Trading Solution

This is my research workspace for studying when to hold S&P 500 exposure and when to hold cash. I’m interested in simple, low-frequency rules and in understanding what a backtest can actually tell us.

Signals and data

The project builds long-term credit-regime and short-term stress indicators, then uses them to study allocation rules. The codebase includes data-loading and validation tools, weekly reporting, and comparisons with price-based trend rules such as the 200-day moving average.

The current research focuses on rule-based strategies. Earlier machine-learning experiments are preserved in a separate archive so their assumptions and results remain distinguishable from the active research.

Making comparisons meaningful

The research tools track signal timing, execution timing, transaction costs, drawdowns, and trading frequency. Recent experiments compare strategies under a common evaluation framework and enforce an annual action budget.

A recurring challenge is data availability: revised macroeconomic history is not necessarily the information that would have been available at the time. That distinction matters when interpreting historical results.

Current scope

This is a research and reporting project. It does not establish reliable excess returns, and I’m not presenting backtested outcomes as live trading results. My focus here is the process: forming a hypothesis, making the comparison reproducible, and documenting what the evidence supports.

I plan to publish more detailed notes on the strategy design, data assumptions, and findings as the research develops.