Paper Trading Portfolios

The 1987 film "Wall Street" features an antagonist by the name Gordon Gekko. Gordon Gekko is an antihero for the financial world in the aspect of being the ultimate asset manager and wall street entity. Gekko is depicted, initially, as all knowing, all powerful, and rich beyond belief.

As the plot develops we learn Gekko is much more than lucky and employs dangerous/fraudulent practices to profit. Gekko's famous quote from the film encompasses his ego/wisdom: "Greed, for lack of a better word, is good."

This film is an exceptional depiction of the duality found in finance and was released in the same year as "Black Monday", a financial crisis related to the global financial system becoming increasingly intertwined with globalization efforts and risk elevation. Financial markets buckled under extreme stress due to new dynamics of derivative use, excessive speculation, and immense amounts of capital flowing into the United States from around the globe.

Since the early 1980s, financial firms and asset managers as a whole have benefited from the immense improvement in technology. These parties have turned to utilizing technology to improve the efficiency and availability of markets as well as improving strategy deployment.

Algorithmic trading, put simply, is the employment of technology in the management of assets with minimal human interaction. The rabbit hole of algorithmic trading is as dark and deep as they come. These methods are used in every major market and it is difficult to speculate the true proportion of all financial transactions occurring within these methods.

Algorithmic trading provides a multitude of benefits, but a single benefit stands high above others. Algorithms are truly "rules-based" and are not affected by emotion. Humans are emotional creatures. This is especially true in relation to finance. Greed and Fear are exceptionally strong emotions which have plagued humans since the early days of civilization.

Gordon's brain is rather simple. He makes a direction prediction on focused securities every day the United States equity market is open. He does this in two parts. The first being the generation of a model for each target security based on historical data.

Nine parameters are used in this model and features data starting March of 2020 and ending at the end of every current quarter, i.e. the model is updated at the end of every quarter. Gordon's modeling utilizes "current environment data", meaning he validates models on an out-of-sample dataset of the most recent quarter.

The second portion relates to how he utilizes these trained models. These models are static and unchanged until the end of each quarter, but they utilize constantly updating parameters. Gordon also works to forecast each parameter in a t+1 fashion - predicting what these nine inputs will be tomorrow, in order to predict which direction a security will take tomorrow.

Gordon is allowed to trade based on his direction predictions using a mean-reversion based trade execution system. This system looks to enter and exit trades when a given security is poised to return to an "average" value.

The signals to execute are paired with Gordon's direction prediction. When they align, Gordon performs a trade. Pairing both Gordon's predictions and the system's signals allow for a higher probability of positive trade outcomes, in theory.

This trading system is hooked up to a brokerage with 100% margin on invested capital and allows Gordon to buy and short target securities as he sees fit. He is allowed a set amount to buy/short in any new position, cannot enter a trade if he doesn't have adequate capital, and sells full positions if necessary. These rules allow Gordon to be entirely autonomous in his trading.