After stints as a trader and portfolio manager on the buy- and sell-sides, Mahmood Noorani opened up macro research firm Quant Insight, and is now considering making a path to China. He talks to MarketBrains about their tech and China strategy.
MarketBrains: What led you to start Quant Insight?
Mahmood Noorani: I spent most of the 90s in banks, Morgan Stanley and UBS, as a trader and in interest rate derivatives. Then I moved to being a proprietary trader for several years at Credit Suisse, and then a portfolio manager at BlueCrest Capital in London for five years.
Around 2010, I became convinced that the fixed income active discretionary investment processes weren’t really working the way they had for the previous 20 years or so.
I put that down to the world becoming much more complex with regional interconnectedness. If you want to understand one region now you have to look at them all, whether it’s credit markets, FX, rates, commodities, or the (emerging markets) complex.
Now, there are more factors to worry about such as QE, sovereign stress in Southern Europe and I found myself needing to really get my head around 20 to 30 macro variables instead of maybe three to six, which was the early days of the 90s.
I also found myself struggling with the vast amount of available data and sat back and thought: well there’s all this complexity, there’s all this data, and processing power has also become incredibly cheap, so surely there’s an opportunity here.
We started working with professor Michael Hobson in Cambridge, he is an astrophysicist who has done a huge amount of work in Bayesian inferences, machine learning, deep learning, neural nets, and so on.
The result was, we launched QI as a company in 2015, had our first client in 2016, and in the second half of 2016, we are up to 20 subscribers.
We started to get proper traction around this year, we’re around 85 subscribers, we have clients across the equity long/short hedge fund space, macro hedge funds, real money asset managers, family offices, we have insurance companies, pension funds.
“…where we are embarking is step one of a discovery process about China…”
MB: And your next push is to improve your tech and make inroads to Asia?
MN: Part of the next push is, number one, getting somebody like Ryan Prescott Adams onboard to help us further refine our methodology.
We spend a lot of time trying to deal with issues like overfitting, and we’re pretty religious about that. It’s very clear to me that if you take enough data you can explain anything historically, but it’s going to be useless going forward.
One of the big ways we deal with that is we use the investment experience we have to narrow down the set of plausible explanatory variables; if we’re looking at the USD/CHF exchange rate we will not take pork belly futures prices, even if it gives us better explanation power.
So, we’re very much hypothesis-based, but we don’t want to just data mine.
To help us with the overfitting, we’ve done a lot of things under the bonnet with PCA (principal component analysis). Financial markets data is all in different units, and you have some challenges in normalizing that data: like making sure you are getting enough information from your variables, but at the same time not wanting to overfit.
As far as Asia is concerned, on a long term basis, we see Asia as a major growth area and investment activity is growing.
“…we want to start looking at China corporate credit spreads, banking sector stress measures, all sorts of other data”
MB: Can you define Asia? You’ve announced that Duncan Clark, chairman of BDA China, will be on your advisory board. Is China the entry point?
MN: Duncan is an option for the future. Our primary focus is to build out our core institutional user base which is still the US and to a certain degree Europe.
But when we did research, we found that in the US, there are about 50 million online broker accounts. If you look at the eurozone, there are around 25 to 30 million. In China, there are something like 200+ million.
So, when we look at where we’re going to be in a couple of years’ time, it seems to us that the sheer potential size of the online trading world in China is so big that we can’t ignore it. And we’re also very aware that in AI and machine learning, China is very advanced.
“…big Western companies have really struggled to get a foothold in China”
MB: What are some of the challenges you are expecting?
MN: The impression we have about China is that it is a bit of a minefield if you are coming from outside.
It’s pretty well known that big Western companies have really struggled to get a foothold in China, and so in order to help us navigate this minefield it seemed to us that local knowledge was essential. I will do my first trip in Q1 (2018), meeting with financial institutions to test the waters.
We are also getting the sense that Chinese institutions are trading US and European assets more actively and that they will be receptive to this kind of non-opinion based quantitative analysis of global liquidity across all asset classes.
“One of the things I am trying to uncover in China is: if I speak to institutions and they think our methodology and framework is interesting, maybe they have some interesting data.”
MB: When you say financial institutions, you mean the state-backed pension and insurance funds, for example?
MN: Yes, they are state-backed, and there are presumably some private wealth managers.
Duncan really takes us from a zero-knowledge base to 10 meetings in week one. We can get familiarized with whether there is an opportunity there or not, what the time scale is, what the need is, what securities they look at.
Also, we track big China indices, the Hang Seng, and other Asian indices, the Kospi, so it’s also to help us build out our set of securities that we run analysis for. And the other thing that is very important is: how do we get hold of data on China?
Right now, we take data from a company that has a China daily real GDP estimate, so we get a bit of real time-ish growth data for China, but we want to start looking at China corporate credit spreads, banking sector stress measures, all sorts of other data.
What we tend to look for is proxy variables that help us get a sense for what’s going on indirectly, and we try and get daily data. For example, one of the variables we look at to assess China financial stress is the Hibor-Shibor spread: it blows out when there’s capital outflows from China.
And the other aspect for us starting from a relatively low knowledge base is to actually go to China and start to get a better idea.
One of the things I am trying to uncover in China is: if I speak to institutions and they think our methodology and framework is interesting, maybe they have some interesting data.
To be totally honest, where we are embarking is step one of a discovery process about China and that’s really the motivation behind getting Duncan onboard.