It's Time—Deploying The Quant Super-System [Code Included]
We discovered the alphas, tested them, and now, it's time to launch.
A few months ago, we started with the lofty goal of discovering statistically-significant and uncorrelated alphas to ultimately merge them into one, beautiful, dynamic system that would earn us profits year after year.
Well, the time has finally come.
For the first time ever, we’ll be doing a deep dive into the full pipeline of a multi-strat quant system from high-level system design, to automated execution via our designated platforms.
There’s a lot of ground to cover, so without further ado, let’s dive right in.
Connecting The Puzzle Pieces
While just one strategy has many moving parts, there’s no need for it to be complex and esoteric. Before diving into any particular strategy, we first need to establish a general structure of how our systems should work:
Over the past few months, we’ve established a relationship with Polygon’s API and we store our transformed data in SQL databases. Without data sources, there is no system — so naturally, it is our core starting point.
The signal generation stage depends on what strategy we’re aiming to execute. We’ll go over the exact strategies in a moment, but this stage may be something like our volatility trading approach, where we pass in the historical data and get a prediction based on current data.
Based on our pre-set criteria, we determine if the generated output is a signal to trade (e.g., confidence over 65%). If it isn’t, we simply terminate the process and wait until the next session to test again. If it is, then we go on to executing the trade.
This process represents the birds-eye view of how our systems will work. So, with that general structure in mind, let’s walk through implementing our first system.
System 1: Overnight Sector Forests
Our first alpha will be one we’re familiar with, but we’ll tweak the recipe just a bit. Let’s just do a quick recap of the original approach: