Precision Alpha
Exact Approach to Identify Alpha

It is now well established that financial markets generally and individual securities specifically are often not in equilibrium for extended periods of time. Founded in 2015, Precision Alpha has changed the traditional approach to equity prediction by using analytical Machine Learning on historical price information to identify alpha in non-equilibrium financial markets. The company was founded by two veterans of Wall Street Mark Temple-Raston and Michael Lucente, who witnessed first-hand the challenges of non-equilibrium markets owing to the lack of a precise analytical solution. "Today, our proprietary, patent-pending technology platform leverages an exactly solvable (analytical), non-equilibrium machine learning to scientifically analyze and enrich financial market data to offer unbiased advanced insights, trends, alpha and predictive signals for investment environments that are not in equilibrium," says Temple-Raston, who is currently donning the role of CIO and Chief Data Scientist at Precision Alpha.

The Precision Alpha platform can either process closing prices from NYSE or NASD (or 85+ other global exchanges), or, Precision Alpha can process the customer’s own data – for any market, including non-equity markets - through Precision Alpha’s self-service interface.

Precision Alpha for Exchanges delivers a daily text file about 5 hours after market close with all machine-learned measurements for all securities on the selected exchange. For example, six months of NYSE closing-price measurements and non-equilibrium machine learning is used to decompose closing prices into: next day probabilities, market energy, market power, market resistance, market noise, market temperature, and the market energy available to move prices (market free energy). "Our calculations forecast everyday moves for all stocks listed on multiple exchanges worldwide, to eliminate estimation, bias/assumptions, and guess work from the calculation," says Temple Raston.

For the Precision Alpha self-service API, the Machine Learning calculations are performed so that the customer’s data is not stored on the Precision Alpha platform."All of our mathematics is implemented in a Lambda function on AWS that enables us to process everything in memory and then push it back into a clients' S3 bucket upon completion.

Our proprietary, patent pending non-equilibrium ML technology scientifically analyzes financial market data to offer advanced insights, trends, alpha and predictive signals.

Meaning, we never have to store the customer’s data. We do the scientific processing and give it right back," says Temple Raston..

Integrations through AWS are not difficult, even for self-service. "We just need to set up the AWS security policies with the client, and we are good to go." adds Temple Raston.

To put things into perspective, a long/ short on open/close investment strategy for US equities generated 35+% returns per annum over three consecutive years with Sharpe Ratios above 5 by one of our clients. The returns are impressive precisely because of the non-equilibrium trading strategy. They have consistently been able to see price moves 1 – 30 days in advance of the actual price changes. One portfolio manager commented, “We were able to see the 2021 market correction in QQQ that began on September 8 before it occurred, allowing us to buy protection, and also see the turn higher from October 4. The changes in individual securities are not always dramatic but we see extraordinary opportunities offering pure alpha every day.”

Markets are plainly not in statistical equilibrium. In fact, Precision Alpha’s current alternative data clearly shows that equities behave as dissipative structures. That is, they oscillate far from equilibrium, round a “steady state” that is itself remote from statistics. Both tornados and hurricanes are examples of dissipative structures. Using Precision Alpha, we see dissipative structures throughout our financial markets, affecting price movement. In Spring 2022, Precision Alpha will extend their data set to include new dissipative structure specific measurements.