Quants worth following: Vivek Viswanathan

"Quants worth following" is our latest interview series highlighting thought leaders in the quantitative trading industry actively sharing their knowledge and resources with the community.

We recently spoke with Vivek Viswanathan, Quantitative Portfolio Manager at BTG Pactual. Prior to BTG Pactual, Vivek spent nine years at Research Affiliates as a quant researcher, seven years at Rayliant Global Advisors as the Head of Research and Portfolio Management, and one year at Delphia as Head of Quantitative Research.

In our conversation, Vivek shares details about his untraditional quant background, daily job responsibilities, industry trend predictions, career advice, and resources he recommends to stay up-to-date in the industry.

Vivek discusses how his economics and finance background has been advantageous in his career working in mid and low-frequency quant equities. Although most quants come from a mathematics or computer science background, Vivek mentions that his background has given him a better understanding of differentiating between effective and ineffective ideas.

"One somewhat unusual part of my background is that I came from an economics and finance background instead of a more traditional quant background that might be more focused on math, physics, or computer science. That was originally a hindrance, I think, because practically the most important thing you need to know is how to code. However, I think in mid and low-frequency quant equities, which is primarily what I'm specializing in, having a finance background is actually extremely helpful for one big reason. You have a much better sense of which proposed ideas are most likely to work.

So, for example, you have a cross-sectional equity strategy, and you're trying to implement some sort of forward-looking ball control. Your first thought is looking at aggregate implied vault since that will be sufficiently, highly correlated with individual equity volatility that it can serve as a good first pass. If you're looking to predict underlying financial variables, you'll know that the market responds more to sales and earnings than to operating cash flow. And that is a result of financial accounting making earnings as relevant a variable as possible.

That knowledge is actually useful and I think it's harder to pick up on your own than coding and machine learning which there are currently a ton of resources for online. So I did have to pick up a lot on my own, but those things were a lot easier to pick up in my mind than all the sort of financial theory, which generally you have to pick up from books."

Vivek shares the evolution of his workflow from primarily writing papers and performing exploratory research to solely focusing on risk-adjusted returns.

"Throughout my career, my day-to-day has changed considerably. There was a period early in my career where most of my time was spent writing papers for practitioner journal publications, doing exploratory research that was many steps removed from trade implementation, and doing creative custom analytics.

Now, everything I do is focused on this simple idea of what is the next thing that I can do that will add the most risk-adjusted return. I'm setting up a new strategy currently that involves an outsized amount of data and infrastructure. But in normal times, it's something like 5% infrastructure, 25% data, 30% feature engineering, maybe 20% machine learning and optimization, and 20% trade implementation. Obviously, that's an approximation, but that's how it roughly splits out."

Breaking into the industry has changed over the years, with the barrier to entry being higher than before. Vivek speculates that it's due to coding efficiency being much higher due to off-the-shelf solutions and coding assistant tools. He provides an example:

"I began trading a strategy built from scratch in 2.5 months, with just one other colleague who joined six weeks into the process. That would not have been possible five years ago and definitely wouldn't have been possible when I started 17 years ago."

Vivek has a few suggestions for early career applicants: expand your pool of possibilities, work in a place with a PnL cut, and it's easier to work at a firm trading its own capital.

"Not every quant job is at Jane Street, RenTech, and Hudson River Trading. There are quant jobs at large firms that don't primarily do quant. There are small quant firms. There are quant jobs that predict over a longer horizon. There are quant jobs that focus on every asset class you can think of. There are buy-side quant jobs and sell-side quant jobs. And yes, you would have gotten paid more working at Jane Street or RenTech, but you could still make good money working somewhere else doing interesting work. So, one recommendation is just to expand the pool of possibilities. You don't need to get in at a tier one firm."

"If you believe you're skilled, it's better to work in a place with a PnL cut, a place that's either trading with their own capital or at least the job involves very little marketing or client interaction. Work in a multi-pod shop where you're insulated from any of that work. You might only achieve this later in your career, like I did, and that's fine. You don't need to have everything right out of the gate."

Vivek explains why a PnL cut is important, going beyond the main reason people might think it is which is to make a lot of money.

"Having a PnL cut keeps you focused on the right thing. I've worked in places where comp was completely discretionary and it's incredible what people will focus on. So a research idea might just catch someone's fancy. It might just seem particularly interesting or compelling. It might be a feature with a fascinating story or a new technique that seems fancy or poor, an interesting dataset, or whatever it might be.

If you have a PnL cut, you will be entirely focused on how to increase profit and mitigate the risk of losses. And if I were to come up with an idea that my colleague or employee did not think was the profit maximizing thing, they will tell me, right? They will say, is that the thing that we want to focus on here? That just makes things very clarifying."

"If you can, it's easier to work at a firm trading their own capital than one trading client cap. The reason is you spend a good amount of time on analytics and communication instead of improving the strategy. It's actually very hard to get away from this because even if you are just trying to focus on improving the strategy, someone's gonna ask you, well, how does this new technique work? And can you describe it in a way that clients can understand? Can you produce some analytics related to it and so on and so forth. So that ends up becoming a massive time suck. And it takes you away from the value adding activity of just generating alpha."

Vivek recommends his top three resources that he uses to stay up-to-date in the industry:

"As far as quant podcasts, Corey Hoffstein's Flirting with Models is excellent. So depending on the particular guest, you can actually get useful, practical implementable knowledge there. On a non quant-related note, _Goldman Sachs Exchanges _ is good for general financial news. For newsletters, Justina Lee's quant newsletters is good because it gives a list of recently published or uploaded quant papers. What you can do there is you can just like quickly scan through and figure out if there's anything that's relevant and read up on it or not as the case may be. Those are the three things that I generally use."

Before diving into his trend predictions, Vivek shares his thoughts on pre-existing trends:

"I don't think I'll be saying anything that anyone doesn't already know, but I expect machine learning to become increasingly important, even more so than it already is. The ability to encode text as features will become increasingly important. Trading and information corporation will become faster as it has over the past couple of decades. The alphas of today will get arbitraged away requiring increasing sophistication."

Providing insights on strategies he currently considers effective, Vivek predicts that these will continue to become more valuable in the future.

"I think multi-strategy approaches will become more popular. They seem to be the most consistent performers, this is largely due to diversification. Maybe slightly more controversially, I expect single pod multi-strategy approaches to become more popular. So, this is, you know, shared PnL across the group running many different strategies across many different asset classes. I think there are two reasons why this is, there's a lot of value and shared information between groups. So that's reasonable. So, fully isolated pods can create suboptimal models.

The second thing is you have this issue of PnL cuts in multi-pod where let's say one pod's up 20 million, the other pod's down at 20 million. You still need to give a PnL cut to the plus 20 million pod. And so you end up being net negative. With the single pod multi-strategy that sort of solves that problem. So that's a bit of a guess, right? My job isn't predicting where the quant industry is going. So I don't actually know if that's going to be accurate, but I think that's something that could be a strong value add going forward."

Watch the full interview on our YouTube here.