A Talk I Gave to Columbia University’s Practitioner's Seminar Series — For the last four years it has been my pleasure to be one of the guest speakers to the program for a Master’s in Financial Mathematics given offered by Columbia University, in New York. Mathematics of Finance at Columbia University - Program - General Information The Department of Mathematics at Columbia University offers a Master of Arts program in Mathematics with specialization…www.math.columbia.edu

Good survey of the reality of the data. The market is not Normal. Student's t is a good try, but it's analytically problematic for asset allocation strategy (utility calculations are sensitive to the moments of the distribution and Student's t will *always* have divergent moments at some level in the real world). I like the Generalized Error distribution (a/k/a Generalized Normal or Exponential Power distribution), which doesn't suffer from this problem. Also, clear evidence for Heteroskedasticity which will act to decrease kurtosis when properly accounted for.

Using a Synthetic Alpha to Remove the Composite Hypothesis Problem — Backtesting, Alphas and Trading Strategies For quants, trading is the process of repetitively taking positions in the markets as a function of information known about the future distribution of the returns of an asset. In the process of analyzing this, I generally draw a distinction between: alphas, which is formally the expected mean of future…

What’s the Utility of Future Returns to Traders? — My work on trading strategy analysis departed from Mean-Variance Optimization, as studied by Markowitz and developed in most finance theory text books under the name Modern Portfolio Theory, the moment I had to consider three aspects of trading in the real world: you want to be trading perpetually; transaction costs…

The Autocorrelation of NASDAQ–100 Member Returns — In a prior article here on Medium, I showed how to compute a benchmark reference table for scoring the performance of machine learning algorithms that purport to predict the performance of NASDAQ–100 stocks. …

Here’s How I Know That and Why It’s Important — Inflation is an increase in the general level of prices. As Milton Friedman declared, this is a monetary phenomenon. There are reasons why this is bad, such as the wage-price spiral and the observation that inflation is autoregressively heteroskedastic (meaning that changes in inflation lead to increased uncertainty about future…

Generating a Table of Forecasting Scores for the NASDAQ-100 — There are many articles here on Medium that purport to show how to use machine learning algorithms to predict stock prices. Many exhibit common flaws that show that the writer has no actual experience in alpha generation. Two that are particularly prominent are: predicting prices not returns; and, using simple…

An Example of Online Learning with a Discrete Time-Series — One of the experiments I’ve been running on Peter Cotton’s Microprediction site has been to track the performance of a Tactical Asset Allocation strategy that offers the choice between investing in Meme Stocks or treasury bills. I’ve already written about this on Medium, and the article is linked to below: Adventures with Microprediction Financial markets are the ultimate laboratory for the Wisdom of Crowds. This is the concept that a group of people,…link.medium.com

Financial Data Almost Might have been Designed to Beat You — The internet is full of forums, such as Towards Data Science here on Medium, that are full of eager articles explaining how to use today’s modern “AI” tools to predict stock prices. I find the articles they publish follow a common pattern: The author provides code snippets, almost exclusively in…