The Model

Today’s markets are increasingly characterized by the continuous arrival of vast amounts of information. The role played by private information in the price discovery process is now well documented. The rate at which informed traders exploit private information varies from split second time frames to steadily across trading days.

I continuously follow eight datasets, including:

• periods of accumulation and distribution by institutional investors who execute trades off-exchange … within venues known as dark pools,
• measures of option market makers’ delta-hedging requirements as they relate to changes in price, volatility, and time,
• institutional demand for protective puts and timing related to call overwriting,
• cross-asset order flows between S&P 500 index options and S&P 500 futures,
• steepening in the volatility skew for out-of-the-money options,
mispricing of S&P 500 index options*, which might reflect “…need or greed,”
• auction market-related price responses, combining the market profile graphic and cumulative volume delta plots, and
high speed algorithmic trades and information arrival rates.

Price Prediction Using Machine Learning

My dataset is composed of a set of variables, evaluative of institutional order flows and volatility in the S&P 500 index options market. Using a supervised machine learning technique, 12 multivariate scenarios indicative of changes in sentiment and predictive of imminent reversals … run simultaneously.  By superimposing multiple signal maps atop a graphical representation of price, I am able to visualize the market’s predisposition to a change in course.