JP Morgan Alternative Asset Management (JPAAM) is on the hunt for short-term quantitative investment strategies, according to Pierre Chabran, the group’s head of hedge fund solutions.
Chabran told the Nikko NZ Investment Summit audience in Wellington last week that JPAAM is favouring quant managers that adopt ‘machine learning’ tools to eke out market alpha.
He said while the current market was supportive of quant strategies in general, technological developments and the rise of ‘big data’ offered particularly interesting opportunities for statistically-based investment approaches.
“The explosion in computing power and availability of data has opened the door to lots of research and innovative strategies,” he said. “These are well-understood techniques but they haven’t been applied to financial markets before.”
JPAAM divides quant managers into two fundamental investment styles: prior-based, and; machine-learning.
Chabran said ‘prior-based’ quant styles assume “an economic or behaviour relationship exists that can be quantitatively verified”.
Prior-based managers follow a series of logical steps starting from formulating the investment idea through to portfolio construction and optimisation.
However, ‘machine-learning’ managers allow the data to direct investment decisions without applying any ideological framework.
“Machine-learning is a sophisticated form of data-mining,” Chabran said. “It lets the data build the investment models, which might make intuitive sense or not. But [the models] need a high level of statistical significance, which leads to a shorter time-frame.
“The time horizon is important. The shorter time frame strategies are less correlated and that’s why we favour those managers.”
But he said machine-based quant strategies tend to be “less scalable” than most others making it difficult to allocate a large portion of a portfolio in one hit.
Chabran also refuted claims that the inherently opaque style of algorithm-driven investments obscured meaningful research.
“We think the ‘black box’ concerns are unfair,” he said. “You can understand these investments if you put in a lot of work. It’s not about gaining access to the algorithm – you’re not buying the system, you’re buying the investment process. You have to understand the principles, the team and the process.”
According to Chabran, current market conditions were accommodative for quant strategies in general based on a number of factors including:
- High levels of volatility;
- High trading volumes;
- High, and increasing, levels of dispersion across and within markets; and,
- Strong trends across several markets (for example, currencies and oil).
But he warned quant investors have to consider a number of risks such as:
- Operational risk – or ‘fat finger’ failures;
- Data fitting;
- Adaptability – to ‘regime shifts’ and underlying market conditions;
- Defence mechanisms against unusual market conditions;
- Drawdown management plans;
- Financing structures – for example, leverage and liquidity; and,
- Crowding – where competitors jump on successful strategies and potentially erode their edge.
Nonetheless, hedge fund-of-funds such as JPAAM were spoiled for choice of quant managers with the underlying supply increasing in recent years, he told the Nikko audience.
Chabran said as banks have been forced to spin off their proprietary activities post GFC, the pool of talented traders offering quant products has increased.
The barrier to entry into quant management has also lowered as a wider range of algorithms become available, he said.
However, Chabran said it was important investors didn’t pay hedge fund fees for access to “alternative beta”.
He said investors have to clearly distinguish between genuine non-correlated quant strategies and simpler, factor-based approaches masquerading as hedge funds.
Nikko offers JPAAM as its alternatives manager to New Zealand investors.