Antonio Marrazzo is a quantitative researcher with a background in economics and actuarial science, focused on factor investing, portfolio construction, market regimes, data analysis, and machine learning in financial markets.

Originally from Argentina, Antonio began applying quantitative methods to investing before formally discovering the quant profession. He translated concepts such as Markowitz portfolio optimization into Python, built his own research pipelines, and developed a systematic approach to understanding markets through data.

In this episode, Antonio joins us for a detailed conversation about how quantitative strategies are researched, tested, and transformed into realistic investment models.

The central question is:

How do you know whether a factor is real, tradable, and persistent rather than the result of statistical chance?

Antonio explains why a factor should be supported by economic reasoning and a plausible relationship with returns. We discuss multiple-testing bias, false discoveries, factor orthogonality, risk premia, and why testing thousands of ideas will almost always produce something that appears statistically significant.

We also explore why factors must be evaluated using realistic and tradable investment universes. Antonio explains how attractive historical results can disappear after accounting for microcap exposure, liquidity constraints, bid-ask spreads, short availability, transaction costs, and other practical limitations.

The conversation goes deeper into multi-factor portfolio construction and dynamic factor allocation. Antonio shares why each factor may perform differently across market regimes, including why momentum can behave better in lower-volatility environments while short-term reversal may become more relevant when volatility increases.

We then examine one of the most important parts of quantitative research: data quality. Antonio discusses missing values, outliers, time-zone alignment, interpolation, point-in-time data, reporting dates, survivorship bias, look-ahead bias, financial statement revisions, and why researchers must understand exactly when information became available to the market.

Finally, we explore machine learning in quantitative finance. Antonio explains why feature engineering can matter more than selecting the most sophisticated model, why classification may be more practical than directly predicting returns, and how models such as logistic regression, random forests, and XGBoost can identify relationships that traditional linear methods may miss.

We also discuss expanding training windows, stationarity, realistic labeling, meta-labeling, purged cross-validation, embargo periods, and the importance of including trading costs throughout the research and validation process.

A technical and practical conversation on factor investing, regime-aware allocation, data engineering, machine learning, portfolio research, and the discipline required to avoid fooling yourself with attractive backtests.

*DISCLAIMER*

The information shared on this podcast is for educational and informational purposes only and reflects the personal opinions of the hosts and guests at the time of recording. Nothing in this podcast constitutes financial, investment, legal, tax, or trading advice, and nothing should be interpreted as a recommendation to buy, sell, or hold any security, cryptocurrency, derivative, or financial product.

Trading and investing involve substantial risk, including the possible loss of all or part of your capital. You are solely responsible for your own decisions, and you should consult a qualified professional before making financial decisions. By listening to this podcast, you agree that the hosts, guests, and producers are not liable for any losses or damages arising from the use of any information discussed.