054: Rishi Narang – Components of a Black Box, Humans Versus Computers, and High Frequency Trading

Chat with Traders is brought to you by Trade the Pool. Did you know that every decade the market reinvents itself? Online brokers opened the doors, mobile apps made trading seamless, and commission free trading erased barriers. Now a new era has begun. Meet, trade the pool, limited risk trading. And now you also have unlimited time to reach the profit target. From now on, your trading risk is capped. and your trading opportunities are limitless.

Trade the pool funds home-based stock traders with up to$200,000 in buying power. That means you can trade larger positions and scale your strategies without risking your own savings. It's time to trade with more capital, making it truly worth your time and effort. Ready to trade the pool? Click the link in the description and join the stock trading revolution today.

Are you ready to get serious about trading? Then join Tasty Trade, Investopedia's best platform for options trading in twenty twenty six. Options, futures, and more. Tasty Trade has everything you trade all in one platform. Get low commissions, including zero commissions on stocks. So you can keep more of what you earn. Trade smarter with advanced charting tools, a pre-built strategy selector, risk analysis tools, and more features. Visit Tastytrade.com slash.

chat for more information. Tasty Trade Inc. is a registered broker dealer and member of Finra, NFA and SIPC. The biggest secret of the best traders in the world is that they're just like everyone else. However, they've worked hard to learn the markets and discover what works and what doesn't. But how can you hear about these journeys and get in on the strategy? Here's your host, Aaron Fifefield.

Hey guys, what's going on? Welcome to the very first episode of 2016. To start the year with a bang, I have a very special guest on the podcast this week who I'd like you to meet. His name is Rishi Narang. Rishi has an impressive background and has been involved with financial markets for well over 20 years now. He originally started out as an analyst at Citibank prior to co-founding TradeWorks with his brother Minoj Narang. TradeWorks was a fintech company turned high frequency hedge fund.

Since 2005 though, Rishi has been the founding principal at T2AM, a fund of funds specializing in short-term quant trading strategies. He's also the author of Inside the Black Box and was featured in a must watch VPR documentary titled Money and Speed Inside the Black Box, which I'll include in a link at chatwithraders dot com forward slash fifty four. During our discussion, which went for well over an hour, we spoke about the non textbook definition of alpha, the components of a black box.

Where humans have an advantage over computers and vice versa. And we also explore the subject of high-frequency trading and speed. So guys, this is a really big episode. There's a lot to take in. Let's jump right to it. You're listening to the Chat with Traders podcast. I'm your host, Aaron Firefield. Here is my guest, coming to you from LA, Rishi Narang. Hi Rishi, welcome. How are you? I'm fine, Aaron. How are you?

I'm doing really well. Thanks heaps for joining me. I'm about halfway through your book Inside the Black Box, which is brilliant. And I really enjoyed the documentary you did with VPR, uh which was also titled Inside the Black Box. So all in all, it's great to be speaking with you. Hey, thanks very much and thanks for the kind words.

No trouble man. Now there's a few things I'm really keen to discuss with you today. Uh black boxes, high frequency trading, and your backstory within the hedge fund world. So let's start there. I know your involvement with trading and hedge funds dates back at least twenty years now and prior to this.

you went for a degree in economics. So share with us what motivated you to get an economics degree and what initially sparked your interest in trading. They have two pretty different answers. Uh I'll do the trading one first. So trading uh became an interest back when I was in high school. Uh there was uh not quite required, but I I can't recall exactly, but most everyone had to take economics class in in high school. I had this.

uh old war veteran, I think from World War Two, named Bob Berger, who was my economics teacher. And uh, you know, one of the projects was kind of like a stock picking game sort of project in in those days, uh, just to date myself a little bit. We picked stocks by reading the newspaper and so in the morning in the paper there would be the tables of stocks in the financial section.

And it would show you the ticker and what the price change was yesterday, and fractions of course, and uh and what the current price was and maybe the volume or something like that. And so my dad, who had been dabbling around as many dads do, with investing, uh, he and I kind of worked on that a little bit together. And the first stock pick I ever made was Nike. And this would have been around nineteen eighty seven or eighty eight. I think it was a pretty good pitch.

But that sort of sparked my interest. Uh as far as studying economics then, when I went to Berkeley, my plan was to be a constitutional lawyer. Uh it turned out that, and I mean specifically like arguing cases about constitutional law in front of the Supreme Court. That was really what I wanted to do. I went into Berkeley double majoring well my plan was to double major in math and economics.

Uh and a lot of the aspects of constitutional law I was interested in related to uh economic aspects of the law. And so uh that seemed like a good path. The math thing quickly became uh self evidently too much work, more than I was willing to do. And so I stopped taking math classes and stuck with the economics. And what I found at uh in in my undergraduate years was that econ is sort of like a very philosophical approach to money. It's um a lot of thought experiments and uh

closed environments that are manufactured just to play out thought experiments. Much of econometrics turns out to be just a lot of smart people almost wanking off because the assumptions are so silly to start with. So I didn't I didn't really like my degree that much when I was doing it. I didn't go to my classes all that much or anything.

Then I wanted a job and I'd uh gotten a job as a programmer, though I didn't really know how to program that well. And I was pretty sure I was gonna get fired any minute. Uh And instead, um I'd had an internship previously at Citibank in San Fran in in New York rather. Uh and I was living in the Bay Area near San Francisco and wanted to stay there.

didn't want to take the job in New York. Uh they were pretty keen to hire me in New York to work for this fixed income architecture desk that would have had me uh helping build the trading environments for fixed income traders on the prop desks and the cell desks at uh at city. So I just used their internet page and found what offices were in San Francisco and it turned out that they had this group called Citibank Alternative Investment Strategies.

uh or no citibank global asset management alternative investment strategies and I didn't know what the hell any of that meant. But it had a phone number and so I called it uh literally out of the blue. And as it happened, the managing director of the group

uh picked up his own phone. He had an assistant, but French he pretty much always answered the call and screened his calls very heavily. But he answered his phone. They were looking to hire a a junior. The guy had a favorable opinion of Berkeley. And the fact that I was already a Citibank employee made it very easy for him from a

human resources standpoint. So I sort of accidentally landed in my first job in the hedge fund business right out of school. Right. Okay. So what were you doing at Citibank on a day to day basis? Like what does a global market strategist position entail?

Yeah, so this was also a fund of funds business and uh at such an entity the idea was to look at look prospectively at different uh things that were unfolding in capital market. and ascertain how those things would impact our portfolio of trading strategies. And then my job was to make recommendations to the portfolio manager about how to position the portfolio to take advantage of those themes. So a couple of examples.

uh there was in Japan at the time uh or up till that time this kind of culture of uh propping up really crappy companies just on like a cultural kind of loyalty basis. Uh so let's say that you're a mentor and your mentee has set up a really terrible company. Uh you feel some obligation to look out for him, so you back it and then you keep backing it and pouring good money after bad.

Uh and so it was very difficult to short stocks and make money in Japan because they would just never go away. They would just keep getting more and more lifelines. And what happened was uh we started to see the pressures of a long bleak stock market in Japan take its toll on that very old system of uh patronage and so

uh we made the decision that it would be a good idea to go out and find some Japanese long short equity strategies. Uh also at the time I was there and it was really lucky timing in a lot of ways. So I joined there in nineteen ninety six, right out of school. And in ninety seven we had the Asian currency crisis.

Uh, in ninety eight we had the Russian debt crisis and then long term capital blowing up. And then in ninety nine, right at around the time I was leaving, was uh really getting near the peak of the dot com bubble. So we got to see lots of interesting oh and by the way, also in there was the introduction of the Euro, this Japanese stuff. So there's lots of fun events to try to figure out. Uh and we did make some interesting uh

decisions during that time. But it was very much a discretionary thing. Now I was pretty good at math and uh and some very rudimentary kinds of programming. And by that I mean like Excel formulas and some really random other craft. And so I was kind of the resident quant as well in that shop.

Okay, so where did you acquire, you know, a lot of your programming knowledge? I mean, you mentioned earlier that you did take on a position for a very short amount of time and you, you know, weren't a very good programmer at that stage. Where did you acquire sort of most of your knowledge on programming between then and sort of this point we're at right now? It's there's a combination of inside the family, my brother, five years older than me

uh went to MIT majoring in math and computer science. And this was back when Indians were meant to either become doctors or engineers. They were really not meant to be computer programmers or go into finance. And my brother did both of those things and

much to the dismay of my parents at the time, not realizing that he was in fact a bit of a vanguard among many thousands of Indians who would follow those footsteps later, uh, in the US anyway. Um Somewhat in school, uh more in high school, frankly, even than in college. Uh sitting down programming was never really something I was excited about or interested in. So I I never got good at it. And having to

Especially in those days, the programming languages were so specific about syntax. And that always tripped me up because I'm just not good at that. I don't care enough about Where the period goes. Okay. So at what point did you decide to move on from Citibank and what was your next move from there? Yeah, so the aforementioned brother uh

He called me in the beginning of the year of nineteen ninety nine and said that he wanted to start a fintech company, financial tech company. And that I had to do it with him'cause it was such a crazy good opportunity. And wanted me he wanted to bring me along for the ride. He wanted somebody he could trust. He wanted someone competent. Bear in mind I was twenty four.

Uh, and really happy. I was doing well at City. I was making more money than I ever thought I would make as a completely terrible student in college. Uh and with a great career path ahead, my boss really liked me at the time and uh but he was quite insistent.

And so eventually I capitulated and uh and quit. So that was in March of nineteen ninety nine that we We started what was then actually called StatArb Inc., Statarb being the street name for statistical arbitrage, which is one of the earlier quant hedge fund strategies. Uh, we quickly ditched that name and called it Tradeworks instead.

even then it was hard to find good names and good domains, so we had to spell it with the X at the end. And so that got going as a company that was really trying to do something that was way ahead of its time. Today it's it's uh it would have been a better idea'cause of things like Quantopian and the World Quant trading competitions and so on, but basically the idea was to make professional investing and trading tools and an analytical tools and decision support tools.

and make them available to just straight up retail investors, people who trade for a hobby.

Okay. So so were you doing any trading internally at that firm? Not until much later. So at some point we developed a lot of really interesting tech and decided, well We were having kind of trouble. This is just before the peak in the dot com bubble, right? So we're having a bit of trouble with the business model. Our customers were meant to be places like E Trade and Ameritrade and they were losing business really rapidly.

uh and not taking on new expenses and so on. So we said, well, look, we we've all come from a tr my brother was a prop trader at Goldman and had previously launched a reasonably good, though very small, hedge fund before trade work. So we decided uh it made sense to try to monetize the technology a different way if it if we weren't gonna do it with like software online licensing services.

then maybe we would do it as a hedge fund. So we launched our first hedge fund back in February of two thousand one. And after nine eleven, really the fintech business, especially in New York, really just dried up completely. Uh deals that were all but inked didn't get inked, just n business just stopped.

And so we shut down all the tech and just be and we were just a hedge fund. And I actually left TradeWorks about a year after that and went to work for uh back on the fund of funds side, what was then TradeWorks' largest investor, which is a small fund of funds based on the coast of California. So it was great for me. I wanted to be in California. Uh you know, TradeWorks was sort of changing gears.

uh and uh they had enough quants on staff and I wouldn't have been the strongest among them anyway. Uh so I left uh and and work and I I became a client instead of TradeWorks' And I was there for a couple of years and then set up my own shop at the beginning of two thousand five.

Okay, cool. So I'd I'm really keen to dig into that. Um but just before we do, up until this point, I mean it seems as though your trading background is entirely within the professional space. Has there ever been a point where you traded on the retail side? Yeah, yeah, yeah, yeah. Um Actually, so around the time I was in Citibank, I had a bunch of friends from college who were really active punters in stocks and just like highly addicted to trading stocks and stuff. And so uh

Almost just out of being social with them. Uh I had a, I can't remember if it was a day tech account. And uh and I punted around quite a bit as well. And I I did so fucking poorly in those early days, excuse my language. but so poorly, uh, as a stock picker. And a lot of it, to be honest, was just I didn't pay that much attention uh

uh I was sort of just picking stocks based on how I felt the wind blowing and that's not a really terribly clever way to do it. And what's pretty funny is all these years later, you know, as a Uh and I should be clear, like after that I did put on some really successful and thoughtful PA trades, personal account trades. Uh I remember I went uh long Best Buy and short Sears and made money on both legs of that. I made money being long schwab and short E-trade.

Uh and so I got better at doing the personal trading. uh later, but it was always in these kind of very specific constructed trades. It was never just I know which way Sears is going and I'd make that bet. I was never any good at that, uh never showed any evidence and being good at that anyway.

Um, but all these years later, uh, you know, I now run a a firm that's got a few employees and uh, you know, as an investment, a registered investment advisor in the US, you have to have personal trading policies. And so those you can kind of have just about any valid sort of policy you want.

And we have one, you know, which which is very standard. It requires pre clearance. So someone has to ask our chief compliance officer two business days before they want to make a trade. They have to hold the trade for some number of days. I don't even know what it is, maybe thirty days or something. And then they have to request pre clearance again to get out of the trade. And

Uh, while that's our policy, the unofficial policy is don't fucking trade. And the reason is I don't want my people trading is that we're professional investors and all we do all day is make sure our Traders are really, really careful about how they invest money on our our behalf and on our clients' behalf. It doesn't really seem very consistent with that to have people spend a tiny fraction of their time punting around in stocks or anything else.

sort of like a a chef at a three star Michelin restaurant, you know, spending part of his time cooking at McDonald's or something. It's not Not okay. Uh, it seems an insult to our clients and an insult to the profession to not take it seriously when you put on a trade. But if they have time to take it seriously, it means they're not doing their job, so

I don't I don't trade anymore. Sure, sure. Okay. So let's let's talk a little bit more about the fund that you operate these days. So you founded it in uh two thousand five, I think you might have already mentioned that, and it's uh T two A M. So At this point, I was gonna ask you, um, you know, what's your trading approach? Like, how do you trade today?

But um, you know, I understand that you guys are a fund of funds. So would you mind explaining what actually is a fund of funds? Sure. So fund of funds have been around for some decades now they are effectively uh and their role has evolved over time. Uh in the early days, investors in hedge funds were primarily high net worth individuals uh who were clients of like the big Swiss private bank.

as well as some Asian institutions like Japanese insurance companies and stuff like that. Uh there wasn't that much US money in in hedge funds in the early days. Uh and so the job of a fund of funds was really to find out Like to basically broker for capacity. So they were middlemen who would find good managers.

the best of them would also conduct a lot of due diligence to see which of these managers are really good, which of them are frauds, which ones are really talented, which ones know how to manage risk. Uh and the really best ones would also try to be very thoughtful about assembling a portfolio of these things that made sense together as opposed to just picking a bunch of things that might each be good, but maybe are all very similar.

And so as institutional investors have become much more sophisticated in this space over the last twenty years. And especially over the last ten years or so. Uh it's very difficult to go out there as a fund of funds with that. generalist kind of approach. Um

And uh when I started this company I I had left a firm that specialized in investing in quant tradeworks was a quant hedge fund. I have a reasonably good grass fund. on both math and programming, even though I don't like programming. I manage teams of pr the developers and stuff. Uh and so uh, you know, we really have always been interested in having a much more narrow style and and still to this day it isn't easy for

investors to pick quants. And the reasons are several. They're tend to be pretty opaque about what they're doing. They don't like sharing a lot of information about what they do. They're notorious for it. It's actually why the space is called black box investing. Whenever you see a breathless article in the Wall Street Journal or something, it inevitably says complex formula, complex algorithms, et cetera.

Uh and so that is certainly the perception. I think some of that is true and some of it's overblown. I think the space is easier to grasp than many understand, but it's not that easy still. So I mean we set up as a a niche shop that folks uh focuses on systematic quantitative trading strategies. Uh

Mainly we're interested in shorter time shorter time shorter term holding periods. And I don't mean by that necessarily high frequency holding periods, but you know, a few days instead of some months or quarters or years. Um and there are systematic styles that go out much longer. There are systematic strategies that are truly fast. And we're sort of in between. Okay. So how do you evaluate the trading strategies and the traders that you um allocate funds to?

There's a bunch of pieces of it. The part that's, you know, kind of probably interesting for a conversation like this isn't the operational due diligence, like how do they do reconciliations the next morning and what are what are their personal trading policies and are they staying compliant. That's all there, but it's sort of legwork at the end. The part that's kind of more fun is is picking the strategies and choosing the practitioners and seeing if we

uh can pick good ones. Uh that's the really challenging part. Um effectively it comes down to Good engineers with good judgment. That's really what we're looking for. The level of impact that the discretion of a human has. on the building of a quant strategy is enormous because the computer doesn't know anything. The computer needs to be told everything. And that person who tells it everything, it's teacher, as it were.

has to be really good. You have to have good judgment about research methods, about data cleaning, about all sorts of stuff. Some of that stuff ends up being really mundane and monotonous and detail oriented. Some of it's really philosophical and uh more fun sounding, but uh a lot of it comes from experience, some of it is probably innate. Uh and it's hard to say what what brings good judgment to some people and not to others. But it it isn't that hard if you're willing to do the work to

find find out who has good judgment and who doesn't. And it's a lot of just asking tons and tons of questions and tracking the information really carefully, cross referencing earlier answers with later ones. Um asking the right questions too. So if you look at the chapter headings in the book, those are the areas that we focus on actually. We ask about, you know, their alpha models, we ask about their risk models, we ask about their transaction cost models.

We ask about their portfolio construction models, we ask about their execution algorithms. We ask about their risk their kind of risk management and monitoring. We ask about their research and development and their data. Those are really the areas. If you look at the headings of the questi sections of questions in our standard in-house due diligence questionnaire, those are the those are the editors.

Okay, sure. So so where do you find a lot of these traders? Are they working within other firms already or are they just sort of um guys who are out on their own? Like where where do you find most of the traders? Yeah, it's a it's a good question and the answer is all over the place, both geographically and in in various sort of situations. So We have uh one of our traders is literally a one-man band who works from his apartment. He does no employees.

Uh he doesn't manage very much money. I think he's got fifteen million under management, which I know in the real world is a lot of money, but in asset management land. Uh, to put it in perspective, he's too small for the regulators to want him even to register for oversight.

So it's a very he's a very small shop. On the other hand, you know, with money with a couple of guys and have had over the years money with a couple of guys whose assets under management are in the billions, they have large teams, they have their own floor in some fancy skyscraper in New York. Uh I mentioned to you I was in Australia just recently and that was to do a site visit on a couple of managers in Melbourne and Sydney.

Uh, that are more on the emerging side. So because of how long we've been doing this and our reputation in the space, most most shops that are setting up a professional kind of hedge fund systematic hedge fund wanna talk to us'cause they know that what we do is specialize in quant. Um other investors know who we are and so if we've made an investment and if we're act willing to act as a reference

that can be a handy thing for a manager, for a guy who's setting up. So we get a lot of flow just naturally, but we w have to work very, very hard. So I do Uh between me and my teammates, we do something like 30 or 40 calls per quarter with various contacts who are also looking at managers. and uh are in the flow of managers and just sharing ideas to make sure that we can cover everything as much as possible. Yeah, okay. That that's really interesting. Um

One of the um one of the things I saw on your website uh when I was looking at it before we got on this call is uh you say one of the key components of your strategy uh is described as pure alpha. What does pure alpha refer to? Yeah. So it's a also a good question. The kind of textbook general definition of alpha is sort of the returns

above the market, assuming it's positive or below the market if you have negative alpha. Um, I've never really loved that definition of alpha. It's sort of a definition by uh elimination. So it's whatever the returns that aren't from the market. are that's not a very satisfying definition for me. Um to me, it strikes me as being really intuitive and clear that alpha is actually just about timing. And it's actually a very specific kind of timing. It's uh

So I'll I'll give you my formal version of the definition of alpha and then we'll kind of talk through it. So to me the the real definition of alpha is the returns generated by skill. at timing the selection and or sizing of your holding. So if we unpack that a little bit, uh I'm interested in skill based returns, not uh luck based returns, right?

Um we try to ascertain skill by virtue of again interviewing these traders really thoroughly and trying to ascertain if they've gotten good luck good results by luck. Or if they've gotten good results by a combination of good judgment and a lot of hard work, which is what it really takes. There is luck as well, but you don't want that to be the main thing or the only thing. Um

The timing thing is is really the centerpiece though of this definition. And and I say that alpha is all about timing because let's imagine that um Uh I know some company inside out, and not because of inside information, but just because I'm really a good analyst. I know this company inside out, and I know that it's a fraud. I know that it's gonna go bankrupt. But I'm so far ahead of everybody else. And so I go short it, right? Um

Uh, I'm so far ahead of everybody else that I'm wrong for a really long time. I mean, I can go out of business before the market realizes that I was right. And so from a return standpoint, I wasn't right. I was wrong. Yeah. Uh I went out of business. I lost all my money on a trade that I was somewhere down the line correct about. So I've not shown skill at timing.

Skill at timing comes from being able to anticipate, roughly speaking, a couple moves ahead. Uh what are other investors going to demand?

Uh to illustrate this a different way, think about Warren Buffett. Warren Buffett, when he picks a stop. Uh he doesn't say, well, you know, Walmart or whatever is a great company all the time in perpetuity in perpetuity at any price.

Right. His his bet is that no, right now is a good time to buy this company. Maybe the company has the right management for now. Maybe its price has come down enough that it's a good value, maybe there's enough margin for safety and Whatever, whatever his factors are. It's a timing decision. Well, if other people come around to his way of thinking not long after he's got that way of thinking.

Uh and not long is subjective. It's up to his strategy how long that needs to be, how long he's willing to hold something and how much he's willing to see it suffer before he's ultimately proven right. Uh, then he's got alpha, right? That's and if he'd if other people end up disagreeing, if the consensus is no, this was a crappy company and it was cheap because it deserved to be even cheaper and it was on its way there then it it would not be said it it would not be said that he has alpha. Uh so

To me, alpha is just about timing. So when we say we're after pure alpha, what we mean is that we don't want any returns just from the one way bet on a market. Because a lot of that's just luck. When did you happen to get into the market? Are there any events that force you to sell at an opportune time? Uh there's just so much uh that is random about that. And what we were after was much more as we say, skill based returns. Uh we want our returns to be generated by Skill at timing.

Yeah, I really like that response, uh, Rishi, and that actually kinda reminds me of something uh Peter Brandt said when I interviewed him and he said to be right on a trade you need to be right on both direction and right on timing. So yeah, I think that ties in really well. Um Yeah, now you literally wrote the book on black boxes, so these next few questions will branch off of this, um, as well as some general questions about quantitative trading in the mix. So

First of all, the way you see it, how do you define a quant trader and what is a black box? Sure, yeah. So uh I'll kinda answer both things together and then I'll give a caveat at the end of the answer. Uh To me a a systematic trader or a quant trader or a black box is simply a systematic impl implementation of some investment strategy. So again to unpack that, let's start with the investment strategy. And and while I'm not a big fan of wars,

Uh they give us terms like strategy and tactics. And so we can kind of borrow those terms from that. that arena. So, you know, a strategy is sort of your approach, right? It's more of a philosophical guide guiding direction. Uh so in war there's things like uh You know, taking uh uh one strategy for war is is taking airplanes, uh bombing something, and then following with ground forces. Another strategy is guerrilla warfare or urban warfare and using human shields and that kind of stuff.

Uh and So a strategy in investing is doesn't have quite the same Bloody connotation, obviously, but something like value or GARP growth at a reasonable price. These are approaches to investing. And so you can take a style like value, a strategy like value, and codify it. Well what do you mean when you say a stock is cheap? First of all, what do you mean by a stock?

Is it a US company? Is it every US company? Is it just large cap ones? Is it just meet you know small cap ones? Uh are they in all sectors? You have to feed a universe into the machine. Uh and so the first thing that you're encouraged to do is think about what universe uh and then the style of well, what does cheap mean? Are we talking about some ratio of some fundamental quantity and if so which to price?

uh or some substitute for price, is it today's price, is it an average price over some period? So you by codifying it, you have to specify a lot of stuff. And that turns out to have really interesting and important consequences as you go out building a strategy. The next part of it really relates to

The caveat, right? So a systematic implementation just means that you've taken the emotion out of the day-to-day buying and selling decisions. You've done a brain dump of your idea of how to invest in the market. Into a piece of code. That code is running on a computer and coldly making buy and sell decisions on the back of that instruction. Caveat is that

This is not a binary thing, right? So there are levels of systematization. For example, would we call someone a quant who does all of what we just said But at the end, before he pushes the button or before a button gets pushed or trades are sent out to the market. He kinda goes through the trade list and says, Well, is my model missing any information about one of these names? Like let's say uh

a company's moved up a lot today. And it's moved up a lot today because Warren Buffett said he's just bought a bunch of shares of it. And so on the back of that lots of people are are buying it and it's a really highly bullish thing. Let's say that that's a real thing. Like you really expect now that company is going to go up. Well, the model doesn't know that necessarily because you didn't feed it that data about Warren Buffett's words.

Understandably. And so uh maybe it wanted to short that stock because now suddenly it looks like the price has gone up, the fundamentals haven't changed. So maybe you take out that trade from the list. So is that still a quant or not? Well I mean it's arbitrary to say. So we don't have to worry about it. What we're after is mostly generally speaking, systematization, taking the emotion out.

from day to day buy and sell decisions. Okay. And when you say a black box, does this differ in any way at all from what, you know, many would call a system? Uh well

So the connotation is different, but I don't think the term means anything different when you really boil it down. So, you know, the term black box really comes from a a sense of not knowing what's inside. It's this opaque process like a Rube Goldberg device where, you know, You've taken input in on one side and then all kinds of machinations just happen magically inside and then there's an output, which is a buyer sell decision at the end.

And that's how a lot of investors feel and how a lot of reporters and even until not long ago regulators felt about what systematic trading was. But I don't know if it's generational and people are just increasingly comfortable with programming computers to do things that we didn't think they might do before, like drive cars or what. But uh we are seeing a a less fearful approach to the space over the last couple of Well l even less than the last couple of years, quite recently

We'll see if it lasts, but it seems to be the way things are headed. So yeah, to me the black box thing is just is almost a pejorative term, saying, Well, we don't know what the hell's going on in there, it's inscrutable. So we're just gonna call it a black box. But anytime you hear that said in trading or in finance, they're really referring to just systematic or computerized or quant trading strategies. Right. Okay. Cool. Cool. So

Are you ready to get serious about trading? Then join Tasty Trade, Investopedia's best platform for options trading in 2026. Stocks, options, futures, and more. Tasty Trade has everything you trade all in one platform. Get low commissions, including zero commission on stocks. so you can keep more of what you earn. Tasty Trade is packed with advanced charting tools, backtesting, a pre-built strategy selector, risk analysis tools, and more features to help you trade smarter.

See equities and derivatives with high trading volumes, dividends, upcoming earnings reports, and more with their pre built watches. Or create a custom watch list to keep an eye on the companies and sectors that matter to you. Manage your positions with speed and precision using Active Trader Mode and one click trading and smart order tracking. Plus Tastytrad's Stellar Trade Desk team offers live support during trading hours if you need it. Visit Tastytrade.com slash chat for more info.

KC Trade Inc. is a registered broker dealer and member of Finra, NFA and SIPC. Yeah. Have you ever watched a stock explode and thought, if only I had the capital? Or sat on the sidelines because your account balance felt too small to matter. Good news. With Trade the Pool's limited risk platform, you don't need millions or even thousands to start trading the US stock market. Bypass the PDT and tap into over twelve thousand US licit equities.

From penny stocks to big caps, ETFs, even the newest IPOs, and short anything you like, with zero locate or hard to borrow fees. Start your evaluation, get funded with up to$200,000 in buying power so you can go big without risking your own savings. And now you can also have unlimited time to reach the profit target. It's a game changer. Not ready to trade yet? Trade the pool offers a free demo and educational resources.

Practice on live data, master the platform, and build confidence risk-free before you even pay a cent. Click the link in the show notes to start trading with Trade the Pools Capital.

This is something you cover at lengths in your book, is the typical structure of a black box. Um but for the purpose of this interview, would you be able to give us just an overview of the individual models that come together to form the components of a black box?

and how they interact with each other. I know this might be a little bit tricky without um, you know, visualizations, but you know, just if you can give us a a brief overview. Yeah, yeah, sure. It's actually not all that tricky. If you think about it, uh The trading system is, like any other economist, an allocator of scarce resources.

The scarce resource in this case is capital. You have only so much. And you want to earn a good return on it. Think about it the same way that the CEO of a company thinks about it. Uh if I'm if I've got a a company that does fasteners, because that's what I know, I have some business strategy for that, et cetera. Um when I'm gonna make business decisions, I need to take in various kinds of input.

And then think about how to transact, and I don't mean this in the stock buying and selling sense, but in the real world, rubber meets the road, buying equipment, hiring people, opening an office kind of sense of transaction cost. Legal documents and so on. What are the kinds of inputs that you would take in, right? You would take in what's the opportunity here, what's the upside? that we think of as coming from the alpha model. So the alpha model is the first and

in some ways most important piece because without an alpha model, without a forecast of the future, without a timing model in other words, all hope is sort of lost that you're gonna make money. So why do any of this other stuff if you don't have alpha Well, the alpha model is is forecasting the future. It's telling you here's here's how you can make money. Maybe. Here's your best shot at making it. if we just compartmentalize it, it's ignorant about risk. And so uh then you want

you know, your lawyer, your to be a risk manager, right? In a in a company, to say, Well, wait, yeah, we could try to sell this sort of product to so and so, but here are all the licenses we need and here's how much that costs and Here are all the downside risks of this. And so a risk model is really about controlling unwanted consequences as much as possible.

And then the third piece is a transaction cost model, which you can liken to the sort of accounting department who are really telling you if you're going to do this, this is how much it's going to cost. So is the upside worth it? Right, so now we've got opportunity, we've got risk, and we've got cost. And these are, roughly speaking, the three major kinds of inputs that any decision maker takes in.

when they're deciding how to allocate resources. So those kinds of inputs go into a portfolio construction model, the most typical of which is an optimizer, is not a really useful term because Depending on your approach, everyone uses an optimizer, even if they're equal weighting their positions, they're optimizing for something. But anyway, there's a specific kind of optimizer that people mean, a mean variance optimizer, the kind that Harry Markowitz sort of invented some decades ago.

And uh and this thing is trying to balance. Upside, downside, and cost, those three kinds of considerations. Upside coming from the alpha model, downside coming from the risk model, and cost coming from the cost model. So we have three models that feed into this portfolio construction thing. Uh the portfolio construction model says, right, given those pieces of information, this is the best portfolio to own going forward. That then gets compared to what you own today.

And the differences are what you have to trade. So if you don't have any positions today'cause it's your first day on the job, uh then you run your optimizer and and it takes in all these inputs and poof out comes a a list of things to buy and sell and the whole thing is is your set of trades. Um if you had a portfolio yesterday

It's pretty likely it's pretty similar to the one you're gonna have today. And so just the small deltas, the small differences are what you would then go out to market and trade. And so the last part of the quantitative sort of black box approach would be then these algorithms that actually place orders in the market. Now this is all kind of girdled in research because each of these areas

can be researched and in the best cases are researched very thoroughly. When we talked about at the very earlier part of this this topic, um how when you codify a strategy you're forced to specify a lot of stuff. that active specification again in the best cases, encourages a good practitioner to be not arbitrary and to try to do research. So

Uh what do I mean by value? Well let me conduct some research and find out what I should mean by value. I have some general ideas, maybe even some pretty specific ones, but it turns out I can test those ideas. And a well made test is, you know, not a bad way to kind of get a sense of whether this idea just sounds good and turns out not to work. Whether it sounds good and could work, but only in unrealistic implementation assumptions like costless trading or something.

So research really drives everything. You know, what risks should I control? Well, it depends on what bets I'm trying to make. And all of that kind of needs to be studied very carefully. So that's the sort of cloak over the whole thing of research or an umbrella over the whole thing of research. And then in turn, both each of these pieces uh and research itself, of course, require data. And so these are the things that are outside of the black box itself, but totally crucial to it. I mean

uh data are like the blood in a in the body. You really you can have muscles and a skeleton And organs, but without blood there nothing's gonna happen. Yeah, I think that's a brilliant answer. I really like how you broke that down. So thank you very much for, you know, going into depth and and explaining that. Um obviously you go into it at greater lengths in the book, so if anyone wants to check that out um inside the black box.

Are there any special risks inherent to quant trading, like risks that may not be a factor to other trading types? Model risk. Uh what we have then is uh a situation where we might have said, okay, uh I think that valuation is what drives future prices. Well, that's a very elementary thing. Lots of other things that aren't value might affect future prices. Growth, sentiment. uh a company's balance sheet and whether it's, you know, taking on tons of debt

Uh whether it's been downgraded by a credit rating agency. I mean all kinds of things affect the other. uh the future price of a company, not just the thing I modeled. And so that is really at its heart what model risk means. It means that I've simplified the world. And sometimes things I didn't account for will drive prices and cause me to lose money. And uh lots of times this stuff happens. You know, after nine eleven air airline companies went down and uh defense companies went up.

And if you just happened to have an accidental bet on airlines and defense companies in the wrong way, you would lose money. And by the way, if you accidentally made money, that's that's better, but it's not a lot better, right? It's still an accident. And uh and so what we what we think of as being a specific risk to quants is model risk. Now

Humans, when they have a strategy, have the same problem. And I mean I can give you an example from right now. One of our clients is a wealth management firm. That means they have a bunch of rich people as their clients. Uh they have other sorts of managers and other sorts of spaces. So they have some like long only stock pickers. Uh and one of these long only stock pickers was Long Casinos in Macau.

Okay. And uh seemed like a great bet, fundamentals all lined up. They didn't account for the fact that China was gonna make it really hard to kind of get money out of China by sending it through casinos in Macau. And when they did impose serious restrictions on doing that, and I don't know all the mechanics here, but when China did make some change to its capital flight rules, uh the casinos in Macau started losing business really quickly.

And their stock prices suffered as a result. Well, that's model risk too. Uh it's just we don't think of it that way. Uh In the case of a human, the human can identify that much more quickly and often prospectively. and avoid it. In the case of a a quant strategy, it's very hard to do that because what data could you feed into your machine about China's arbitrary policy?

It's very hard to handle that kind of stuff. And so And you can't really conduct any research about it'cause there's no historical precedent to to look at, you know, in even if you can find analogies. or analogs for this thing. It's not it's not easy, right? So model risk is perhaps a risk to anyone who's got a disciplined investment strategy, but more of a risk for quants than for even a disciplined human.

Because quant what what machines are good at is mass mass production of a concept. What they're not good at is nuance. And that's where the human mind is. is a better tool. Yeah, that's a that's a really good point. And I think this question may tie into that somewhat. Um, this is another topic you write about um in your book, and that's uh common criticisms of quantitative trading, one of which is those who say trading is an art, not a science. How do you respond to this point?

Um I'd say that it's both, and that's the response. And by the way, lots of things which people thought were an art, not a science. including certain actual arts like writing, uh, creative writing. There now are interesting pieces of software run by some of the R and D houses like IBM research and so on, where they're writing poetry

uh computers are writing poetry in an automatic way and uh uh doing like tests with English professors to see if they can really tell what was written by humans or what anyway, leaving that Turing test aside. Um you know, medical records are being read by or diagnosed, sorry, yeah. images, right, like x-rays and CAT scans and that stuff are being read by computers now more than doctors.

Uh, and no one would have thought like if you asked a doctor if that sounded like a good idea five years ago, they would have said, Hell no, that's terrible. Will never work. Uh, driverless cars, et cetera. I mean, so I guess the reality is that um Uh there's room for both approaches. Uh it's fair to say that computers can't understand nuance, but it's also fair to say that humans are really unreliable.

practitioners. Uh human pilots when they when they're in charge of airplanes sometimes take the plane to the wrong airport because they were busy, distracted, playing a Sudoku puzzle. This is a true story. Uh pilot error is a big thing, right? In causing plane crashes.

And weaving aside pilot error and not to trivialize important examples of life and death, but you know, there was that flight in Europe where where the a depressed pilot crashes the plane full of passengers into a mountain. I guess that's the the findings uh that they had. I mean, it's it's really hard to say that humans are so great. We're good at coming up with ideas. We're not so good at implementing. If you think about it in sports, and I don't know if it's the same in

in uh in some of the sports that are more popular in Australia, but in the US, for example, there's baseball. And when a pitcher's having problems or in American football, when a quarterback is having problems, a lot of the time you'll hear

the coaches and the commentators pointing to his mechanics. What does that mean? It means he's not doing the exact same thing over and over. He's changed his arm angle slightly and it's screwing something up. Or he's changed his release point slightly and it's screwing something up. Uh this is not something that computers generally have problems with. They have the same release point pretty much every time within mechanical tolerance.

Alright, well let's now narrow in on the subject of HFT. As someone who is a great advocate for HFT and you believe there is a lot of misinformation flying around, I'd love to get your take on this. Probably the first thing I should ask is what is high frequency trading and how do these firms make money? So uh there isn't there really isn't a uh uh canonical definition of HFT, the

The definition I use is uh really a strategy that has three defining characteristics, or three characteristics I should say. One of them is that it tends not to forecast out further than a day. In other words, it's trying not it's not trying to time things further than the end of today. Secondly, it tries generally not to take home positions overnight. So it's mainly active intraday and then goes home what we call flat, which is to say largely just in cash.

Um both of those characteristics are more optional. The third part of the definition is really not, and I think is the most universally accepted part of it anyway, which is that the strategy is sensitive. to the amount of latency the uh between um And the latency really is referring to how much time it is between an event happening in the market and your systems being able to process that.

And then in turn, how quickly that decision that you've made on the back of uh the you know, in response to the market is going back out to the market. So the kind of round turn of information into you and then back to the market. is the sort of latency that people care about. There's a bunch of subcomponents there that we can talk about, but that's that's the short of it. So if a strategy really needs

To be implemented on a very fast infrastructure in order to make money, then it's a high frequency strategy. So beyond that, It's a a very poorly defined term because a huge number of things are conflated into this term that that aren't any more similar than car is to driver.

Yeah, okay, that's really interesting. So would you mind expanding on some of those subcomponents that you mentioned there? Um of latency? Uh sure, yeah. Okay. So The first thing that I uh there there's a bunch of them, right? So one of them is that uh an event happens in the market and that piece of information has to reach your server. Okay.

Uh so that's the first node and connection of between those those two nodes, the connection between the event happening and the first node, which is your machine finding, that's the first. uh amount of time you want to minimize. Then you have to process that event, meaning your computer has to make sense of was this how does this message fit in with all the other messages and what type of message was it? and make that information available to your internal decision-making algorithms, right?

Uh that's the second node. And then your decision-making algorithm has to process. So the start and end you know time stamps of that, that's the third segment of time you want to minimize. And then your order has to go back out to the market and be in you know, in the market again. And so that's another so there's a few of the major Uh those are the major uh sources of potential latency, which as a high frequency trader you'd want to minimize.

Okay, so what you're saying, um just going back to your first response there, was that there are various types of HFT. You can't really put every firm who trades within, you know, fractions of a second and who relies on latency um to be minimised as much as possible. You can't put everyone in the exact same category.

I mean you could, um, but it's just ill defined. Um and so it it'd be like saying vehicles as a category. I mean there are many types of vehicles and different types do different things. So uh uh you know Uh not to be too coy about it, let's just dive right in on what types of things are. We can do some subcategories of types of HFTs that are sensible. Uh so one type of HFT And one that I think is the most familiar to people is what we call like a regulated market maker.

uh or contractual market maker. And so these are folks who uh for example, if you have an Ameritrade account, this is one of the big online brokerages here, uh, and you place a trade uh the odds are quite good that that trade is going to uh a big market making shop. part of whose existence is to take the other side of your trade. And to do it quickly and at a at a very fair price. Um they're required to take the other side of small trade.

Uh, they are required to fill as best they can uh larger trades. There are some shenanigans in that area that we can get into, but the um

one of the practices that I think is at at least fair to say questionable is this practice of payment for order flow. So uh I'm I'm I'm some broker. That broker has a bunch of clients and those clients have a bunch of orders. And I, as the broker, rather than connecting those orders directly to the marketplace, I'm routing them to a market maker. Well, why am I doing that? Partially because the market maker is going to pay me.

And so one of the line items of revenue on like Ameritrades and ETrades, most other brokers uh you know statements. Ironically Schwab also. If you look at Schwab who were very outspoken outspoken opponents of HFT after the Michael Lewis book came out. Ironically, they engage in payment for order flow, which is to say they receive payments from HFTs for routing those trades out to HFTs. Uh this is a a practice that I think deserves some scrutiny.

It's a very old world practice. It's an old school practice. It predates H F T as we know it. HFT as we know it, especially in in the US equity market really didn't exist and it's in in anything like the form it does now. Uh until about 2007. But this payment for order flow stuff goes back way further than. Uh kind of wholesale market making operations go back decades. Um the next kind is a non-contractual market maker. So there are strategies which

sort of loosely provide liquidity to the marketplace. They're very skittish though because it's a If you're not very careful, it's a highly unprofitable thing to do. It is a great way to lose money. Uh and so you want to be very careful if you're gonna provide liquidity. Um to the market, uh that way. A third kind are what we call uh just fast alpha traders. And so these are taking the same kinds of strategies and concepts that

uh slower trader, someone who holds positions for days or weeks might use, for example, mean reversion. It's a bet that similar things will have their price converge. So if you have, you know, two airline stocks that are both large cap in the same country, like American and United, uh all else equal, especially on a short timescale. any divergence in their price is likely to be some short term liquidity driven thing. And so if you fade that move, if you

buy the underperformer and sell the outperformer, you have a fairly low risk trade and maybe you'll make money on it. Um and so there's some of this kind of stuff that goes on in the sort of fast alpha category and and lots of other little games and trade. um, ideas for alpha. They just have a short horizon.

And the last are the most latency sensitive, which are the true arbitrages. So things like index arbitrage. Funny enough, they are the most latency sensitive, but also the ones which violate the first two components of the definition of HFT I gave earlier, especially the one about Particularly the one about not taking home positions overnight. So in an index ARB trade, let's say you have an index of fifty stops. Uh I think the Euro stocks has fifty stocks in it. Um

So there's an index that trades on a futures market. There are ETFs that track this. uh uh this index as well. And then it has the fifty underlying components. Well if you can see in real time that the index isn't tracking its components properly. Then you can buy the underperformer and sell the outperformer. And you've effectively locked in a riskless profit. The problem is you may never be able to exit that trade.

Because when you go exit, you're gonna have to pay transaction costs to get out, and that will wipe out the profit. And so what you see often with index ARB is it builds up really huge positions, but they're riskless because your long A and short the components of A, which is the same thing, the sum of them is the same thing as being long A. So your long A and short A. and riskless, but you've locked in this little profit by being very, very, very

Um so those are the most speed sensitive traders. The second most speed sensitive traders are the voluntary market makers, the sort of non contractual market makers. Uh they are they have to be fast because one of the ways that you have to be very careful is leaving stale orders out. You get what's called picked off, which is to say you've put up a bid

to buy at$100. And if you're not really careful about getting rid of that bid as soon as it's time to get rid of it, then someone will take very serious advantage of you and and you'll be uh buying at 100 when you should have been buying at 99. Or ninety eight or whatever. Okay. So just to help us better understand those last two types of HFT, the ones that require reduced latency.

What type of volume are these firms transacting on sort of a daily basis and how much money are they really making? I know it's a very broad question, but just to give us some sort of idea.

Well, the I dun I've never seen a breakdown of any of these subcategories because they're categories I sort of came up with in thinking about the space and trying to decompose it in a more helpful way than just some monolithic title of HFT, which doesn't even really make much sense given the heterogeneity inside. That label.

Uh that being said, there's a group called the Tab Group. I think they're based here in the States. Uh Larry Tab is the guy at the head of it. And they've done as good a job as anyone, I think, uh trying to tabulate results. of uh on on both questions. And so, you know, the kind of expectation is somewhere between half and sixty percent of volumes in US equities are done by HFTs of various types.

And uh in terms of profitability, that number has come down uh, you know, pretty dramatically over the years. So uh according to the tab guys, and this is from two thousand fourteen, in two thousand nine, the aggregate profitability and this is in US equities only, but that's a pretty big market obviously. Uh was seven point two billion dollars. And H F T. Uh it has declined every year subsequent.

And the most recent figures they had in twenty thirteen were one point one billion dollars for the exact same thing. So it's, you know, uh one seventh the size it was, something like an eighty percent or so reduction, eighty five percent reduction in profitability in the industry overall. Uh and to put it in per share perspective, we're talking about zero point zero zero one dollars per share, so a tenth of a penny per share.

uh is the aggregate profitability of HFTs relative to the volume that they do.

Okay. Interesting. So sorry, just one thing you said there. Did you say seven billion dollars in two thousand and fourteen and then No no two thousand nine. Oh two thousand nine, okay. Yeah, down to one billion dollars in twenty thirteen. The study came out in two thousand fourteen. Right. So so why do you believe that numbers on the decline?

Well, uh a few reasons. One, volumes on the decline. Uh and especially through that time point. I mean, I haven't tracked exactly how volumes have evolved this year, but uh secondly competition. Um there's so much more HFT going on than there was that uh so many more players competing to provide liquidity, as it were.

that the price of liquidity has come down. If you think about it in basic econ one supply and demand stuff, in that framework, liquidity is a thing for which there should be some equilibrium price.

And as there's more supply of liquidity, that price goes down. As there's less supply of liquidity, that price goes up. And uh sort of a secular trend over the last several years has been uh an increase in the supply of liquidity relative to an actual decrease in the demand, as you can see from volumes having gone down. So that's a bad combination for the price of liquidity, right? Which is the price that HFTs kind of get paid. Now, episodically, when you have a panic in the market,

the price of liquidity can temporarily shoot up. So this kind of accounts for that on average, right? So for example, the flash crash, which happened in two thousand ten, Uh that was still a less profitable year for HFDs than two thousand nine. And O nine was less profitable than O eight and so on. Um so it's been a pretty much one-way street. And and to be clear, there are some other there cer are some interesting public pieces of information out there.

Uh if you look at the financials that get go, which was one of the biggest couple of hedge uh uh of HFT uh in the world. Uh they had to make their financials public when they acquired Night Group, which was another of the biggest HFTs. Both of these are large market making houses.

And so when night sort of blew up uh in the last couple of years and then Get Go came in and bought them to form what's now KCG, uh They made their financials public and I don't remember the figures offhand now, but there was something like a ninety percent decline in their profitability. Over the course of a few years. Right. That's that's pretty huge. Yeah, and so I mean it makes uh one of the things that was particularly annoying

Uh, and I mean that in the sense of like a fly is annoying, uh, about the Michael Lewis book was its timing. It was sort of like I mean, f four years ago this book might have been interesting. But today you're looking at this hyper efficient market where you basically don't get paid to provide liquidity anymore, or you do, but it's it's very hard to do it anymore.

And there's been a lot of consolidation of businesses, a lot of guys going out of business, et cetera. No one's crying over that. I'm just saying that the timing was a bit funny.

Yeah, and that's one of the things that, you know, you've spoken about a little bit in the past was uh Michael Lewis's book, Flash Boys. Um you disagreed with many of the claims that were made inside that book. Um could you share an example of what you thought lacked evidence? Almost everything. I mean the idea that HFTs are front running as an example is uh on the whole completely idiotic.

Um and even the people he cites as having the expertise making that claim, IEX, don't make that claim, have severely backpedaled from that claim in the year or so since. uh the book came out and have uh actually got HFTs on their Exchange slash uh dark pool. It's uh it's a completely ludicrous claim. Uh it's a bit technical to get into exactly how, but not that technical. Um if you think about

why he says it. He says it because uh an HFT can pay the exchanges an extremely expensive cost on the order of a couple hundred thousand dollars per year. To get access to direct feeds that show how those order books on the different US equity exchanges are evolving in real time. uh someone who wants to not pay that couple hundred thousand dollars a year can for something like uh

A thousand or two thousand dollars a year get what are called real-time feeds. And this is really the national best fit offer. So it's a consolidated. Market across all of the books, uh, all of the order books at all of the different exchanges, and it just gives you the best fit and offer in the last trade. the relatively poor architecture of that national best bit offer feed, that real-time feed that costs only a couple of grand per year for a professional investor.

And to be clear, many retail investors, like if you go on Yahoo Finance or Google Finance, you can get the real-time price for free. So a professional investor pays two grand, a retail investor gets this for free. That thing is delayed by about one millisecond versus what you could do if you went direct to all the exchanges, processed it as fast as possible.

You'd be one millisecond faster than the national best bid offer. Now, to Michael Lewis, this means that you're front-running because when an order happens on exchange A. The$2,000 feed doesn't see that for one millisecond. The person who takes the feed at exchange A sees it right away. Okay?

The problem with this being called front running, besides it just being stupid, is that It's like saying that I read the newspaper before you this morning and therefore I knew what happened I I know what's going to happen today.

the the the newspaper tells me what already happened. So the fact that I'm making a real point about finding out what already happened faster is like totally different from front And by the way, the reason that I have to do that as a market maker is what I referred to earlier about the perniciousness of leaving orders, leaving stale orders out there in the bid offer book.

And so uh a stale limit order is just an invitation to get fucked. Uh there we we at TradeWorks did a pretty interesting study on this, and there's a pretty huge uh cost. to being slow. It's something like a penny per share. over something like a two minute period. Actually I have it in front of me now. It's 1.7 cents per share difference between fur between being first in the order book and last in the order book.

And bear in mind that's more than ten times, like seventeen times, the expected profitability of a good HFT. And that that's not a good that's that's not a good way to make a living, right? Uh so it makes sense that you want to be fast because if you find out you've placed your order and you're not near the top of the queue. For at that price. And then you get hit, you're almost certainly going to lose money on that share, uh on that trade.

And if you do this over and over, you'll be out of business in no time, and you'll have paid hundreds of thousands into the millions of dollars in infrastructure for that. So this is why these guys have to be fast. It's nothing to do with front running.

The irony is the only kind of HFT that could front run is the contractual market maker. And to my knowledge, not one contractual market maker has ever been sued. Not even in all of the noise after the Flash Boys book.

for misusing their clients' information to front run them. Okay. So I mean Hain Bodek was on uh the other week for uh episode forty-nine and He from what I understood about his story was that Uh the some a certain certain HFT firms were using special order types that did allow them to jump cues. Um, I mean I understand those order types might have been called hide not slide orders, which have now been um taken away.

Was that the case, you know, like in in your opinion, was it the case that when HFTs were using those autotypes they were able to uh jump cues. And they're now not able to do that. What's your take on that? It's not exactly the case that it was cute jumping. Uh and it gets really technical. Like a discussion between me and Haym on this would go over just about everybody's head. It wasn't

Like a hardcore practitioner. It's and by the way, it would be a discussion, not not a fight. Um we agree entirely, by the way. uh he and I that hide not slide orders and all these there are many exotic order types. are pretty much a direct result.

of certain inefficient aspects of market structure, like a slow national best bid offer, like a totally arbitrary ban on locked markets, which is saying that you can't have the best bid and best offer both exist at the same price without them interacting.

So if the best bid is a hundred dollars point one hundred point zero zero, the best offer can also be one hundred point zero zero. Those two orders would have to interact and both players would have to trade. So the minimum bid offer spread is a penny.

Well, it's pretty arbitrary because if we like a penny spread, wouldn't we like a zero cent spread even better? Uh and Because of the way that the NBBO is the national best bit offer is created using this outdated piece of technology called the SIP, which just stands for Securities Information Processor. Um You have to find workarounds for the situation where you haven't locked a market, but the SIP thinks you've locked a market.

And so most of the funky order types are a direct result of working around this flaw in the structure. That being said, The term cue jumping is a little loaded. Uh what it means to jump a queue is there was a line there. I show up later and I hop to the front of that line. That isn't actually what happens. What happens is my order is there, it's hidden, and then when it gets to a certain price.

it lights up and is first in line. But the order was there, right? So it's not the case that someone got to put the order in later. There is a timestamp and time priority is given. It gets into some technicalities. I understand why Hame doesn't like them. I don't like them either. Um but I I don't view this as being uh

uh some big problem. Right. Okay. Yeah. And I mean this is sort of starting to get over my head a little bit, but no, I really appreciate you trying to explain that. And I think um you've done a really good job of that. Just to summarize for us

Is there anything about high frequency trading that you believe, in your opinion, is unethical, or are you totally for it? Well, I I don't think it's I don't mean to be a pain. I just don't think it's a well defined question and I don't and so what I mean is that Um there are aspects of market structure that I think need improvement.

By improving those, a lot of the bad behavior that exists would go away. There are bad players, right? There are people uh and shops who uh contract with data vendors to lock out other players from getting data faster so that they can uh have first mover advantage on certain data types.

There are players who uh make orders without any intent of getting those orders filled. It's called spoofing. And they're just placing these orders rapidly and canceling them to try to make other people make mistakes. Um this stuff does happen. It's just the vast minority of the time that's what's going on. Even in specific instances where the regulators have and they now have the technology to do this, have looked at instances that look like

bad behavior. They go in and find out there is no bad behavior. Uh there are just some inefficiencies in the system. So I mean I can give you a few sort of things that I've talked about as being areas for improvement. One is getting rid of the ban on locked markets.

uh it's something I just referred to. The second is make the SIP as fast as getting direct feeds. And by the way, that's entirely possible. You would just have to do what HFTs do. When they do this, they get all the direct feeds. They process them well. If we had a central uh order book that did that, we wouldn't need to do that on the HFT side. Uh and then there wouldn't be any advantage to having the direct feeds. Get rid of payment for order flow.

Get rid of off-exchange transactions for exchange trade instruments. This is something that's again a little technical, but there are things like wholesalers, these contractual market makers. And Dark Pools, which got lots of press. One of the really funniest ironies about the Uh, Michael Lewis book is he touts how two-tiered and and uh clubby the markets are because HFTs have this great advantage. And

uh poor institutional investors with their billions and billions of dollars and hundreds and hundreds of employees are like the second tier. The irony is that institutions interact largely on dark pools, like the one run by IEX, who are the heroes of the Lewis book. IEX is a dark pool, which means it's an invitation only club effectively for doing transactions and exchange traded securities that will never hit the tape on the exchange. So they do the public no good.

They give off no information. They can have rules that are not actually that fair. And so on. Uh uh to me, if you have an exchange, the purpose of the exchange is that's where uh transactions are done for those kinds of instruments. We don't have dark pools in futures, we don't have dark pools and lots of other kinds of exchange traded instruments, we just have them for U.S. equity.

And this has been something which is really a response to special interests, large institutional investors in this case being a special interest, who've clamored for this kind of market structure. It's been this kind of two tiered structure for decades. Uh used to be called the upstairs market. Now it's Dark Pools, same thing. It's a computerized version of the upstairs market. Uh so I get rid of all that stuff.

At least I'd seriously take a look at it. Uh flattened rebates for liquidity provision are currently tiered. Uh and finally vigorously prosecute insider deals where a data vendor is giving special access or an exchange is giving special access to data or technology to one kind of player than another. On the whole, though, Aaron, I would say that it is Absolutely to fly in the face of all of the evidence.

to say anything other than the following, at least about the major US markets and the major European markets and other modern markets out there. We have easily the most egalitarian fair markets we've ever had in our history. So to bring it all the way back to the beginning, you asked me how I got involved in the space and I pointed out that I was picking stocks by reading the newspaper the next day.

The guys who were trading on Wall Street had quote machines in front of them that told them prices in in near real time. That was a big advantage over my dad me and and me making stock there. The advantage that the most fancily outfitted HFT has in speed, just in plain speed, over the average mom and pop investor today is measured in milliseconds. Liquidity is cheaper than it's ever been. It's abundant.

Uh, the kind of bad practices of manipulation and so on still go on, but they're less and less every year. Uh we have like much bigger problems as a society than HFT uh and fixing market structure. Like I'm all for rounding out the the corners, the rough edges on the inefficiencies, but Uh to be clear, all of this, like you know, if you want to talk stock market problems, I think short-term incentivization of of uh

of CEOs to manage earnings in the near term without really worrying about the long term consequences on their shareholders on the planet, et cetera, of their actions, those are these are really big issues in capital markets regulation, not sort of Some tiki tack. nonsense, and it is mostly nonsense, about how HFTs aren't fair'cause they have to invest millions of dollars a year in infrastructure for the privilege of maybe, if they're good at it, making a tenth of a penny per share.

Yeah. All right, Reishi, well, this has been a a fascinating conversation. Uh thank you very much for coming on. Um let's let's leave it at this. I mean I mean, I know we've gone a little bit over time here. Um, but I really appreciate you um, you know, making the time to do this. It's been a a lot of fun speaking with you. Where can listeners go to find out more about you? Well, as hedge funds I'm not really allowed to advertise anywhere. I do have a a Twitter, but I don't tweet very much. Um

And uh I don't really write anything that's allowed to be consumed by the public. So sadly it's not there's not well, maybe it's not sad. There's not that much fun to learn more about me anyway. Well, I'd definitely encourage listeners to check out your book inside the black box if they'd like to find out more about, you know, quantitative trading and um, you know, algorithmic trading, that type of thing. Um, awesome book. I'm halfway through it, like I mentioned at the beginning.

Um and also inside the black box the documentary by VPro is also a really good one to check out. Um that was that was a really well put together documentary. So I'll link to both of those in the show notes at chatwithraders.com. And um, you know, if listeners wanna find out more, uh all the links will be there all in one place. All right, Rishi, will you enjoy your evening and uh let's speak again soon. Thanks, Aaron. Hope so.

You've come to the end of this episode of Chat with Traders, but don't worry, more great episodes are on the way. Updated with each great new episode. On Shat with Traders.