Regression-based statistical learning and transaction cost analysis for systematic developed-market FX forward trading with macro factors

Research evidence excerpt

Regression-based statistical learning and transaction cost analysis for systematic developed-market FX forward trading with macro factors

Products & Services J P M O R G A NJPMaQS

03 July 2026

Regression-based statistical learning is a convenient and transparent method for combining trading factors into composite signals.

Sequential statistical learning considers only the data available at each time point to choose and parameterize the “best” model

and to generate signals without hindsight bias. Yet assessing PnL potential in backtests also requires estimates of transaction costs

as a function of the notional amounts traded. Such estimates can be generated by (i) using LLM prompts to retrieve key cost

parameters and (ii) feeding them into appropriate Python functions for cost imputation.

This article illustrates regression-based learning and transaction cost analysis for systematic developed-market FX forward

trading with macro factors. The sequential learning process generates signals with highly significant predictive power. Risk-

adjusted returns may look modest, but they have been uncorrelated with major market benchmarks, and the statistical probability

of long-term value generation is very high. Moreover, transaction cost analysis attests to the scalability of a low-frequency

strategy in liquid markets, showing very large average annual PnLs over time.

Full post and python code here

Empirical findings in this note are based on the J.P. Morgan Macrosynergy Quantamental System (JPMaQS), a joint dataset

produced by J.P. Morgan and Macrosynergy delivering point-in-time macroeconomic indicators for systematic trading research.