The opinion
UNBINE
ELECTRONICALLY FILED
VIA ECF | BOC | October 22, 2020
DATE FILED:__ 10/23/2020
The Hon. Stewart D. Aaron
United States District Court for the Southern District of New York
500 Pearl St.
New York, NY 10007
Re: = Allianz, et al. v. Bank of America Corp. et al., 18-cv-10364-LGS
Dear Judge Aaron:
Joint update on search terms. Pursuant to Your Honor’s October 2 and October 17, 2020 Orders,
the parties submit this joint letter to provide the Court with a status report on the parties’ search
term negotiations. The parties are happy to report that they have reached agreement on a set of
search terms that defendants will test on the agreed-upon or Court-ordered custodians. The parties,
however, disagree on a few topics. The parties’ respective positions are set forth below.
With respect to the issues raised in this Joint Letter, the Court hereby ORDERS, as follows:
1. As an appendix to the letter to be filed with the Court on 11/5/20, the parties shall list the names of each
custodian at each Defendant for whom audio recordings are sought, indicating whether such custodian
was indicted, terminated and/or suspended. Each of the Defendants that has a name on the list shall use it
best efforts to determine if audio recordings exist for its custodian(s) and, if so, whether they are readily
accessible, and shall report its findings as part of the appendix. In addition, each Defendant shall address
in the appendix the burden and expense of audio collection and review so that the Court can weigh the
extent of appropriate cost-shifting, if production of any audio recordings is ordered.
Bank of America, Barclays and Deutsche Bank shall run Plaintiff name searches over documents
previously collected but not produced, with the limitations already agreed to by Plaintiffs. Once hit counts
are obtained, the parties shall meet and confer, as appropriate, to minimize any undue burden associated
with production of documents identified from such searches.
Each of Plaintiffs and Defendants shall make diligent inquiries to determine if they have reasonably
accessible shared drives/network drives that may contain relevant information and, if so, shall conduct a
reasonable search of such shared drives, to the extent they have not already done so.
As asecond appendix to the letter to be filed with the Court on 11/5/20, Plaintiffs shall supply to the
Court what they refer to as the "targeted list of chat rooms" and Defendants refer to as "hundreds of
chatroom ID numbers and names,” along with the date range(s) sought.
SO ORDERED.
Dated: 10/23/2020 ai
Plaintiffs’ Position. After three months of negotiations, the parties have resolved most search
term disputes except for the three issues discussed below. Specifically, Defendants (1) seek to
shield from discovery “smoking gun evidence” (including audio recordings for indicted or
terminated employees) by not searching all readily accessible sources of custodial documents; (2)
refuse to run plaintiff-specific terms over already collected late-period documents, and (3) refuse
to pull complete transcripts for a targeted list of now infamous chat rooms used to execute the
conspiracy. Defendants’ attempts to withhold highly relevant and proportional discovery should
be rejected.
Defendants are refusing to collect early period audio for a targeted set of indicted, terminated
or suspended employees and to collect other readily available custodial materials for all
custodians. Instead, they want to limit their searches to just chats and emails. Defendants’
position is unreasonable and would allow Defendants to hide numerous incriminating materials.
For example, Defendants refuse to produce early period audio recordings for employees who were
indicted, terminated or suspended in connection with FX investigations, unless Plaintiffs identify
specific calls referenced in chats and emails. But we know from regulatory investigations that
Defendants’ traders frequently coordinated over the phone, and not all such calls will be referenced
in documents. See, e.g., TAC ¶¶ 290, 294, 296. The charge against indicted Citi trader, Rohan
Ramchandani, accused him of “using chat rooms, electronic message and telephone calls to
coordinate [the] collusion.” Id. ¶ 297 (emphasis added). At Mr. Ramchandani’s trial, the
prosecution played phone calls that Mr. Ramchandani participated in, where the conspirators
“planned how to most effectively manipulate EUR/USD.” Id. ¶ 300. When Mr. Ramchandani
was prohibited from accessing “The Cartel” chatroom1 in January 2013, he told his co-conspirators
that “I have to leave this room and will call you.” Id. ¶ 313.
Defendants complain that audio is “generally not searchable,” cannot easily identify participants on
the call, and will require listening to recordings “at the pace of natural speech”; but these are based
on nothing more than self-serving speculation. Again, our request here is not for audio files for the
entire list of custodians, but only for a small subset of traders who were indicted, terminated or
suspended in connection with FX-related investigations. Defendants claim this set includes 87
custodians, but by lumping all Defendants together, Defendants mask the true burden for each
Defendant. Four Defendants actually have zero custodians to search; some have as little as one to
three; and only three Defendants have to search more than 10 (UBS 11, Barclays 14 and Citi 20).
That Defendants jointly claim undue burden (when some have zero audio to search) highlights the
unsubstantiated nature of their complaint. Defendants have not even disclosed whether this discrete
set of custodians have readily accessible early period recordings, much less their volume.
Defendants’ blanket refusal to even investigate these basic questions is unreasonable. See Kleiner v.
Burns, 2000 WL 1909470, at *4 (D. Kan. Dec. 22, 2000) (granting motion to compel defendant to
provide all ESI omitted from initial disclosures including audio recordings).2
1 “The Cartel” was the name of an electronic multi-bank chat room used by traders from various banks, including
Defendants Citi, JP Morgan, and UBS, to coordinate their manipulation of FX benchmark rates. TAC ¶¶ 284, 468,
481, 503.
2 Defendants claim Plaintiffs “have flatly refused to collect or review audio for their custodians.” But Plaintiffs
actually informed Defendants that most of them do not have audio, and in any event, there is no parallel here given
Plaintiffs have no traders who were indicted, terminated or suspended in connection with FX investigations—the
targeted universe for whom Plaintiffs are asking Defendants to collect audio.
While Defendants wax on about the difficulties of searching audio, technology exists to transcribe
recordings. Defendants have retained sophisticated law firms with access to the latest software
that can circumvent the need to manually listen to each audio call. Veritone, Intelligent Voice,
and Everlaw are just some of the available options for reviewing audio files without having to
listen in real time. Defendants in fact produced audio to regulators and in FOREX—cases
involving similar allegations and a much larger universe of custodians.3 They do not explain how
they were able to collect and review these late period recordings, the very tasks they now claim
are cost-prohibitive or unreasonable to do in this case. Citi, for example, told Plaintiffs it collected
audio recordings for specific traders by simply pulling all calls on the traders’ phone lines and their
“internal speakerbox audio.” Defendants do not explain why they cannot similarly pull calls on
their trader’s lines. In any event, if any particular Defendant has a unique burden, the parties can
negotiate a tailoring of this request, just as they have agreed to negotiate adjustments to search
terms that result in an “unreasonable number” of hits.
The same applies to instant messages, text messages, shared drives or network drives containing a
custodian’s documents.4 See Paisley Park Enterprises, Inc. v. Boxill, 330 F.R.D. 226, 234 (D. Minn.
2019) (“In the contemporary world of communications, even leaving out the potential and reality of
finding the modern-day litigation equivalent of a ‘smoking gun’ in text messages, e-mails, and
possibly other social media, the Court is baffled as to how Defendants can reasonably claim to
believe that their text messages would be immune from discovery.”). If such materials exist and can
be produced without undue burden, Defendants should search for these categories of materials, just
as Plaintiffs are doing. And to the extent a particular Defendant believes that searching a particular
source would be unduly burdensome, it should provide sufficient details so that the Court and
Plaintiffs can assess the claim.5 This is how custodial discovery normally works, and there is no
reason Defendants should get special treatment.
Bank of America, Barclays, and Deutsche Bank are refusing to produce late period Plaintiff-
specific materials. As background, Plaintiffs have largely adopted the late period productions
Defendants made to regulators and in FOREX. However, these productions have an obvious gap—
neither the DOJ nor FOREX plaintiffs were focused on Plaintiffs, and so these prior productions
were not designed to capture Plaintiff-specific materials. Among other materials, these terms are
3 Defendants’ argument that requiring them to produce audio recordings would violate Judge Schofield’s
presumptive 160 hour rule on ESI discovery is absurd. Defendants know full well that the 160 hour rule does not
apply to complex cases like these—their many demands to Plaintiffs have already resulted in Plaintiffs incurring
more than 160 hours to collect, search and review documents.
4 Defendants complain that Plaintiffs only identified shared drives in recent meet and confers. But Plaintiffs have
consistently asked Defendants to produce all readily accessible custodial materials. There was no need to categorize
the different types of applicable custodial materials – nor could Plaintiffs do so since Defendants claimed they could
not provide such a list until custodial negotiations were complete. In any event, Defendants’ belated and equivocal
agreement to search shared drives does not eliminate the need for a Court order that requires Defendants to search
all reasonably accessible custodial materials.
5 See In re Novartis and Par Antitrust Litig., 2020 WL 3317203, at *5 (S.D.N.Y. June 18, 2020) (“Claims of undue
burden must ‘explain the manner and extent of the burden, as well as the consequences of compliance.’”); Montesa
v.Schwartz, 2015 WL 13173164, at *2 (S.D.N.Y. Feb. 20, 2015) (“While the Defendant District states that it will
suffer financial burden to include these terms in its electronic search, the claim is conclusory and therefore
insufficient to bar discovery.”).
designed to capture traders’ exchanging sensitive information concerning Plaintiffs’ FX trades, one
of the hallmarks of FX manipulation according to regulators.
To address this gap, Plaintiffs have asked Defendants to run Plaintiff name searches over
documents previously collected but not produced. Contrary to Defendants’ claim, this request
does not seek to expand the custodians from whom documents are collected, because they involve
already collected documents. And to minimize any review burden, Plaintiffs also agreed that
productions should (1) exclude blast emails, interparty communications, and chats with more than
15 participants, (2) be run over only a subset of custodians with the highest hit counts (10 max),
and (3) adjustments can be negotiated if a term results in an “unreasonable number” of hits.
Yet, these three Defendants have steadfastly refused to run these searches, claiming Plaintiffs seek
to “side-step” the Court’s recent orders on custodians because we are “unhappy” with them. This
is false. Plaintiffs consistently maintained throughout the parties’ negotiations that they would
seek to compel Defendants to supplement late period productions to identify Plaintiff-specific
materials. This request was always separate from any custodial negotiations, which were focused
primarily on collecting new documents from the early period. Indeed, most other Defendants
have agreed to run these searches, demonstrating that Plaintiffs’ request is reasonable, not
burdensome, and independent of the parties’ custodial agreements.
Defendants refuse to produce all chats in chatrooms known to be used by their traders to
manipulate the FX market. Plaintiffs have asked Defendants to pull the complete transcripts for
a targeted list of chat rooms which Plaintiffs and regulators have identified as containing “smoking
gun” evidence of the conspiracy. Defendants have agreed to produce these transcripts only to the
extent they fall within the materials of the agreed-upon or Court-ordered custodians. But
Defendants’ proposal would exclude highly relevant and incriminating chats, which given their
intended strategy to argue that there is no overarching conspiracy but only isolated chats, will
deprive Plaintiffs of critical evidence to rebut one of their primary defenses.
Defendants claim Plaintiffs seek to expand the number of custodians ordered by the Court. Not
true. Chatroom transcripts are not a custodial source. In fact, what Defendants are proposing to
do—sift through chats to identify ones that relate to agreed-upon or Court-ordered custodians—is
actually more burdensome than simply pulling the complete transcripts from a narrow list of
chatrooms and producing them. Nor is pulling the transcripts an insurmountable task—Defendants
can easily get the transcripts from Bloomberg.6 Defendants also insinuate that Plaintiffs are
seeking all chatrooms or claiming all chatrooms are illicit. Not so. Plaintiffs are not asking
Defendants to produce transcripts of all chatrooms; rather we provided Defendants a targeted list
of chatrooms, which they in fact agreed were proper chatrooms to search.
Defendants’ existing chat productions highlight the problems with having Defendants artificially
manipulate a non-custodial source into a custodial collection. Many of the incriminating chats
referenced in regulatory findings or criminal trial transcripts are missing from Defendants’
productions. See, e.g., TAC ¶¶ 501, 588, 609. Further, when Plaintiffs applied the agreed upon
6 Defendants’ current position on burden is noticeably different now that the shoe is on the other foot. When it came
to Plaintiffs’ production, Defendants argued that review shouldn’t be a significant burden because any document
hitting on search terms and related to FX should just be produced. See 06/11/2020 Kolbe email to A. Alden.
chatroom number terms7 over existing productions, they found that for 73% of the search terms,
Defendants had produced fewer than 100 documents per term, suggesting custodial discovery is
not an appropriate vehicle for gathering these chats, which are non-custodial documents, akin to
centralized materials.8 They are similar, for example, to the FX trade data system; traders have
different logins to use the data system much like they do to access different chatrooms. Even
though different traders input information into the data system, it is still considered a centralized
source. And it would be unreasonable to claim that producing trade data entered by non-
custodians was seeking to expand on the Court’s orders on custodians. Nor would it be
reasonable for Defendants to sift through trade data and only provide Plaintiffs the trades that are
attributable to agreed-upon or Court-ordered custodians.
7 Chatrooms have a chatroom ID or number assigned to them by Bloomberg. The parties have agreed to search
terms that include such chatroom IDs.
8 Defendants’ analogy of Plaintiffs’ chatroom request to a search for all emails with a specific domain name is
misguided. Unlike email, which can be sent to anyone in the world with an email address, the chatrooms used by
Defendants’ traders to coordinate FX prices and benchmarks were used by a defined number of participants over
specific periods of time. Because they were created for illicit purposes, most chatrooms Defendants used to engage
in wrongful conduct included only a few trusted members to avoid unwanted detection and disclosure. It is
reasonable to assume that if a trader created a chatroom to improperly share confidential information with a rival
trader, most (if not all) chats in that room will be connected with the illicit purpose for which the chatroom was
created.
Defendants’ Position: Defendants have agreed to conduct broad supplemental custodial
discovery, including testing more than 15,000 individual search terms over the readily accessible
emails and chats of approximately 500 custodians.9 This supplemental discovery—in addition to
the over 2.5 million documents defendants have already produced—exceeds what is proportional
here. Plaintiffs, however, demand more.
Review of 5 Years of Audio Is Disproportionate To the Needs of the Case. Last week, plaintiffs
told this Court that “chats or emails [are] where conspiratorial communications are most likely to
occur.” (ECF No. 610 at 1 n.2.) Defendants have agreed to run plaintiffs’ search terms over the
readily available chats and emails of their custodians.10 But defendants object to an unduly
burdensome and disproportionate wholesale review of 5 years of audio recordings, much of which
is likely not to be reasonably accessible (e.g., archived on back-up tapes requiring restoration).
Unlike written documents, audio is generally not searchable. And, even if audio was reasonably
accessible, plaintiffs’ proposal would likely require defendants to review millions of hours of
irrelevant conversations to identify potentially responsive calls. Audio is also inherently more
time-consuming to review compared to written documents because audio generally must be
listened to in its entirety at the pace of natural speech. The audio here will also be filled with
trading jargon, as well as slang unique to various jurisdictions, which will further slow the pace of
review. In sum, an audio review is disproportionate to the needs of a case where plaintiffs claim
relevant conduct was “most likely to occur” in chats or emails. No doubt recognizing these
burdens, plaintiffs refuse to collect or review audio themselves.
Defendants offered to discuss any targeted requests for audio recordings if plaintiffs identify
communications in which a relevant discussion may have continued over audio (plaintiffs have
search terms designed to identify precisely these kinds of discussions in chats and emails, such as
“call me” OR “call my”). Plaintiffs rejected that proposal. Plaintiffs’ argument that they are only
seeking a “discrete set of custodians” does not resolve the proportionality concerns. Plaintiffs are
requesting 5 years of audio from nearly 90 custodians, which is facially overbroad and
disproportionate to the needs of this case. Plaintiffs also argue that technology exists that can
transcribe audio recordings. But that technology does not work for open lines on trading desks
where participants discuss and execute trades without identifying themselves and often use
unintelligible trading jargon. Also, contrary to plaintiffs’ assertion, only a few defendants
produced audio and none of the defendants produced new audio in FOREX (other than Credit
Suisse, which is still litigating that case) and this says nothing about the burden of doing so. For
example, Barclays spent over two years and hundreds, if not thousands, of work hours, collecting,
reviewing and producing the audio it produced to regulators and subsequently to plaintiffs.11
Similarly, plaintiffs’ request for “shared drives” or “network drives”—which they made for the
first time this past week—makes no sense in the context of custodial discovery, since shared
9 As used in this section, “custodian” or “custodians” refers to both agreed-to and Court-ordered custodians, unless
otherwise specified.
10 Defendants agreed on a recent meet and confer to investigate the availability of text messages and instant messages
and to schedule a meet and confer with plaintiffs once they had more information, so there is no dispute on this topic.
11 A full discussion of the extraordinary burdens of audio collection and review is not possible in this letter, especially
because plaintiffs changed their demand at 1:00 pm Eastern on the day this letter was due. If the Court requires more
detail to resolve this dispute, defendants respectfully request the opportunity to make additional submissions.
drives/network drives are non-custodial (i.e., not linked to a particular custodian). In addition,
plaintiffs belated request upends the parties’ negotiations and the Court’s custodial rulings, which
evaluated the burden of custodial discovery in connection with the collection and review of
electronic communications—not shared drives/network drives. The inclusion of those non-
custodial categories significantly expands the overall burden. Plaintiffs new characterization of
shared drives/network drives as “custodial” is also puzzling because the parties have spent the past
year discussing non-custodial discovery from centrally stored locations, including shared
drives/network drives. In any event, as explained to plaintiffs, defendants are willing to make
reasonable inquiries to determine if they have reasonably accessible shared drives/network drives
that may have relevant information and, if so, will conduct a reasonable search of such shared
drives (if they have not already done so).12 That is the way that discovery is always conducted,
including in this case, and plaintiffs have offered no reason why discovery should now be handled
any differently.
Plaintiffs Are Not Entitled To Additional Custodians Beyond Those Ordered by the Court. The
Court recently set the number of custodians for each defendant involved in an unresolved dispute.
Unhappy with the Court’s rulings, plaintiffs now demand that several defendants—including those
whose custodians the Court just addressed—run so-called “plaintiff name” search terms over
additional custodians beyond the Court’s orders.
Defendants have agreed to run “plaintiff name” search terms over the documents for Court-ordered
custodians for the entire eleven-year period, including the later period. As relevant here, Bank of
America has agreed to do so for 30, Barclays for 43, and Deutsche Bank for 30 in response to the
Court’s order from last Saturday. (ECF No. 614.) This is more than enough to fill any purported
“gap” in defendants’ prior late-period productions. Yet, plaintiffs argue that Bank of America,
Barclays and Deutsche Bank should also run those terms over additional custodians13 not included
in the Court-ordered custodian lists. The parties just litigated custodian limits, and the Court ruled.
Indeed, plaintiffs argued that their custodian demands were necessary, in part, to search for
“documents that hit on plaintiff name terms.” (See, e.g., ECF. No. 610 at 2.) Plaintiffs should not
be permitted to sidestep those rulings in the guise of a search term dispute. Plaintiffs’ argument
that their requests regarding plaintiff name searches were “separate” is disingenuous. The parties
have been negotiating search terms and custodians as a package for more than two months. There
were no “separate” negotiations. If plaintiffs thought there were, they should have informed the
Court (and defendants) during the briefing on custodians.14
Plaintiffs’ Request for Entire Chatrooms Is Meant To Side-Step the Court’s Custodian Rulings
and the Parties’ Negotiations. Defendants agreed to run hundreds of chatroom IDs and chatroom
names over the custodians. Not satisfied—and despite these negotiations’ express focus on
12 Plaintiffs should do the same. For example, having deleted the platform that formerly hosted TARs, BlackRock
should conduct a thorough search of any shared drives used by sampling custodians who received TARs.
13 Plaintiffs have demanded 10 more custodians each from Deutsche Bank and Bank of America, and 9 from
Barclays.
14 If plaintiffs are correct that defendants’ prior searches missed “hundreds of thousands to over a million documents”
that are relevant (which they are not), the incremental burden of searching almost 30 additional custodians’ worth of
documents is incontrovertible. The fact that certain other defendants have agreed to assume this burden for their own
reasons is irrelevant.
custodial discovery—plaintiffs demanded for the first time this week that defendants collect,
review, and produce all chats from hundreds of chatroom ID numbers and names, regardless of
the individuals in those chatrooms. In other words, plaintiffs want to treat a custodial data source
(chats) as a non-custodial source to backdoor additional discovery that is untethered to custodians
or search terms. This sweeps aside the parties’ months-long custodial and search term discussions
on this issue and would require defendants to collect and review tens (or possibly hundreds) of
thousands of chats, irrespective of whether any custodians participated in those chats.
Defendants’ existing productions include millions of pages of chats. Plaintiffs have these
documents and could—and presumably did—use them to identify chat participants that plaintiffs
believe are important and should be custodians. Assuming that is true, defendants’ proposal of
running the chatroom names and IDs over the custodians will capture the relevant chat
communications. Indeed, plaintiffs’ request would only result in the production of chats among
individuals whom plaintiffs have not claimed to be purported “bad actors” and who are not even
among the hundreds of additional custodians for whom defendants have already agreed to produce
documents. To the extent that plaintiffs decided not to include other chatroom participants as
custodians, they should not be able to walk back that tactical decision, their negotiated resolutions
with several defendants, and the Court’s rulings on custodians by demanding that defendants
collect entire chatrooms without regard to custodian.
Plaintiffs attempt to justify their proposal by falsely claiming that “many of the incriminating chats
referenced in regulatory findings or criminal trial transcripts are missing from Defendants’
productions.” But this is not true. In fact, during the course of briefing this letter, plaintiffs
repeatedly identified supposedly “missing” documents only to have defendants identify the bates
numbers for those documents in our response. Their most recent citations fare no better. If
plaintiffs simply reviewed the more than 2.5 million documents they have, they might realize that
nothing is missing. Plaintiffs proposal should be rejected because they have not offered a valid
reason for deviating from standard custodial discovery in which the parties apply negotiated search
terms to custodians.
Plaintiffs seem to indicate that there is no burden to their request because defendants can simply
collect tens (or possibly hundreds) of thousands of chats and turn them over to plaintiffs without
reviewing them. As plaintiffs and the Court know, that is not how discovery works. Defendants
will have to review the chats for, among other reasons, responsiveness and data privacy issues.
Defendants agreement to run plaintiffs’ chat IDs and chatroom names over custodial documents is
burdensome enough without unnecessarily increasing that burden by sweeping in documents from
non-custodians. The Court should reject plaintiffs’ overreach.
Finally, contrary to plaintiffs’ assertion, there is nothing improper about participating in a private
chatroom. The banks’ traders were not only permitted to participate in such chatrooms with other
market participants, but they were expected to in order to facilitate trading in a fast-moving market.
As plaintiffs know, every defendant in this case had multiple private chatrooms with traders at
every plaintiff in this case because that is how FX traders communicate and execute trades.
Plaintiffs insinuation that all private chatrooms are illicit is demonstrably false.
Respectfully submitted,
QUINN EMANUEL URQUHART & SHEARMAN & STERLING LLP
SULLIVAN, LLP
By: /s/ Jeffrey J. Resetarits
By: /s/ Daniel L. Brockett Jeffrey J. Resetarits
Daniel L. Brockett
Richard F. Schwed
51 Madison Avenue, 22nd Floor
Adam S. Hakki
New York, New York 10010
Telephone: (212) 849-7000 599 Lexington Avenue
Fax: (212) 849-7100 New York, New York 10022
danbrockett@quinnemanuel.com Telephone: (212) 848-4000
jeffrey.resetarits@shearman.com
Anthony P. Alden (pro hac vice) ahakki@shearman.com
Jeremy D. Andersen (pro hac vice) rschwed@shearman.com
Johanna Y. Ong (pro hac vice)
865 South Figueroa Street, 10th Floor
Attorneys for Defendants Bank of America
Los Angeles, California 90017
Corporation, Bank of America, N.A. and Merrill
Telephone: (213) 443-3000
Fax: (213) 443-3100 Lynch, Pierce, Fenner & Smith Incorporated
anthonyalden@quinnemanuel.com
jeremyandersen@quinnemanuel.com
johannaong@quinnemanuel.com
Counsel for Plaintiffs