Purpose

Merlin's aims to predict whether the monthly price change of the S&P 500 could be negative using fundamental and macroeconomic data.

Model Type

Merlin is a 'top-down' research tool, not a trading system. Its development started in 2007. It combines macroeconomic models and machine learning algorithms. These algorithms are used for regression and classification purposes.

Merlin analyzes the risks associated with: changes in consensus EPS for the S&P 500, the pace of US/Global economic growth, inflation expectations, monetary policy adjustments, tightening of financial conditions and credit availability, absolute and relative valuation of the S&P 500, investment portfolios flows and Investors / businesses / consumers’ confidence.

Key Output

On the first day of the month, Merlin estimates the probability that the performance of the S&P 500 Index will be positive for the month that begins. This probability is updated during the month.

Live Use

The current version of Merlin went live in September 2018.

Performance

The average predictive accuracy is greater than 70%. The exposure to the cash S&P 500 is 0% or 100%. Please refer to the "Results" section for live results, and long-term out-of-sample simulations.