5.Joseph Pimbley — Direct/Cross
717 linesDIRECT EXAMINATION BY MR. EVERDELL:
MR. EVERDELL: Good morning, Mr. Pimbley.
JOSEPH PIMBLEY: Good morning.
MR. EVERDELL: Mr. Pimbley, what do you do for a living?
JOSEPH PIMBLEY: I am a consultant for financial issues.
MR. EVERDELL: Okay. And what does it mean to be a consultant for financial issues?
JOSEPH PIMBLEY: It means I have clients who hire me for a variety of both skills and tasks they need, which does include litigation for financial disputes; it also includes building models for complex investments and giving advice on investments and risk management.
MR. EVERDELL: And where do you work?
JOSEPH PIMBLEY: Well, New York State. I live in New York State and I work out of my home office.
MR. EVERDELL: Are you affiliated with any consulting companies?
JOSEPH PIMBLEY: Yes. For this and several engagements, I'm affiliated with PF2 Securities.
MR. EVERDELL: And what is PF2 Securities?
JOSEPH PIMBLEY: It's a firm that specializes in litigation dispute services.
MR. EVERDELL: What is your educational background, Mr. Pimbley?
JOSEPH PIMBLEY: Education is all in physics. I have a bachelor's, master's, and PhD.
MR. EVERDELL: Where are those degrees from?
JOSEPH PIMBLEY: Rensselaer Polytechnic Institute.
MR. EVERDELL: When did you receive those degrees?
JOSEPH PIMBLEY: 1980, a bachelor's degree in math——minor. I'm sorry. Bachelor's in physics, minor in math; master's degree in physics, 1981; PhD, theoretical physics, 1985.
MR. EVERDELL: Can you describe your work or professional background, please.
JOSEPH PIMBLEY: Yes. I started my career at GE Research Center in upstate New York. I was a physicist for semiconductor devices, an expert on that topic.
I then, later in the '80s, became an assistant professor of applied math at Rensselaer Polytechnic Institute and worked there for several years.
I joined Citibank in 1993, 30 years ago. They were looking to hire quantitative finance people and needed somebody with my background.
MR. EVERDELL: Can you explain what quantitative finance is.
JOSEPH PIMBLEY: Yes. It's——of course finance, like many fields, has many branches, many important aspects. I was hired, you know, for Wall Street type problems, in both investments and derivatives, but those areas often require a lot of code writing, solutions of mathematics problems, and to translate the math and the computer code into practical instructions or practical devices, practical investments, practical analyses for institutions like Citibank.
MR. EVERDELL: So you mentioned you were hired for some of your particular skills. Does that include the coding skills that you're discussing here?
JOSEPH PIMBLEY: Yes, sir, it does.
MR. EVERDELL: And can you describe your familiarity with coding and databases.
JOSEPH PIMBLEY: Yes. In particular I've been writing computer code really my whole career, starting at age 20 or so, so that's quite awhile. But in my——in my physics work, also as a math professor, I would be required to——I actually——shouldn't say required. But my research would need computer code. I actually taught courses in computer code, as well as other more typical mathematics classes. And Wall Street, in the financial field, data is supremely important, so all the models I built would have to interact with the data——read input data, write output data——and so the database work, which is a much more formalized use of a high volume of data, became much more important to me when I worked with a large investment firm in the early 2000s.
MR. EVERDELL: Can you describe your work with that firm. And what firm was that?
JOSEPH PIMBLEY: That firm was ACA Capital. I joined in 2002. I had——I was originally a portfolio manager for a complex type of security called a collateralized debt obligation——CDO——and I was——I built several of these CDOs, billions of dollars, billion-dollar sizes each, with, you know——working within my firm to sell to investors and manage for them. Lots of data is needed for that, and the models need to interact with that.
But two years later, they promoted me to the executive vice president level to——really, to lead all of risk for the firm, and that——the most important component of that is the database system——I'll call it the data system——that we needed to rebuild for that firm. So——
MR. EVERDELL: Can you describe what that was.
JOSEPH PIMBLEY: Yes. And what we——what we and my team of people, which included the information technology group, that underlying database was using a product called SQL server, and we also used a——what's called a code base or code platform, computer code, that needs to work with that database. It actually manages, inputs the data, extracts data, in very quick time, does all the calculations that a financial firm needs, so that a data system really means this database, this formal database, but what I call a code layer that goes over that database.
MR. EVERDELL: You mentioned a few concepts there. Let's see if I can break that down.
First, you mentioned SQL; is that right?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: What is SQL?
JOSEPH PIMBLEY: SQL is not the word, it's just the letters S-Q-L, and it stands for "structured query language," and it——essentially what it is is what it sounds like; it is a special language that a person can use, sitting in front of the computer, just type a few commands for particular data you want to extract from a database.
MR. EVERDELL: And are you familiar with SQL?
JOSEPH PIMBLEY: Yes, I am.
MR. EVERDELL: Are you able to craft SQL queries on the database?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: You also used the terms "code platform" and "code layer." Can you describe what you mean by those.
JOSEPH PIMBLEY: Yes. By "code platform," what I was thinking as I said that, something called dot-net, and programmers——when I use the word "programmer," the word "developer" means the same thing. Somebody who writes computer code is called a programmer or developer. So developers will understand that this is an extensive Microsoft product which essentially just has different languages you can put under one program. We happened to focus on one particular language, but it's broad enough for several languages. That's the code platform. And one of the most important elements of that type of code is how it's able to get data out of the database quickly but also write data into the database.
MR. EVERDELL: Okay. So can you describe briefly how the code interacts with the database.
JOSEPH PIMBLEY: Yes. Just as you asked a minute ago about SQL——"sequel" is how we say SQL——that's a language you can sit in front of a computer, type a command, get answers out of the data. What the code does is it does these SQL commands inside its code. It's a——it's a variant on the SQL that you might see separately, but it's very important for that code language to have the ability to query. We still call it a query when we put it in that code. It queries the database.
MR. EVERDELL: And are you familiar with how to write queries to interact with databases to extract data?
JOSEPH PIMBLEY: From the code as well, yes.
MR. EVERDELL: Okay. One last term. Are you familiar with the term "relational database"?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: Can you describe what that is.
JOSEPH PIMBLEY: Yes. Essentially the dominant form of——when I say database, within the financial world is what's called the relational database. It's a concept that's almost 50 years old now, but it was a great concept, and what makes the database relational is the question of, well, how is the data structured or stored inside this——lots of hardware that has lots of memory, and the answer is that it's configured as tables. Think of a table as just rows and columns for something that's important. It might be the users in those systems, such as FTX, that all the users have to be listed somewhere. Each row in the table is a different user. And the columns going across are just——they may all be simple but just different simple attributes of the user that needs to be stored somewhere. Maybe it will be changed when a new user is added, but, you know, the database has to be increased in size. That's one table. But a database, a relational database will have many tables. We'll talk about FTX later, but it's normal to have tens to hundreds of tables within one particular database.
Finally, what makes the word "relational" matter is that each of these tables has an ID number——I'll call it a key——a key that lets me easily go to another table to get information I need elsewhere to use. If I'm also getting information from table No. 1, I can also couple that or pair that with information from a different table. That's the relational aspect.
MR. EVERDELL: I stopped at your career path. Did you use any of your expertise in databases and code in the rest of your career following ACA Capital?
JOSEPH PIMBLEY: Yes, I did. If I could——
MR. EVERDELL: Please go ahead.
JOSEPH PIMBLEY: Okay. It seems essentially all roles that I've had have certainly used computer code one way or another. Sometimes it's the most important aspect, sometimes it's only a secondary aspect. But databases, after——after ACA——if this is what you meant——after the work that I did leading that effort and creating what I think was a great data system for that firm, I only used databases when I have a client who wants that, and I've had, you know, essentially one or two clients in those last ten years or so who asked for that capability.
MR. EVERDELL: Okay. Have you ever taught or given lectures on financial data systems in your professional capacity?
JOSEPH PIMBLEY: Yes, I have.
MR. EVERDELL: Please describe that.
JOSEPH PIMBLEY: Yeah. The largest financial risk management organization in the world is called Global Association of Risk Professionals. There's a——there are second and third kind of organizations, but they're all important. Just, again, what it sounds like, a professional society. I was a member of this global association for many years, and I wrote articles for them on quantitative finance as a regular column. I contributed at all their——or many of their conferences as a speaker. They invited me to be the first person to create a video series for them on how one should understand credit risk in the financial world and how to manage it, and how to assess it. And one of the key videos I created out of the nine of that series focused on financial data systems. And the short story is, financial data systems are just absolutely required to manage——to even just run your firm, okay, or not just manage risk but to run your entire firm, you need a very high-integrity, high-capability financial system, and the database is the first key element of that; the code layer is the second key element of that.
MR. EVERDELL: And have you ever served on the board of any companies that involve financial data systems?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: Can you describe that briefly, please.
JOSEPH PIMBLEY: Yes. I served on the board for years 2013 to 2019——something like that——for a firm called SOLVE Advisors. They have since rebranded as SOLVE. They were a startup. I was their only early investor. So I was on the board for many years. The founders are actually colleagues of mine at ACA Capital, where we felt we had built this——the financial data system for ACA, and my colleagues ended up going to a business that specializes in an excellent database that they use for what's called fixed-income pricing data.
MR. EVERDELL: Your Honor, at this time the defense offers——
JUDGE KAPLAN: You don't need to do that.
MR. EVERDELL: Thank you, your Honor.
All right. I'll move on.
MR. EVERDELL: Mr. Pimbley, what is your connection to this case?
JOSEPH PIMBLEY: I was retained by your law firm in August of this year.
MR. EVERDELL: And to do what?
JOSEPH PIMBLEY: I was asked to assist with their extraction of data from the database.
MR. EVERDELL: Which database are you talking about?
JOSEPH PIMBLEY: Oh. I was provided with what was represented as a snapshot of the FTX database that's hosted on Amazon Web Services, so we often say AWS, Amazon Web Services; and in fact, I'll often call this the AWS database or the FTX database.
MR. EVERDELL: Okay. You were asked to extract data from the database, from the FTX database?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: To your knowledge have you ever met any of the witnesses in this case?
JOSEPH PIMBLEY: I don't believe I have, no, sir.
MR. EVERDELL: Have you ever met Mr. Bankman-Fried?
JOSEPH PIMBLEY: No, sir.
MR. EVERDELL: And do you have any personal knowledge of the facts of this case other than the work you performed with the database?
JOSEPH PIMBLEY: Other than the work I've performed, I have essentially zero knowledge beyond——it's really beyond what I've seen in headlines. I don't follow the stories, but what I know about headlines. The trial is going on, so I know that. I'm limited to that kind of knowledge.
MR. EVERDELL: Okay. Mr. Pimbley, are you being compensated for your work on this case?
JOSEPH PIMBLEY: Yes, I am.
MR. EVERDELL: What is your hourly rate?
JOSEPH PIMBLEY: $720.
MR. EVERDELL: Okay. And roughly how much have you spent on this case?
JOSEPH PIMBLEY: I would say about 70 hours thus far, something to that effect.
MR. EVERDELL: Is your pay in any way dependent on the opinions you give in the courtroom today?
JOSEPH PIMBLEY: No, sir.
MR. EVERDELL: And is it dependent in any way on the outcome of the case?
JOSEPH PIMBLEY: No, sir.
MR. EVERDELL: Okay. Let's go back to the AWS database that you mentioned a minute ago. What is the AWS database?
JOSEPH PIMBLEY: It's the database that represents all of the——essentially all of the data that FTX as a going concern had to have——for example, who are the users, who are——which could be customers, it could be other people, but who are the users, what are the coins, and other positions that are traded on the database, what are the values of these; all the information you can imagine to make that business run.
MR. EVERDELL: Okay. And you said it was a snapshot. You used the phrase "snapshot."
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Was the snapshot at a particular date?
JOSEPH PIMBLEY: Yes. And I——I don't know a precise date. I think it's November 12, 2022, but it's some day in November 2022.
MR. EVERDELL: And did you become familiar with the AWS database in the course of your work?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: How big is the database?
JOSEPH PIMBLEY: Well, in terms of number of tables, I believe it's almost 300. I think it's 289, but I'll say about 300 tables in the database. In terms of memory, I believe it's like 30 terabytes of memory.
MR. EVERDELL: Can you give a sort of rough estimate in layman's terms of what 30 terabytes means in terms of data.
JOSEPH PIMBLEY: Well, for me——and that's one of the first things I did is, how many users are there in this database, was this, you know, a thousand users; it turns out it's more like somewhere of 9 to 11 million users, so imagine storing, you know, I'll just say roughly 9 to 11 or 10 million users, and what trades they have on and when they——when they put money in, when they took money out, deposits and withdrawals. So to me it's a huge amount. Unlike many problems, you can't just scroll down a screen to see what's going on. Really, it's a huge amount.
MR. EVERDELL: Okay. How did you familiarize yourself with the AWS database?
JOSEPH PIMBLEY: Well, I downloaded software——first of all, I was given access to the database, not——it doesn't——it's not on my computer, it's just a remote link that I'm able to do——to use to AWS to see the database. But I also needed special software that lets me run queries, but also just review the tables even without running queries, and so I can see all the table names, I can choose any one I want and go look into it, and I——I did a lot of that just to understand how can I, you know——I'm going to get my arms around this, so let me look at——look into all the tables I can see here.
MR. EVERDELL: So is that background work to get ready for your——the specific projects on this case?
JOSEPH PIMBLEY: Yes, it was. That was background, but also it's good, you know, to jump in and work on the real problems as you're learning at the same time, and so that was my philosophy there.
MR. EVERDELL: And so you did that work?
JOSEPH PIMBLEY: Yes, I did.
MR. EVERDELL: Are you familiar with the term "query"?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Can you explain what a query is in relation to a database.
JOSEPH PIMBLEY: It's really like it sounds. It's, you ask the database to tell you something. I gave the example a minute ago. The simplest example might be, how many users are there. So you can write a query that will answer that question, okay? And that's what a query is. You're asking the database a question.
MR. EVERDELL: Okay. And what language do you use to write those queries?
JOSEPH PIMBLEY: SQL.
MR. EVERDELL: Did you familiarize yourself in any way with the code that relates to the FTX database?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Can you describe how you did that.
JOSEPH PIMBLEY: Yes. I was also provided——I believe it's on discovery in this case——what I'll call the code base for the computer code. It's printed in language called Python, which means I was given, you know, a folder of size I forget, but maybe 25 megabytes in size, 17,000 individual files, each of which is Python code; and I also have software called developer environment ID——or it's a type of developer environment. Visual Studio. But I already had this code that lets me view essentially any part of this Python code that I want, so I became very familiar with the Python code also.
MR. EVERDELL: And are you yourself familiar with Python code?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Okay. All right. Now in connection with your work on this case, were you given any specific projects to complete relating to the data in the AWS database?
JOSEPH PIMBLEY: Yes, sir.
(Continued on next page)
MR. EVERDELL: How many projects were you given?
JOSEPH PIMBLEY: Three.
MR. EVERDELL: And can you just, at a very high level, summarize each of the three projects you were asked to complete.
JOSEPH PIMBLEY: Yes. I'll order them so I can say it this way.
First was, at a high level, we want to know the in-use line of credit, which was just stated as LOC, line of credit, what was the in-use line of credit of Alameda's entities that you can extract from the database.
MR. EVERDELL: Alameda, you are referring to what?
JOSEPH PIMBLEY: I'm sorry. Alameda Research as being one of the dominant accounts within the FTX database.
MR. EVERDELL: When you say in-use line of credit, what do you mean by that?
JOSEPH PIMBLEY: Well, to me the simplest -- a simple and good analogy is credit cards. Many of us have credit cards. The credit card company will give us a maximum amount we can charge, but that maximum may not actually be what you are using. You might be charging much less, purchasing much less.
So the in use is the portion of that credit card that you're actually using, and that's the part you pay interest on. You wouldn't pay interest on the full size if you are not using that money.
MR. EVERDELL: We will get to the specifics of that project in a bit.
You said there were two other projects, is that right?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Can you just, high level again, describe what those two projects were.
JOSEPH PIMBLEY: Yes. They asked me, please, from the database, we want you to tell us the total balance, which is total of invested amount, but that's again at a high level, that's a simple description, but excluding Alameda and excluding any FTX entity. So it's, essentially, what's the total amount of everybody who is not at Alameda or FTX. And we want that number divided down into a few different groups, and that involves all coins versus some small number of coins, and coin, we may discuss, is like a currency, and also a special request about certain -- whether certain accounts are margin enabled or not.
Essentially, it was get us the total balance but without Alameda or FTX, and a few more requests.
MR. EVERDELL: We will discuss those in a little bit more detail in a second.
Let's talk first about that first project, the one that was related to the line of credit for Alameda entities. OK?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: What specifically were you asked to do in connection with Alameda's line of credit in use?
JOSEPH PIMBLEY: Well, it was essentially, we are looking at a specific -- out of the 300 tables in the AWS database, there is one specific table that tells us, from October 2021 to November 2022, what the in-use principal was, the in-use LOC amount was, and we wanted only those for the Alameda entities, the total for that time period.
MR. EVERDELL: What was the name of that table?
JOSEPH PIMBLEY: It was LOC interest charges.
MR. EVERDELL: Did you construct a query to extract that information from the database?
JOSEPH PIMBLEY: Yes. But I should clarify that I was working with a colleague. My colleague provided the query. I took the query, ran it, tested it, derived the output, but I did -- with that colleague, yes, sir.
MR. EVERDELL: You verified the query yourself?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: First explain to the jury what the query was that you used to retrieve this information from the database.
JOSEPH PIMBLEY: Well, the query -- I'm focused on just one table. Again, I said it was called LOC interest charges. So a table, as I said, is just rows and columns, very few columns, only five or six columns going across, but the rows essentially were all the accounts. But I was able -- with a query you would say, look, ignore everything that's not Alameda because we just want Alameda. I tell it, just give me Alameda. And then there is a column that says principal, and that principal is what we want. That's the amount of line of credit that's in use.
There is a column next to it called size, and I mention this because it is relevant that it shows the amount of interest to be charged on that day to that account because of this principal amount, the in-use LOC.
MR. EVERDELL: How did you satisfy yourself that the amount in the principal column was the in-use LOC?
JOSEPH PIMBLEY: How I satisfied myself was, the table in the database on its own is suggestive of the in-use LOC, but to really be confident that's what it's meant to be -- because the label principal, it could be different things -- is, I went to the python code. It's really in the Python code where it actually interacts with this particular table that you can see what the code is doing and it helps you understand.
For example, and this is -- this may be a very good example -- the function in this Python code that was used to generate that column was called get LOC in use. So, in a sense, that's what computer code writers do and developers, programmers, whatever we call them, when they write their code, they try very much to name things in a descriptive way so it helps them remember -- not just somebody like me -- it helps them remember what the purpose of the function is.
I ran -- since I can read the code, I read all the lines of code to understand how it was doing its calculation, and I can see that, yes, that's what they were doing and that's what they called the principal.
MR. EVERDELL: Once you verified that the query gives you the in-use LOC for Alameda, what did you do next?
JOSEPH PIMBLEY: Well, again, ran the query in concert with understanding and validating, like you just said. I got the output results, because the result of running the query is for every day from October 21 to November 22, it would give me an amount that might be a billion dollars or so. It varied over time. So I generated that output. I compared it to what my colleague that I mentioned before, what he had gotten for his output.
It's clearly a good idea for two independent people to compare their results. I had agreed with what he had done and I agreed with the numbers that came out.
And from that I then made a plot that I have provided to this case here that shows that graph. It's really a graph over time of what that in-use Alameda line of credit is.
MR. EVERDELL: Just one more clarifying question. When you say Alameda, was there only one account of Alameda on the database or were there more than one.
JOSEPH PIMBLEY: The answer to that is, I am going to say yes and no and explain. If I run just the one dominant Alameda account that I see in the database, I get the results that I'm presenting, but I verified that. I checked that because really it's reasonable to define Alameda as a much larger number of accounts also, because there are two different concepts, an account ID and a user ID. A user like Alameda can have more than one account. So my goal was to get essentially all of the Alameda accounts. So I also ran the query using the user ID. So it's a different query and it got exactly the same answers as the first result. That's one of the ways that I checked that the queries are correct.
MR. EVERDELL: Was the data that you were extracted voluminous?
JOSEPH PIMBLEY: Was it voluminous. Three or 400 lines of output. The extracted data wasn't that bad.
MR. EVERDELL: Did you summarize those three or 400 lines in the graph --
JOSEPH PIMBLEY: Yes. They are all in the graph. So the graph is an excellent visual display of what the output is.
JUDGE KAPLAN: Mr. Pimbley, would it be accurate to understand what you said as being that the object of the exercise was to find a dollar amount of the line of credit of Alameda for each day from October 2021 through November 2022 and that for each day there was a single number? Is that right?
JOSEPH PIMBLEY: Yes, sir.
JUDGE KAPLAN: Thank you.
MR. EVERDELL: I want to show you, Mr. Pimbley, what's been marked as Defense Exhibit 1617 for identification.
MR. EVERDELL: Just for the witness, please.
JOSEPH PIMBLEY: Yes, sir. That's the appendix I recognize, the graph.
MR. EVERDELL: Is that the graph you produced from the data that you extracted?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Does this summarize the data that you extracted?
JOSEPH PIMBLEY: Yes. As I said a few times, I start in October 2021 to November 2022. Those dates are simply because that's what was in the table in the database.
MR. EVERDELL: Your Honor, the defense offers Defense Exhibit 1617.
JUDGE KAPLAN: Received.
(Defendant's Exhibit 1617 received in evidence)
MR. EVERDELL: We can publish that to the jury.
MR. EVERDELL: I am not going to spend a ton of time on that, Mr. Pimbley, but if you can just look at the graph you produced.
You see the start date on the left-hand side, is that right?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: End date on the right-hand side?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Just that first time period, just so we know what the numbers mean on the left-hand axis, what does the left-hand axis reflect?
JOSEPH PIMBLEY: It reflects in billion dollars what the in-use LOC for Alameda was.
MR. EVERDELL: In that first time period you see it from the first four months hovering around what number?
JOSEPH PIMBLEY: It's just above $1 billion.
MR. EVERDELL: I want to direct your attention to one other thing, if we can fast forward to the month of June 2022.
Do you see that month?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: What do you see with the in-use line of credit for Alameda happening, according to the graph there?
JOSEPH PIMBLEY: Well, it starts close to 3 billion in June, but then it drops very precipitously about mid June.
MR. EVERDELL: We can take that down now.
MR. EVERDELL: I want to turn now to the second and third projects that you were asked to do, Mr. Pimbley. OK.
Let's start with the second --
MS. SASSOON: Your Honor.
JUDGE KAPLAN: We will take a 15-minute break, folks.
(Jury not present)
(Recess)
JUDGE KAPLAN: Mr. Everdell, before we bring the jury in, I just want to note that we long ago left behind your estimate for 15 to 20 minutes on this witness.
MR. EVERDELL: I apologize, your Honor. I am going to try to streamline the next two more charts.
JUDGE KAPLAN: That would be a good idea.
MR. EVERDELL: Yes, your Honor.
JUDGE KAPLAN: Let's go.
(Jury present)
JUDGE KAPLAN: Defendant and the jurors are all present, as they have been throughout.
Let's continue.
MR. EVERDELL: Thank you, your Honor.
BY MR. EVERDELL:
MR. EVERDELL: Mr. Pimbley, when we left off we finished the first project, right?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Let's talk briefly about the second and third projects you were asked to do.
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: Did both of those projects involve pulling data from the database?
JOSEPH PIMBLEY: Yes, they did.
MR. EVERDELL: Did you create visual representations of the data that you pulled for those two projects?
JOSEPH PIMBLEY: Yes, sir, I did.
MR. EVERDELL: I want to show you what's been marked for identification as Defense Exhibits 1618 and 1619, if we can put those side by side.
Do you see what's in front of you, Mr. Pimbley?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Do you recognize what those are?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: What are those?
JOSEPH PIMBLEY: These are two pie charts that I made from data extracted from the AWS database.
MR. EVERDELL: And these relate to the second and third projects you were asked to do?
JOSEPH PIMBLEY: Yes, they do.
MR. EVERDELL: Your Honor, the defense offers Defense Exhibits 1618 and 1619.
JUDGE KAPLAN: Received.
(Defendant's Exhibits 1618 and 1619 received in evidence)
MR. EVERDELL: We can publish those to the jury.
MR. EVERDELL: Mr. Pimbley, let's start with the one on the left. Can you explain what this second project was.
JOSEPH PIMBLEY: Yes. I was asked to run a query that would find the total balances on the AWS database of all accounts but excluding Alameda and FTX entities.
MR. EVERDELL: What do you mean by balances?
JOSEPH PIMBLEY: It's essentially the dollar -- I'm sorry. It's -- balances is, first, the number of coins that each account holds, and coins is a term that they use in the database, and by reading the database and the code you realize it's a different way to say it would be the currency. I use the word coins, but it would be easier to understand if I said currency.
The dominant currency there is just U.S. dollars. You can own U.S. dollars in your FTX account, as per the database that I see, but you can also own other currencies, like you can own Japanese yen, you can own Swiss Francs, you can own Bitcoin. Bitcoin is a cryptocurrency. If we just think of all these as being just currencies, then there is a specific coins table in the database that lists all the possible coins. It actually has 700 entries, so one could say there are hundreds of coins that you may own.
When I say total balance, what I really mean and what the query does, it says, look at each account that's not Alameda and not FTX and what's the dollar value of the sum of all of its coins that it owns. So if it owns Japanese yen, you have to convert that yen into dollars to give its dollar equivalent. If you own Bitcoin in the account, you convert that back to dollars, and you add up all those dollars and the conversion factors are in the database itself, so using the data that's in the database.
MR. EVERDELL: Once you get the balance number, what was the next part of the analysis?
JOSEPH PIMBLEY: Once I get the balance number properly excluding Alameda and FTX accounts, then it was to please form some different groups on that. For example, the first group on the left said, I get the total balances, but I am also going to restrict ourselves to only four coins, again, four currencies, if I can call it that, out of the 700 or so. But the four that I was asked to isolate were the dollar, U.S. dollar, Ethereum, Bitcoin, and Tether as the names -- these are often --
MR. EVERDELL: Just to be clear, you are tallying up the coin holdings for these four categories of coins for the users on the FTX database that are not Alameda and not FTX?
JOSEPH PIMBLEY: That's correct.
MR. EVERDELL: What about the one on the right?
JOSEPH PIMBLEY: The one on the right says the same thing. Everything is the same except instead of just the four coins, take all the coins, all the balances.
MR. EVERDELL: What do the different portions of the pie chart represent?
JOSEPH PIMBLEY: And then that's a separate way to get -- create subgroups. So the two -- each pie chart has two groups. One group says, take only those accounts that have something called spot margin, that the accounts are identified as being enabled for spot margin or spot margin lending or have -- the accounts have had futures activity, which means the accounts have done some long or short of futures at some point in time, according to the database.
MR. EVERDELL: What is the other category?
JOSEPH PIMBLEY: The other category is simply those accounts that fall out of the first category.
MR. EVERDELL: In the charts that we are looking at here, which category, blue or red, is the one that have the spot margin enabled and the spot margin lending enabled versus everything else?
JOSEPH PIMBLEY: It is the blue which is the larger in both cases and the percentages are about the same in both cases, but the blue larger one has those spot margin enabled, spot margin lending enabled.
MR. EVERDELL: Just a few clarification questions. What are the dates of the balance information for these charts that are reflected here?
JOSEPH PIMBLEY: This would be the snapshot date in November 2022.
MR. EVERDELL: When you talk about the coins, just looking at the left-hand side with the four coins you selected, one of them is USD?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: What does that refer to?
JOSEPH PIMBLEY: That is the dollar --
MR. EVERDELL: It includes currencies, not just cryptocurrencies?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: ETH is there. What does that refer to?
JOSEPH PIMBLEY: ETH. I believe that's Ethereum.
MR. EVERDELL: BTC?
JOSEPH PIMBLEY: Bitcoin.
MR. EVERDELL: And USDT.
JOSEPH PIMBLEY: Is Tether.
MR. EVERDELL: Let's just look at the numbers you were able to get to with your database extraction.
Looking at the left-hand side, which is the four coins, how much total were the balance information as of that date in November that you found from doing your data pull?
JOSEPH PIMBLEY: That was like 5.8 billion.
MR. EVERDELL: Of that 5.8 billion, roughly, how much of that were in accounts that were enabled for spot margin, spot margin lending and had futures activity?
JOSEPH PIMBLEY: About 4.54 billion was in that category.
MR. EVERDELL: And then the rest left over is 1.3 billion in the other category?
JOSEPH PIMBLEY: That's correct.
MR. EVERDELL: What are the rough percentages breakdown?
JOSEPH PIMBLEY: 78 percent in the category was spot margins, spot margin lending enabled and the futures activity, 22 percent that are not in that category.
MR. EVERDELL: Skipping over to the right-hand side where you were considering all currencies, all coins, what are the total number -- what is the total balance number you arrived at as of that date in November?
JOSEPH PIMBLEY: That total was $8.9 billion.
MR. EVERDELL: Again, the portion that was with those categories enabled is what?
JOSEPH PIMBLEY: 6.91 billion.
MR. EVERDELL: The rest is 2.03, right?
JOSEPH PIMBLEY: Yes.
MR. EVERDELL: One moment, your Honor.
Very briefly, your Honor.
MR. EVERDELL: Mr. Pimbley, you were asked to perform these queries that generated the charts that we saw, correct?
JOSEPH PIMBLEY: Yes, sir.
MR. EVERDELL: Did you do any independent analysis in the database apart from verifying the data that you did for these projects?
JOSEPH PIMBLEY: Yes. First, primarily run the query, validate that it works and it gets the results that not just are consistent with what our colleague may have found, but also do the results make sense. The queries do generate the output without -- very clearly.
The second part is, when there is an intent of what the query should be, I did as much as I could to look through the code as well to understand where the code was interacting with that particular data I was extracting, that the code also was consistent with my understanding of what it should mean.
MR. EVERDELL: Understood.
Apart from that, did you evaluate the opinions of anybody else involved in this case as part of your work on the case?
JOSEPH PIMBLEY: No, sir.
MR. EVERDELL: Nothing further, your Honor.
JUDGE KAPLAN: Thank you.
Cross-examination.
CROSS-EXAMINATION BY MR. REHN:
JOSEPH PIMBLEY: Good morning.
JOSEPH PIMBLEY: We have not met.
MR. REHN: I believe you testified on direct that you have never met the defendant, Sam Bankman-Fried?
JOSEPH PIMBLEY: Yes. I have testified to that, yes.
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: I had no conversations with him.
MR. REHN: So you don't know what was in the defendant's head when he was running FTX and Alameda, right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, sir.
JOSEPH PIMBLEY: No, sir.
MR. REHN: And you don't know what information he was actually looking at while he was running FTX and Alameda, right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, sir.
MR. REHN: And you can't say whether the queries you ran in the FTX database were ever actually used by the defendant at any point in time, can you?
JOSEPH PIMBLEY: No, sir.
JOSEPH PIMBLEY: No, sir.
JOSEPH PIMBLEY: That is correct, I did not.
JOSEPH PIMBLEY: That is correct, I did not.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That is correct, I did not.
MR. REHN: So your testimony is based on certain queries that you ran on certain tables in the FTX database?
JOSEPH PIMBLEY: That's correct.
MR. REHN: But you don't know how those tables were actually used in the course of business at FTX, right?
JOSEPH PIMBLEY: That's correct.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, sir.
MR. REHN: So you didn't investigate, for example, whether the decisions were made to include or exclude particular data from the database?
JOSEPH PIMBLEY: No, sir.
MR. REHN: So you can't say that the particular tables that you queried contain all the relevant data for your analysis?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That's correct.
MR. REHN: What your testimony is, is that if you run these queries on these particular tables in the FTX database, you get these results. Is that fair to say?
JOSEPH PIMBLEY: That is correct.
JOSEPH PIMBLEY: Correct.
MR. REHN: In fact, I believe on direct you said you had zero knowledge of this case aside from the data you extracted from the FTX database, is that right?
JOSEPH PIMBLEY: Well, I believe I included headlines. I have seen some headlines. Other than that, correct.
JOSEPH PIMBLEY: Yes, sir.
JOSEPH PIMBLEY: It was -- I presume so, yes, that they -- I presume they did not use any outside contractors, consultants, or whatever. I just don't know. I should say I don't know who maintained the data in the FTX database.
MR. REHN: Is it fair to say the database doesn't, for example, hold any actual cryptocurrency inside of it?
JOSEPH PIMBLEY: I don't want -- I believe almost certainly the answer is no, but I can't say I can testify that I know that's the case.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained as to form.
MR. REHN: As far as you know, the FTX database does not hold actual funds; it just has information about FTX, is that right?
JOSEPH PIMBLEY: As far as I know, yes.
MR. REHN: And you did not look at any bank statements in connection with your testimony this morning, correct?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: I did not look at any bank statements, no.
MR. REHN: And you did not compare the amount of customer deposits held in bank accounts to the amount that was reflected on customer balances within FTX, did you?
JOSEPH PIMBLEY: I did not.
MR. REHN: So nothing in your testimony would address whether there was a deficiency between what was in FTX, Alameda, and North Dimension bank accounts and what was reflected in the FTX database being owed to FTX customers.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: The fiat@ account was not part of my testimony, that's correct.
MR. REHN: In fact, you did not engage in any financial tracing for your testimony today, did you?
JOSEPH PIMBLEY: I'm sorry. Could you repeat the question. Me engaging in financial trades? I want to make sure I understand your question.
MR. REHN: You did not engage in any financial tracing in connection with your testimony this morning, correct?
JOSEPH PIMBLEY: Of course not, no, no, sir.
MR. REHN: For example, you did not address whether FTX customer fiat currency deposits were used to pay for investments.
JOSEPH PIMBLEY: I did not address that, no, sir.
MR. REHN: And your testimony did not address whether FTX customer fiat currency deposits were used to pay for real estate, correct?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Correct. My testimony did not address the issue you just named.
MR. REHN: And it did not address whether those customer deposits were used to purchase Robinhood shares for the defendant, correct?
JOSEPH PIMBLEY: That's correct.
MR. REHN: Moving on from fiat currency tracing, you also did not engage in any cryptocurrency tracing, did you?
JOSEPH PIMBLEY: I did not engage in any cryptocurrency tracing.
MR. REHN: For example, you didn't look at the Blockchain to see what was actually happening to the cryptocurrency?
JOSEPH PIMBLEY: I did not look at the Blockchain, no, sir.
JUDGE KAPLAN: Asked and answered.
MR. REHN: To be clear, your testimony did not address whether the amount of crypto held in FTX crypto wallets was less than the amount of crypto that it owed to FTX customers?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That's correct.
MR. REHN: And you also did not look at whether FTX customer crypto deposits were used to pay for Alameda expenditures?
JOSEPH PIMBLEY: I did not look at that, no, sir.
MR. REHN: And you did not do Blockchain tracing, for example, of how Alameda's lenders were actually paid back in the summer of 2022, is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Apologies, sir. Could you repeat the question, because I just want to make sure I'm answering your question.
MR. REHN: You did not do any tracing on the Blockchain of how Alameda's lenders were actually paid back in the summer of 2022?
JOSEPH PIMBLEY: I did not do any tracing on the Blockchain of the nature you just described.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained as to form.
MR. REHN: In your testimony you did not offer an opinion as to where the crypto that was used to pay Alameda's lenders back came from, is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Again, I apologize. Could you summarize that question again, please.
MR. REHN: In your testimony you did not offer an opinion about where the crypto that was used to pay Alameda's lenders back came from, is that right?
JOSEPH PIMBLEY: I did not offer an opinion on that topic, no, sir.
MR. REHN: So you can't say one way or another whether those payments came from FTX customer funds, right?
JOSEPH PIMBLEY: I can't say. I also have no knowledge of that, so I certainly can't say.
MR. REHN: Let's look at what you did do.
I want to start by asking about your first slide, which I believe was marked as Defense Exhibit 1617 of Alameda's LOC, I think it said, on the chart. Is that right?
JOSEPH PIMBLEY: Sounds right, sir.
JOSEPH PIMBLEY: That's my interpretation, yes, sir.
MR. REHN: And I believe your testimony is that this chart shows the amount of line of credit that was used by Alameda between October 2021 and November 2022, is that right?
JOSEPH PIMBLEY: Yes.
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: It was called LOC_interest_charges was the name of the table.
JOSEPH PIMBLEY: It was -- first, that was suggested to me by the colleague I worked with. He looked at that. I also looked at all the tables in the database, including that one, to make my own -- satisfy myself that that was the -- really I think it was the only source of in-use LOC that I knew, but I did look to see if other tables were also relevant.
JOSEPH PIMBLEY: That's correct.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: Is it fair to say you don't have any information from the company that would indicate what the actual purpose of that LOC interest charges table was, do you?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained. I think you have covered this.
MR. REHN: In connection with this table you did not run a query of the actual balances of Alameda accounts, is that right?
JOSEPH PIMBLEY: I did run queries of, again, on the balances table, if that's your question. I am not sure it is.
JOSEPH PIMBLEY: Well, that's true, yes.
MR. REHN: But to generate what you showed us in Defense Exhibit 1617, you did not actually query Alameda's balances, is that right?
JOSEPH PIMBLEY: I would not agree to that statement. If the question is, did the specific query I used reference a table other than LOC interest charges? No, that query didn't reference a different table.
But I'm wondering if you are asking, did I ever access Alameda balances anywhere else? I am not sure.
MR. REHN: No. I'm asking about the basis for the table that you presented during your direct testimony.
JOSEPH PIMBLEY: The basis for the table is my belief, judgment, opinion that that is the appropriate table that has the in-use LOC for Alameda, and also for other entities if I had chosen to look at other entities.
MR. REHN: Just to be clear, that table does not tell us what the overall aggregate balance of Alameda accounts was, is that right?
JOSEPH PIMBLEY: That is correct, it does not tell you that.
JOSEPH PIMBLEY: Yes. A different query in the balances table would be more appropriate for that question.
MR. REHN: By the same token, it doesn't tell us what the aggregate balance in Alameda accounts that had the allow-negative flag was, is that right?
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: It does not.
MR. REHN: That means those balances in those Alameda accounts could have been more negative than the line-of-credit amounts that are reflected in the table you presented, is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Let me have the question back, please.
(Record read)
JUDGE KAPLAN: Rephrase it.
MR. REHN: So the aggregate Alameda balances could be more negative than the amount that's reflected in the line-of-credit balances that you presented, is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Yes. If the concept of a negative asset balance is appropriate. As you know, the LOC -- the in-use LOC is a debt. It's a liability of whoever has drawn that amount. That's a liability of Alameda. Now you are asking me to compare a liability of Alameda to assets of Alameda, and of course I say that assets and liabilities are two ends of the same coin, if I can use that term.
But if you are asking me to go into something I didn't study, which is that Alameda assets in some sense could be negative, which is -- it's an ambiguous way to talk about the financial condition of a company. Since I have not studied it, I really don't think I can give you much more of an opinion.
MR. REHN: You didn't look at the aggregate balance of Alameda accounts that had the allow-negative feature.
MR. EVERDELL: Objection. Asked and answered.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That is correct.
MR. REHN: On the table that you did generate, you identified the LOC in-use amount for Alameda accounts, is that right?
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: I did it two ways and I got the same answer both ways. One was with user ID and one was with just the single-account ID for Alameda Research.
JOSEPH PIMBLEY: That was the account ID, was number 9.
MR. REHN: The account ID was number 9. If you put that account ID into the LOC in-use query, you would get the results that you presented?
JOSEPH PIMBLEY: That is correct.
MR. REHN: Now, you said that you looked at the code that underlied that particular query, is that right?
JOSEPH PIMBLEY: Yes, sir.
JOSEPH PIMBLEY: Yes, sir.
But may I clarify my previous answer? May I reanswer? It would be more clear if I answered that previous question differently.
MR. REHN: I am not sure which previous question. If you could just answer my question.
Did you examine the code to determine how the data responsive to your query was generated?
JOSEPH PIMBLEY: Yes. I like the way you asked that question. It was a little bit different than how you phrased it before. I am going to say yes to that question.
JUDGE KAPLAN: Mr. Rehn, you're getting a good grade.
MR. REHN: If we could pull up for the witness Government Exhibit 1737.
If we could expand the part, the top there.
MR. REHN: Mr. Pimbley, do you recognize this as a portion of the FTX code that would be used to return the data responsive to the query you ran?
JOSEPH PIMBLEY: Yes, I do -- it doesn't have all the identifying information that's helpful for me looking at code, like the name of the file of the code, but, yes, this is a function that I remember from my review.
JOSEPH PIMBLEY: Yes, sir.
JOSEPH PIMBLEY: Yes, sir. But I would -- if you don't mind my qualifying, there were two versions of this function that are very close to each other in the file, and both are important, but this is one of them, yes.
JUDGE KAPLAN: Isn't this already in?
JUDGE KAPLAN: Hearing no objection, it's received.
(Government Exhibit 1737 received in evidence)
MR. REHN: Mr. Pimbley, do you see that this has some code relating to if there is a query in that get LOC in use relating to user account PMMID?
JOSEPH PIMBLEY: Yes.
JOSEPH PIMBLEY: Yes. Elsewhere in the code that variable is hard coded to mean 9, and 9 is the Alameda account ID, so, yes, that is the Alameda account ID.
MR. REHN: So this portion of the code tells how to generate data responsive to a get-LOC-in-use query if the query relates to Alameda, is that right?
JOSEPH PIMBLEY: I am going to say yes, but I am going to, if I can, modify how I might say that if I were you. I would say, in the case where the account happens to be this one Alameda account, this code says do something a little bit different.
MR. REHN: And this is the Alameda account that you used to run the query that supports the data in your chart, is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Go ahead.
MR. REHN: So this Alameda account, you used that account to query this database to generate the data in your chart, correct?
JOSEPH PIMBLEY: That is correct.
MR. REHN: If we look at the code here, it says: If that user account is identified, it directs the database to pull some other accounts. Is that right?
JOSEPH PIMBLEY: That's what it says, yes, sir.
MR. REHN: And that means those are accounts that wouldn't be pulled if it was only pulling the Alameda accounts, is that right?
JOSEPH PIMBLEY: I think that's a fair statement. A code is essentially -- I'm not going to disagree strongly with what you said. The code is essentially saying, look, in the case of this Alameda account, we are going to -- for the purposes of this function, we are going to imagine or state or hypothesize or decree that additional accounts need to contribute.
I am going to answer yes, but it is a deliberate aspect of the code that whoever the developer was, they did this on purpose, yes.
MR. REHN: You are saying the developer of this code deliberately chose to add two other accounts to a query for the Alameda account, is that right?
JOSEPH PIMBLEY: Yes. In fact, you can look at the account IDs all the way over to the right-hand side. That's true, yes.
JOSEPH PIMBLEY: I did have to do that, yes.
MR. REHN: So looking at that first one, the one that says 1405310, is that the Cotton Grove trading account?
JOSEPH PIMBLEY: I remember that one of the two was called Cotton Grove. I don't know which one, but I will submit that you may be right, yes.
JOSEPH PIMBLEY: I couldn't find the identity of that within the code, but it had some other special significance elsewhere in the code, but I am going to answer no and say I didn't know anything about Alameda ventures.
MR. REHN: In your chart, to be clear, it presents the data responsive to this query, in other words, including these two accounts in the Alameda line of credit, is that right?
JOSEPH PIMBLEY: I believe that is true, yes.
MR. REHN: And so the balances of those accounts would affect the numbers that are on the chart at Defense Exhibit 1617?
JOSEPH PIMBLEY: When you say balance, I wouldn't agree to balance. I would say that -- as you can see at the end of the function, where it says return, it's saying now take this information and call this other function, but my point being that it's going -- the code will say, of all these accounts, including these two, then perform the in-use LOC calculation, including those accounts. That's what it will say.
MR. REHN: In other words, the data you presented on Defense Exhibit 1617 was using a query that pulled the Alameda account plus these two additional accounts.
JOSEPH PIMBLEY: Yes. The query was doing exactly what we wanted it to do, which was identify Alameda, and this is what the code does. I see that point, yes.
JOSEPH PIMBLEY: Well, I believe -- actually, I believe I did, not for balances. I might have looked to see how do these contribute to the in-use LOC because that's what mattered. My recollection -- again, I'm not -- my recollection is that they had no material impact or was actually zero, but it's a little bit hazy for me.
MR. REHN: That's not reflected anywhere in the chart or the data underlying the chart that you presented today, is it?
JOSEPH PIMBLEY: No. Actually, let me be clear about that. The chart says I ran this query, I got this result.
JOSEPH PIMBLEY: No. What it says is, the AWS system treated -- I ran this query saying Alameda account, account ID equals 9, and that's the definition the data system is using. I'm saying the code plus the database itself is leading me, as the user -- even somebody who gets to see the source code is leading me to say, this is how we define Alameda.
The query is doing what it was supposed to. The question is how expansive is this definition of Alameda.
(Continued on next page)
BY MR. REHN:
MR. REHN: But just to be clear, you did not prepare a separate chart showing what it would have looked like without these two extra accounts included that the code——
JOSEPH PIMBLEY: Well, that's correct, because my recollection is that the diligence I did said this had no meaningful impact, or it was negligible or zero.
MR. REHN: Now before you did your database queries, did you review any of the notes of the government's interviews with Gary Wang?
JOSEPH PIMBLEY: No, sir. I had no access to anything other than the code base and the AWS database, so no, sir.
MR. REHN: The defense counsel didn't provide you with any notes of interviews with Gary Wang or Nishad Singh?
JOSEPH PIMBLEY: The answer is no. My understanding of my assignment from my defendant-client was stick to the database and the code base. That was my assignment.
MR. REHN: In general, when you're trying to evaluate data in a code base, is it fair to say that understanding why it was designed in a particular way is relevant?
JOSEPH PIMBLEY: It is relevant, yes.
MR. REHN: So if you had known that Mr. Wang said that the defendant directed the inclusion of these subaccounts to make the used line of credit appear lower, would that have been relevant to your analysis?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That statement, representation, allegation, whatever I call it, is unknown to me. I have no knowledge of that.
MR. REHN: I'm saying if you knew that that was an aspect of how the code was designed, would you have taken that into consideration in your analysis?
JOSEPH PIMBLEY: No, sir. Simply because my analysis here——you say would it is have changed my analysis. My role here was, to my defendant counsel-client, please pull this data from the database, okay, as——you know, with this intent of Alameda account. So I don't think it——I don't think I could start using what did the developer actually intend when he or she wrote the code. I think that would have been pretty speculative of me and pretty far afield.
MR. REHN: Okay. Now let's go to an account that is not included in the query you ran.
Are you aware of an account called fiat@ftx?
JOSEPH PIMBLEY: I have, in my——yes, in my work through the database and the code, I had some awareness of that, yes.
MR. REHN: And that is an entry in the FTX ledger that reflects how much fiat currency was owed to FTX customers?
MR. EVERDELL: Objection, your Honor. Beyond the scope.
MR. EVERDELL: We had a lot of discussion about things that are not included but——
JUDGE KAPLAN: I'm sorry?
MR. EVERDELL: Your Honor, beyond the scope.
JUDGE KAPLAN: And why is it within the scope?
MR. REHN: Your Honor, I'm attempting to cross-examine the witness regarding what this chart actually shows and what it doesn't show.
JUDGE KAPLAN: Well, I think you can rephrase it to ask more directly.
BY MR. REHN:
MR. REHN: So the fiat@ftx account was not included in the line of credit chart that you created to show Alameda's line of credit, correct?
JOSEPH PIMBLEY: That is my understanding, that it was not included in either of two ways I ran that query, but that's my——to be honest, since I don't have the code in front of me or the——and the table——the database tables in front of me, I can't say for sure, but my impression was it was not included, no.
JOSEPH PIMBLEY: No, sir.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: In the course of your review of the FTX database, did you look at the FTX fiat old account?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
He has told you what he did at length.
MR. REHN: Yes, your Honor. I was attempting to evaluate whether he included this account as well, but we can move on.
JUDGE KAPLAN: Yes.
BY MR. REHN:
MR. REHN: So looking at your chart, in terms of what is actually included here, is it fair to say you calculated that Alameda was in fact using more than a billion dollars on its line of credit throughout 2022?
JOSEPH PIMBLEY: Yes, sir. Maybe there's a very brief dip below a billion dollars at Jan 7, but it's fair to say a billion dollars or more, yes, sir.
MR. REHN: And your analysis also shows that the amount of its line of credit that Alameda was using generally went up over the course of 2022; is that correct?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
The chart speaks for itself. We can all see it.
MR. REHN: So looking at October 2022, Alameda is using between 4 and $5 billion of its line of credit; is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: Is it fair to say that based on your calculation, Alameda went from having a $1 billion use on its line of credit to having more than $4 billion between January and October of 2022?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Look, you have an opportunity to do a closing argument in this case, and the numbers are right in front of you. Let's move it on.
MR. REHN: Does Defense Exhibit 1617 list of Alameda accounts that were used to generate this chart include any accounts under the email address seoyuncharles88@gmail.com?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Yeah, I'm sorry. I know you gave an account or a chart number, but can you ask me, are we talking about this in use loc query?
MR. REHN: I'm talking about, for the account you used to generate this particular chart, did it include an account under the email address seoyuncharles88@gmail.com?
JOSEPH PIMBLEY: I have seen that account name, okay, and I believe it was not part of this Alameda group, but I can't——as I'm sitting here, I don't have the database in front of me, and I can't testify to that precisely.
MR. REHN: When you were preparing this chart, did you consider whether any other customer of FTX had the ability to have a line of credit that ran into the billions of dollars?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: Your Honor, he's testified about the line of credit that Alameda had. I think some cross-examination on whether other customers had similar lines of credit is warranted.
JUDGE KAPLAN: Look, he has told you, and he told Mr. Everdell, that essentially the charts he prepared added up certain numbers in the table, date by date, and he told you what numbers in the table were added up, and obviously the ones that he did not enumerate as having been added up are not included. Why isn't that self-evident at this point?
JUDGE KAPLAN: Okay.
MR. REHN: Let's talk about your second and third charts. These deal with the sum of FTX user balances for user accounts; is that right?
JOSEPH PIMBLEY: Yeah. If I can——I think I'm probably saying what you just said, but it's the sum of balances of the non-Alameda, non-FTX, other users.
JOSEPH PIMBLEY: That's correct.
MR. REHN: And then the third chart was the same concept but it included the sum of FTX user balances for all coins; is that right?
JOSEPH PIMBLEY: Yes. Again, with the caveat, I don't know the term "FTX user balances." User——again, we're excluding Alameda and FTX entities, but I guess that's clear.
JOSEPH PIMBLEY: Yes, sir.
MR. REHN: So that means these are the values of FTX user balances as of around November 11, 2022?
JOSEPH PIMBLEY: Yes, sir.
MR. REHN: That's when FTX declared bankruptcy and stopped processing customer withdrawals; is that right?
JOSEPH PIMBLEY: I don't know that. I don't have that information.
MR. REHN: But your testimony with respect to these two charts is just a snapshot of that moment in time?
JOSEPH PIMBLEY: Yes, sir, and it's the only snapshot I received in evidence in this case, yes, sir.
MR. REHN: You didn't look at what was happening with user balances at any earlier point in time; is that right?
JOSEPH PIMBLEY: Essentially, no, but I will volunteer and say that there was a table in the database that tried to capture balance history.
JOSEPH PIMBLEY: For these charts——for these charts, I did not use that table, correct.
MR. REHN: And to calculate the value of the customer balances that was reflected on your charts, you didn't look at any outside sources, right?
JOSEPH PIMBLEY: Correct.
JOSEPH PIMBLEY: That's correct.
JOSEPH PIMBLEY: Yes, sir.
MR. REHN: And so you didn't take, for example, the value of Bitcoin from an outside source and then multiply it by the balance of Bitcoin that FTX users had?
JOSEPH PIMBLEY: I did not take the value of Bitcoin, say, in US dollars from an external source. I used precisely what was in the AWS database, yes, sir.
MR. REHN: And you did not evaluate whether the valuations inside the FTX database reflected real-world values, right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: That is correct.
MR. REHN: And in fact, around the time of FTX's collapse, there were some pretty significant fluctuations in crypto values, right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
JOSEPH PIMBLEY: I——I can't testify to that, no. No.
MR. REHN: It's fair to say your analysis did not evaluate whether the approx fair value in the FTX database is a realistic valuation of FTX user balances?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: For generating the charts, the pie charts that we looked at, you just took the internal FTX database valuation, right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: And then what you did was, you identified the accounts with certain things enabled in them to generate the pie chart; is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: Yes, sir.
JOSEPH PIMBLEY: Yes, sir.
MR. REHN: And accounts with spot margin enabled and accounts with spot margin lending enabled, right?
JOSEPH PIMBLEY: That's correct.
MR. REHN: And then you put all of those accounts together and you took the total balance of those accounts; is that right?
JOSEPH PIMBLEY: Yes, remembering, as I'm sure you have, that we're excluding Alameda and FTX accounts, but——but, yes, the grouping you just mentioned——spot margin enabled, spot margin lending enabled, or——these are ors——or futures activity——that formed one group, and then the second group was just all of the——all those cases that fell outside the first group.
MR. REHN: So starting with futures trading, those are customers who engaged in buying and selling of crypto futures on FTX; is that right?
JOSEPH PIMBLEY: Sir, that's actually going beyond my specific knowledge of how FTX worked. I think to be fair to me and to——to the entire court, I should just be clear. We developed a specific way to query the database to isolate those accounts that had never traded futures at all versus——and of course the others would be those that had at one point in time traded futures, and it——these just come however the database interpreted that. Did futures include Bitcoin futures or these other futures; it's however the table in the database——whatever signals it gave is what we used.
MR. REHN: But customers who participated in futures trading but did not participate in the spot margin program, they did not agree to lend out their assets; is that right?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: Did you evaluate whether customers who engaged in futures trading on FTX agreed to lend out their assets?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: So moving on to the spot margin program. So spot margin lending enabled, is that the feature that allows FTX customers to lend money on the exchange?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained. Look——
JUDGE KAPLAN: I understand that. He explained in detail how he did it. And it has, I imagine you will argue, certain inherent limitations.
Okay. You know, when I was a kid, I worked in my father's deli, and if he gave me the assignment to go add up the column that showed how many pounds of smoked fish we sold last month and I gave him a number, it means pastrami is not included, neither is cole slaw, or macaroni salad, and we can keep on going through every SKU in the deli. But we're not taking the time to do that.
BY MR. REHN:
MR. REHN: Can I ask, who decided to exclude these three, or to lump together these three categories?
MR. EVERDELL: Objection, your Honor.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: It was the Cohen & Gresser request to me that made that decision. I did not make that decision.
MR. REHN: And you didn't evaluate whether there's any significance to putting together those three types of accounts, did you?
JOSEPH PIMBLEY: No, sir, I did not.
MR. REHN: And you did not form an opinion on why those three types of accounts would be lumped together on this chart, did you?
JOSEPH PIMBLEY: No, sir, I did not.
MR. REHN: And just to be clear, you did not conduct any analysis of how much money was actually being lent in the FTX spot margin lending program, did you?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
JUDGE KAPLAN: This is getting to be time to wrap it up.
MR. REHN: Does this chart reflect how much money was actually being lent in the spot margin lending program?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, it does not.
MR. REHN: And you also did not evaluate how much money Alameda was actually borrowing in the spot margin program, did you?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Sustained.
MR. REHN: Did you present any opinions or charts today about how much Alameda was actually borrowing in the spot margin program?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, sir.
MR. REHN: And so your testimony did not address whether Alameda's balances on FTX were connected to borrowing through the spot margin program, did it?
MR. EVERDELL: Objection.
JUDGE KAPLAN: Overruled.
JOSEPH PIMBLEY: No, sir.
JUDGE KAPLAN: Thank you.
Mr. Everdell.
MR. EVERDELL: Nothing from the defense, your Honor.
JUDGE KAPLAN: Thank you. Thank you, sir.
JOSEPH PIMBLEY: Thank you.
JUDGE KAPLAN: You're excused.
(Witness excused)
JUDGE KAPLAN: All right. We'll break for lunch till 2.
Counsel, please remain for a minute.
COURT CLERK: Would the jury please come this way. Bring your notebooks with you, please.
(Continued on next page)
(Jury not present)
JUDGE KAPLAN: Be seated, folks.
Okay. Mr. Cohen, is there anything else on the defense case other than the testimony of your client?
MR. COHEN: That's it, your Honor.
JUDGE KAPLAN: Okay. We'll resume at 2:00.
MS. SASSOON: Your Honor, at this point do you have a sense of when you intend to conduct the hearing?
JUDGE KAPLAN: Based on what Mr. Cohen has told me, I do not expect him to finish the direct on the subjects other than those that are implicated in the hearing today, and we will probably have to finish that tomorrow morning and we will then have the hearing. Now I am going to think over the lunch hour about what to do with the jury while we're having the hearing, which might be to send them home and come back Monday, and it might not.
Mr. Cohen, how long a hearing should I be planning for?
MR. COHEN: If we could confer over lunch and I can come back with a more precise answer for you, your Honor?
JUDGE KAPLAN: Okay. Thank you. 2:00.
MS. SASSOON: And your Honor, perhaps, depending on the length, we could conduct the hearing in the morning and have the jury come in a little later, even if the rest of the direct hasn't been completed.
JUDGE KAPLAN: That's possible too.
COURT CLERK: All rise.
(Luncheon recess)
AFTERNOON SESSION 1:57 p.m.
(In open court; jury not present)
JUDGE KAPLAN: Okay. Counsel, what's the estimate as to how much time we need for a hearing?
MR. COHEN: I think, your Honor, 60 to 90 minutes.
JUDGE KAPLAN: Government?
MS. SASSOON: We expect to cross-examine maybe for 45 minutes.
JUDGE KAPLAN: Given those estimates, it seems to me the best thing to do may be to send the jury home now, do the hearing this afternoon, and start fresh tomorrow. What do you think?
MR. COHEN: Whatever your Honor wants. We're ready to start, if you'd like.
JUDGE KAPLAN: I understand that.
MS. SASSOON: Government's amenable to that.
JUDGE KAPLAN: Okay. That's what we're going to do. Then we'll bring the jury in and give them what I imagine they're going to welcome as news, as soon as they're all here.
(Continued on next page)
(Jury present)
JUDGE KAPLAN: Okay. The defendant and the jurors all are present, as they have been throughout.
Ladies and gentlemen, I have some scheduling information for you, and what may be a little bit of a surprise. There is something that has to happen at this point in the trial that will take a couple of hours, and that is not something that concerns you. In consequence, you've got the rest of the day off. We'll see you at 9:30 tomorrow morning. You will hear the rest of the defense case starting at 9:30 tomorrow morning. And you can reasonably expect that you will get this case to decide in the first two or three days of next week. We're in the home stretch. Of course I can't guarantee it. No insurance. But there we are.
Okay. So have a pleasant afternoon. And I want to reiterate, do not read or get informed in any way about what's going on. You've heard all those instructions before. It's especially important now.
Okay. Thank you.
COURT CLERK: Would the jury please come this way and bring your notebooks with you.
(Continued on next page)
(Jury not present)
JUDGE KAPLAN: Okay. Be seated, folks.
Now just so that everyone who has taken the trouble to get here today understands what is going on and why, there are a number of areas of potential testimony from Mr. Bankman-Fried that the defense wishes to elicit. The government asserts that I shouldn't hear any of it, or, to be more precise, that the jury shouldn't hear any of it. And despite a great deal of effort on the part of everybody concerned, the amount of information that I have to date is, in my judgment, inadequate to resolve the admissibility of this testimony, in significant part because it's not sufficiently detailed or specific.
I have the authority under the rules of evidence to conduct a hearing so that the defendant can put in the evidence for my ears alone, following which I'll be in a position to rule one way or another as to whether the evidence is admissible before the jury, in whole or in part, and if in part, to what extent; and once that happens, we will then be able to proceed with Mr. Bankman-Fried's testimony before the jury, whatever the scope of it winds up being. That's what's happening.
And so Mr. Cohen, I take it you're going to call your client to testify in this hearing; is that right?
MR. COHEN: Yes, your Honor. The defense calls Sam Bankman-Fried.
COURT CLERK: Please remain standing for a moment and please raise your right hand.
(Defendant sworn)
COURT CLERK: Thank you. Please be seated.
MS. SASSOON: Your Honor, what's your practice when it comes to objections during a hearing of this nature?
JUDGE KAPLAN: Well, not having had a hearing of this nature in quite a long time, if ever, I will hear whatever objections there are. I don't have a practice, in other words. Seems to me appropriate to hear whatever is to be said. SAM BANKMAN-FRIED, the Defendant, having been duly sworn, testified as follows: