FX traders going ‘quantamental’
Euromoney Limited, Registered in England & Wales, Company number 15236090
4 Bouverie Street, London, EC4Y 8AX
Copyright © Euromoney Limited 2024
Accessibility | Terms of Use | Privacy Policy | Modern Slavery Statement
Foreign Exchange

FX traders going ‘quantamental’

A growing number of FX traders are extolling the virtues of combining quantitative and fundamental analyses into an approach known as quantamental.

Traders are constantly looking for new insights into quantitative and fundamental data via more comprehensive data sets, faster delivery or new types of data that provide unique insights into a currency.

Celent analyst, Brad Bailey, notes that most firms already use both quantitative and fundamental data to inform their trading decisions. “Even the most fundamental trader will take advantage of unusual price action to add to — or lighten up — a position,” he says.

However, when quantitative traders use only time series data they often intentionally disregard fundamental factors, even though similar price patterns may be caused by totally different fundamental factors and vice versa.

In this scenario, ‘pure’ quantitative models start to underperform if they were developed only using price patterns and then the pattern falls apart because it was caused by totally different market processes, says Alex Krishtop, founder of Edgesense Solutions.

“I have always developed quantitative strategies based on trading ideas derived from certain real market processes as opposed to abstract mathematical concepts or machine learning,” he says. “Therefore, when the term quantamental emerged I began to use it in this sense and noticed that so did many quant traders. For me, quantamental means finding a trading idea, suggesting a qualitative description for it and then building a quantitative model that reflects this qualitative description.”

David Lopez Onate,
Forex Artilect

Krishtop suggests that fundamental factors — such as changes in the market structure — are perhaps the most influential and yet the most overlooked factors affecting the performance of buy-side FX strategies.

“As it is very difficult to formalise factors related to market structure to use as numbers, most models disregard them,” he says. “However, if this results in reduced performance and the trader doesn’t understand the reason for this degradation, they may find it hard to decide when to stop, or whether to stop at all.”

Changes in market structure (for example, market makers taking reduced risk) can result in greater fragmentation and thinner liquidity, which affects the performance of buy-side FX strategies in a number of ways, explains Vikas Srivastava, head of investor FX and business development at Integral.

“Increased direct and indirect transaction costs will negatively impact the strategy alpha,” he says. “Secondly, the capacity of these strategies will be restrained. In light of these liquidity challenges, it becomes useful for the buy-side to include options such as client-to-client or all-to-all midrate matching in their toolbox. Midrate fills with zero transaction costs can reap big benefits for strategy alpha.”

For those trading mid- to long-term, the underlying premise of the trading strategy must be almost completely reconsidered in the event of structural market changes, adds Forex Artilect founder, David Lopez Onate.  “That is a constant challenge the trader must address to keep his trading system on the edge,” he adds. “The skill of the trader should be to identify the signals issued by the market to anticipate any upcoming shift.”

According to Krishtop, the most efficient way of improving FX strategies performance would be to moderate a quantitative model using a qualitative analysis of various fundamental factors. “In addition, the quantamental approach should produce more stable results with quantitative models compared to those designed using traditional technical analysis or machine learning,” he says.

According to Lucena Research CEO Erez Katz, the most appropriate approach to utilizing machine learning for FX is dynamic model creation, which continually assesses which factors matter based on the most recent historical conditions. “In short, a healthy machine learning approach should have data that supports different conditions in the market and should create a dynamic assessment of the model based on which future price and volatility are assessed,” he adds.

However, Institutional FX Advisory Partners founder Henry Wilkes remains unconvinced, suggesting that it is very expensive to add quantitative analysis with sophisticated systems to analyze mountains of data for an asset class that is for some traders a secondary activity linked to the primary activity of investment securities trading.

“The quantamental approach tends to be used more frequently in the strategy decision by the buy-side rather than the execution phase where they prefer to use the click-to-trade rather than execution algorithms,” he says.

However, there is some evidence to suggest that this may be changing. An online FX e-trading survey conducted among institutional traders by JP Morgan in late 2016 found that while click-to-trade accounted for 83% of trading, almost four in 10 respondents said they would increase their use of algos in 2017 and 2% said they would make less use of click-to-trade.

Flextrade senior vice president FX David Ullrich observes that in a quantitative world, FX traders should never override quantitative trading signals since the introduction of discretion negates the trading framework.

“However, an argument for a better trade execution process measured via implementation shortfall managed by a discretionary trader could easily be made,” he concludes. “Regardless of whether the process is quant or fundamental or a blend, an iterative and analytical review of the trading process via a pre-trade/post-trade t[transaction cost analysis] system is necessary to improve the trading outcomes.”

Gift this article