Report: Quants restrategise after markets plummet

Quant funds have had a hectic month. Many notable quants – Millennium, DE Shaw, Two Sigma – suffered losses at the end of March, leading some commentators to question if another “quant quake” like that seen in 2007 could be on the horizon. Things have been looking slightly brighter for quants over the past two weeks, …

by | April 29, 2020 | bobsguide

Quant funds have had a hectic month. Many notable quants – Millennium, DE Shaw, Two Sigma – suffered losses at the end of March, leading some commentators to question if another “quant quake” like that seen in 2007 could be on the horizon.

Things have been looking slightly brighter for quants over the past two weeks, with Two Sigma and DE Shaw making slow gains in April according to Institutional Investor. But as the dust settles, some market participants begin to question a reliance on market data and purely quantitative strategies.

“Structural limitations have caused the majority of quantitative funds to perform poorly during the current market crisis,” said Daniele Grassi, CEO, Axyon AI in an email.

“While patterns of varying complexity can be found in asset behaviour, black swan events like the current coronavirus crisis can see asset behaviour break down completely.”

According to Grassi, quantitative models will continue to provide unreliable market predictions.

“This makes it very difficult for managers to make the right moves to navigate through the crisis.”

March and April performance proved Grassi right, but not all funds faced the same gloomy fate; several quants experienced gains while the rest of the market was plummeting. While the big funds took hits, smaller funds from Rotella Qdeck, Markin Asset Management and 1798 Q Strategy performed well according to Bart Kellerman, CEO, Global Capital Acquisition.

Grassi distinguished two categories for the few quant funds that performed well.

“While some had more sophisticated risk management models that were able to automatically detect the market chaos and allow them to quickly reduce exposure by selling off assets, others were simply lucky with their pre-crisis positioning on the market,” said Grassi.

Vickram Tandon, head of US business development at Tardis Group, agrees.

“The purely quant funds weren’t ready for it … They’re going to get hammered for a while.”

Rethinking risk models

An urgent and versatile approach to risk management is becoming increasingly important for quants. Markin Asset Management, which manages a portfolio with long term-only investment strategies, saw steady numbers in March and April, owing its success to a strategy that encompasses several models. During a Battle of the Quants webinar on April 16, David Marra, managing partner at Markin, said his fund’s risk model contained both pre-emptive and reactive responses to risk.

“It’s very important that before you’re in a time of uncertainty like this to be prepared for the unknown. And the way that we break this down is we think about it as knowing in advance where your trading model parameters are. Where is it that, in the market context, your model works and you know it works and know when market conditions are beyond the strategy parameters that your model is comfortable operating in,” said Marra.

According to Kellerman, this type of multi-faceted model hinges on reacting to violent markets by switching over to a new risk regime, and then switching back to the previous model when markets normalise again. Robust risk management also helps place faith in quants as favourable additions to investment portfolios, especially as trust takes a hit in coronavirus markets.

Grassi agrees.

“The adoption of new risk management technologies will also serve as an important marketing tool, instilling confidence in investors, who will see quant funds as more stable during future crises,” said Grassi.

From forecasting to nowcasting

As market data becomes more regularly available and black swan events threaten the accuracy of forecasts, forecasting may lose relevance. According to stats by, total spending on alternative data by buy-side firms will reach a projected $1.7bn this year, indicating a growth in firms’ desire for data.

An alternative to forecasting could be nowcasting, a practice that depends on quantitative models utilising vast amounts of data acquired through quarterly number releases. The practice reacts to real-time data rather than attempting to predict the future. It has recently received press from AQR’s former machine learning head who advised quant funds to embrace the practice.

“Unlike passive strategies, these strategies bypass and circumvent the delay and immediate information he receives and is able to react both on a systematic basis and a human trading basis. So it’s a combination of both – you get the human evaluation which is decided upon provided by quantitative analysis,” says Kellerman.

A 2019 survey by Lowenstein Sandler found that 80 percent of firms were using alternative data. With 90 percent of all data in circulation being created in the past two years, the boom in data availability makes the possibility of successful nowcasting a reality.

“And every day there’s an exponential growth occurring in data, it’s astounding when you look at the numbers and how much data is being mined. And only two to three percent of all the data in the world is actually being harnessed and categorised, so there’s still a massive opportunity there with regards to data and measuring things that move markets,” says Kellerman.

Alex Tsyplikhin, senior artificial intelligence (AI) engineer at Graphcore, is finding more and more quantitative finance companies utilise nowcasting through Graphcore’s Intelligence Processor Unit (IPU), which speeds up training for advance AI models. According to Tsyplikhin, nowcasting requires updating quantitative models as new data comes in.

“Excellence in data management, as well as making use of that data correctly is vital to the success of quants,” said Tsyplikhin in an email.

“In finance, data is typically very noisy and heteroscedastic. Traditional approaches are inefficient for estimating alpha and risk. We are now seeing a rise of probabilistic machine learning methods, for example, Markov Chain Monte Carlo (MCMC). This approach has been long considered too computationally intensive, but we’ve recently managed a 26 time speed increase, using our IPU processor.”

The human touch

While smaller funds seemed to have bypassed recent market chaos with nowcasting and robust risk models, Kellerman believes the poor performance of large funds is largely owed to statistical arbitrage. Statistical arbitrage, or the practice of buying pairs and relying on mean reversion to get them back, first gained notoriety in the late 1980s when it was employed by DE Shaw. The strategy has made the fund the fifth highest-returning hedge fund of all time, according to Institutional Investor.

DE Shaw’s hedge fund Valence – a fund based on statistical arbitrage – fell more than nine percent in March and has seen a slow move towards recovery.

“If [funds] were highly leveraged, if they had the wrong strategy, they got taken out. Very simple,” says Kellerman. “And there will be quant funds that have failed. And the mega-funds – the Millenniums, the Two Sigmas, the Systematicas, the DE Shaws – these massive funds that have long term quant trades and lot of statistical arbitrage trades, a lot of them got hurt.”

Yet Kellerman points out that despite losses, quants have on average out-performed discretionary funds over the past month’s turbulence.

“Quants typically succeed in a market that is moving, that is volatile, because they are putting on a lot of trades with a lot of companies. When markets don’t move, they have to lever up on leverage. They have to leverage tremendously in order to have small market moves make a difference,” says Kellerman.

“But when markets are volatile and moving all over the place, they can use less leverage and achieve their returns. Lately markets over the last five years, volatility has gone down and there are less traders in the market. But now all of a sudden that’s all reversed, and this is a market that the quants love.”

Quants will likely continue to grow and change in accordance to coronavirus market data. Some believe this will spur a move away from fully automated quant funds, and towards more hybrid funds.

“The models can’t take into effect this whole health scare: that’s the human element that has to be put in,” says Tandon.

“Some of these people that are constantly modelling their models, they’re sitting there and tweaking them consistently; they have a little bit of a human approach in addition to the quant approach and are the ones that are doing well and will always be in demand. I think they’ll be more in demand now than they have been in the past.”

1798 Q Strategy, a quant under Lombard Odier Investment Management, also exhibited a steady performance throughout the coronavirus market. Qaisar Hasan, the fund’s portfolio manager, believes its hybrid approach set it apart from the quants heavy funds that tanked.

“We actually tend to take a man plus machine approach, which allows us to be nimble and allows us to re-do our investment process and the way we manage risk very quickly, so we’re not just dependent upon finding historical patterns. How do you find historical precedents for the crisis we’re in today? There are none,” said Hasan during the Battle of the Quants webinar.

“In this environment we’re very focused on understanding what happens when different countries remove [coronavirus] lockdowns and people start to come back and resume normal patterns. Which industries tend to benefit more than others? Which industries continue to struggle? The key is just to make sure you have the right data and that you’re pointing it in the right direction and teaching the system how to evaluate the data correctly in a very unique environment.”

Morningstar reported that in August 2019, passive equity assets exceeded those run by humans for the first time ever. Whether or not these numbers will change due to recent passive fund hits is yet to be seen.

“Not having seen this before and this violence in the market of point swings, this whipsawing … a lot of quants just didn’t know how to trade this and pulled out of the market and suffered losses. The experience is going to be modelled, it’s going to be analysed, and it’s going to be plugged into every single model out there that wasn’t prepared for it so that when this happens again to this extent, they’ll be ready for it,” says Kellerman.

According to Tandon, quant funds will use this experience to innovate and improve, but the way they are perceived by portfolio managers will likely change.

“People are going to pause and reset how they look at [quants]. They’re not the be-all-and end-all anymore, there are a lot of more moving parts to it. I think people will be much more considerate,” he says.

“We’ve never really had a phenomenon like this before. Now you have to bring into account – what are the other phenomenon like this that can happen?”



Regulatory reporting: 7 Questions with Philip Flood, Gresham Technologies

Other | Behavior detection & predictive analytics Regulatory reporting: 7 Questions with Philip Flood, Gresham Technologies

Gresham Technologies
Real-time payments tech put pressure on banks

Best Practice | Behavior detection & predictive analytics Real-time payments tech put pressure on banks

Managed Services in 2021: Poised for Lift-Off

Best Practice | Behavior detection & predictive analytics Managed Services in 2021: Poised for Lift-Off

SmartStream Technologies