:19-68:6, 84:12-
85:9, 100:7-19.) Of course, this notion that the Supreme Court decided in Halliburton II that the
proper approach to efficiency is to reject the prevailing scientific view would turn Daubert on its
head. Dr. Werner’s fundamental misunderstanding of what must be shown to reach a conclusion
of efficiency pollutes all of his opinions and requires exclusion. See Welding Fume Prods., 2005
WL 1868046, at *7 (excluding opinion of expert who applied wrong legal standard).
In fact, the Eleventh Circuit has made clear that “a corollary of the efficient
market hypothesis” is that “disclosure of . . . information already known by the market will not
cause a change in the stock price.” FindWhat, 658 F.3d at 1310. In other words, the stock price
generally should not move (in significant amounts) in the absence of new, value-relevant
information. Dr. Werner testified that the Eleventh Circuit’s statement of the legal standard in
FindWhat is correct under the academic standard of “fundamental efficiency.” (Ex. 1, Werner
Dep. Tr. 84:12-86:10.) He further admitted that he did not apply this standard in his work
because, according to his lay opinion about Halliburton II, this is no longer the legal standard.
(Id. at 65:19-68:6.) But that is simply wrong for the reasons explained above.
Dr. Werner’s current premise is also belied by his own prior sworn testimony.
For example, he has testified in prior cases that “if I saw all sorts of crazy stock price movements
on no information, that might inform me that the stock wasn’t efficient.” (Ex. 3, Hi-Crush Dep.
Tr. 124:25-125:4.) When confronted here with that prior testimony, Dr. Werner bizarrely
refused to either agree or disagree with his own prior testimony. (Ex. 1, Werner Dep. Tr. 89:4-
90:21, 97:1-98:12.) Moreover, in his reports in this case, as well as at deposition, Dr. Werner
made clear that he understands the importance of comparing stock price movements on days with
news to days without news. (See, e.g., Ex. 1, Werner Dep. Tr. 145:16-146:4 (explaining that it is
important to effect such a comparison “[b]ecause you are trying to determine whether or not the
stock price reacts differently to news days as opposed to non-news days”); Ex. 8, Werner Report
¶ 73 (“After the high information flow and lesser information flow dates have been identified,
statistical analysis is used to compare the stock price reactions . . . .”); Ex. 9, Werner Rebuttal
19
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 25 of 28
¶¶ 49, 51 (stressing the importance of observing a difference between the frequency of
statistically significant price movements on days with and without news).)
Finally, in the Alibaba report that Dr. Werner cited as authoritative in his
deposition, Dr. Tabak provides a cogent explanation of the methodological problem with
Dr. Werner’s rebuttal approach. In particular, Dr. Tabak explains why it is “necessary” to
compare the results of news and “non-news” days:
Establishing two such groups is necessary because even if the market for a
stock were not efficient, there would generally still be some news days
that were randomly associated with stock price movements. Selecting a
few examples of such instances and claiming to have thereby found an
association between news and stock price movements would be clearly
incorrect.
(Ex. 7, Alibaba Report ¶ 37 & n.29.) Here, given the significant price movements on most days
from January 21 to 27, 2021, including days without news, significant price movements on days
with news provide no evidence that the price movements were caused by that news.
Prof. Grenadier confirmed this methodological deficiency in Dr. Werner’s
binomial test. When Prof. Grenadier compared the stock-price movements on the days (from
January 21 to 27, 2021) he identified as having potentially value-relevant news for each
company with the stock-price movements on the days without such news, he found no difference.
(Ex. 13, Dep. Ex. 288, Analysis in Response to Werner Table 7.)4 Put another way, the five
trading days from January 21 to 27, 2021 for the Affected Stocks have a disproportionately large
number of significant stock-price movements, on both days with and without news. Because
Dr. Werner’s binomial test in no way helps to determine whether there is a difference between
“news” and “non-news” days, it cannot help answer whether there is a cause-and-effect
relationship between news and price movements. It therefore does not inform the question
whether the market was efficient during the Class Period and should be excluded.
CONCLUSION
For these reasons, Dr. Werner’s market efficiency opinions should be excluded.
4
Plaintiffs’ counsel has objected to Prof. Grenadier’s analysis as untimely. (Ex. 1, Werner
Dep. Tr. 326:13-327:7.) But the binomial test is a new test that Dr. Werner presented for the first
time in his rebuttal report and is different in kind from the tests he presented in his opening
report. Prof. Grenadier prepared his analysis and provided it to Plaintiffs’ counsel before both he
and Dr. Werner were deposed. Thus, if any expert analysis here is untimely, it is Dr. Werner’s
binomial test, which should have appeared in his opening report.
20
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 26 of 28
REQUEST FOR HEARING
Pursuant to Local Rule 7.1(b)(2), Robinhood hereby requests an evidentiary
hearing on this motion. A Daubert hearing is particularly appropriate in “complicated cases
involving multiple expert witnesses.” United States v. Hansen, 262 F.3d 1217 (11th Cir. 2001).
This is such a case. Efficiency lies at the intersection of contested legal, economic and factual
issues. The parties have together proffered three experts, who have performed a substantial
number of statistical and other analyses. These analyses and the conflicts between the experts on
these issues can be better elucidated through oral questioning and testimony. Moreover, this
motion raises issues of credibility, determinations of which are particularly well suited to a
hearing. See Ramirez v. E.I. DuPont de Nemours & Co., No. 8:09-CV-321, 2010 WL 2921619,
at *4 (M.D. Fla. July 23, 2010) (scheduling a hearing where a credibility determination was
required to decide a Daubert motion).
If the Court also grants a hearing on Plaintiffs’ Motion for Class Certification,
Robinhood requests that the hearings be held in conjunction. The parties’ arguments with
respect to the two motions are closely intertwined. Accordingly, a joint hearing will provide
further clarification on these issues and promote judicial efficiency.
LOCAL RULE 7.1(A)(3) CERTIFICATION
Pursuant to Local Rule 7.1(a)(3), undersigned counsel hereby certifies that on June 2,
2023, we conferred with counsel for Plaintiffs in a good faith effort to resolve by agreement the
issues presented in this motion, but counsel were unable to reach an agreement.
21
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 27 of 28
Dated: June 7, 2023 /s/ Samuel A. Danon
HUNTON ANDREWS KURTH LLP
Samuel A. Danon (FBN 892671)
Tom K. Schulte (FBN 1025692)
María Castellanos Alvarado (FBN 116545)
333 S.E. 2 Avenue, Suite 2400
Miami, FL 33131
Telephone: (305) 810-2500
Facsimile: (305) 810-2460
sdanon@huntonak.com
tschulte@huntonak.com
mcastellanos@hunton.com
CRAVATH, SWAINE & MOORE LLP
Antony L. Ryan
Kevin J. Orsini
Brittany L. Sukiennik
825 Eighth Avenue
New York, NY 10019
Telephone: (212) 474-1000
Facsimile: (212) 474-3700
aryan@cravath.com
korsini@cravath.com
bsukiennik@cravath.com
Counsel for Defendants Robinhood Markets,
Inc., Robinhood Financial LLC and
Robinhood Securities, LLC
22
Case 1:21-md-02989-CMA Document 566 Entered on FLSD Docket 06/07/2023 Page 28 of 28
CERTIFICATE OF SERVICE
I HEREBY CERTIFY that on June 7, 2023, I electronically filed the foregoing document
with the Clerk of the Court using CM/ECF. I further certify that the foregoing document is being
served this day on all counsel of record via transmission of Notices of Electronic Filing
generated by CM/ECF or in some other authorized manner for those counsel or parties who are
not authorized to receive Notices of Electronic Filing.
Dated: June 7, 2023 /s/ Samuel A. Danon
Samuel A. Danon (FBN 892671)
23
File and source
- File
- gov.uscourts.flsd.590042.566.0.pdf
- Size
- 279,053 bytes
- SHA-256
- 0fa519b2affe0c15847d0130725f33bce3b507cd4c1e2258f000877b23af6c1d
- Original
- archive.org