Forge FateQuant & Data Research
Methods for evaluating investment ideas through data, backtesting, costs, risk, and reproducible research records.
Can Logistic Regression Beat the Baseline? What Makes a Model Comparison Fair
Define timing, data splits, retraining, and cost rules for a fair comparison of logistic regression and baseline strategies, before testing actual performance.
Would Adding Transaction Costs Change the Conclusion? Checking Accounting and Risk Paths in a Synthetic Portfolio
Use a synthetic portfolio to check transaction-cost accounting and distinguish what ending returns, turnover, and maximum drawdown reveal.
Start a Backtest with Positions, Not Returns: One-Period Accounting for Buy and Hold vs. Momentum
Check holdings, cash, and ending values for buy-and-hold and momentum using three synthetic assets, establishing accounting rules before market testing.
Why shift(1) Is Not Enough: A Time Contract for Signals, Execution, and Labels
Separate signal, execution, return, and label timing. Use synthetic unit tests to check information leakage and preprocessing boundaries.
Part 2. What to Check Before Trusting Investment Data: Acquiring and Validating U.S. ETF Data
Plan access terms, price definitions, raw-data preservation, and quality checks before acquiring U.S. ETF data, including missing values and adjusted prices.
What to Decide Before Backtesting: Questions and Rules for U.S. ETF Quant Research
Set the research question, trading rules, baselines, and time splits before a first U.S. ETF study. Record the experiment contract before seeing results.
Does AI Actually Help with Quantitative Investing? A Conditional Conclusion and Standards for Validation
Separate AI's research value from investment profitability and define evidence for usefulness through chronological testing, baselines, and reproducible records.
11. A Safe Way to Do Quant Research with LLMs: From Questions to Validation and Research Records
Organize LLM-assisted quant research around questions, evidence, code checks, timing, and experiment records, with human review and documented failures.