Algo Trading

I want to share a framework of trading algorithmically. This page will be updated as often as possible with my template of building my Flobot. This template has evolved over many years as both my skill and the technology has evolved.

For me, trading with an algo means that I turn the algo and don't turn it off except for week-end computer maintenance. If I mess with an algo then the math and data mining on which it was built is invalidated.

I trade a portfolio of algos as I have shown in blog posts. The portfolios are constructed to create the smoothest possible equity curve I can make with the lowest drawdowns.

I have more than one portfolio trading different timeframes. I have a portfolio that generates signals for trading stock and index options.

Here, you can expect to see a step by step process of what I did and do to create a portfolio of robust algos. While I am not a great innovator, I am very good at taking an idea and making it much better. That is what I believe I have done.

My journey to algo trading started in the early to mid 1980s. In those days the algo gaave a signal and I telephoned the floor (broker in the pits) and placed the order. It all started while I was living in Connecticut in the U.S. and read an article in the New York Times or Wall Street Journal about A guy called David Aronson who was using computers to create optimized trading strategies. I tracked him down in New York City. It turned out that it cost $25,000 to do a single optimization. I passed, sadly. I then came across Bob Pardo who was selling software for the new IBM PC that could optimize. I ended up doing some testing for Bob.

While the early 1980s algos were very primitive compared to what exists now, they really showed the possibilities. The Cruz brothers released some algo software that became TradeStation some years later.

So that's my background. Now let's look at what I do now.

I create portfolios of algos using state of the art genetic data mining applications. There are a number of these applications out there but the one I use is called StrategyQuant. I chose StrategyQuant not only because of its ability to create profitable algos but because has the best tools to comb through the thousands of profitable algos it can create in one run and eliminate the ones that are not likely to be robust.

Let me be clear, using these genetic programming applications makes it possible to create algos with a positive profile. The art in creating algos is being able to discover which algos are likely to remain profitable for a significant period of time. StrategyQuant has the tools and I have create the processes and procedures to use those tools to find the gold amongst the rocks.