How I Got Into Algorithmic Trading (and StrategyQuant)

A while back, StrategyQuant sat me down for an interview about how I ended up doing this for a living. I still like that conversation — but it was a snapshot of one moment, and a good deal has changed since. So here is the fuller version, in my own words and brought up to date.

A note in passing: I work at StrategyQuant as its AI Solutions Architect, and the one StrategyQuant link below is an affiliate link — if you start there through it I may earn a commission at no extra cost to you. This post is a story, not a sales pitch, but I would rather you know that up front.

It started with a typo

The honest origin story is not glamorous. I got into trading, in part, because of a phrase I once misspelled into a Google search. What came back sent me down a rabbit hole, and somewhere in it I found the word for a thing I had already half-suspected about myself: that of all the ways to approach markets, the systematic one fit me best. I did not yet know how to do it. I just knew that rules I could write down and test appealed to me far more than opinions I had to defend.

The desk years

Long before any of the automation, I earned my scars the manual way. I started in 2007 on a proprietary futures-and-options desk at Colosseum, trading live across the world’s major exchanges — CME, CBOT, EUREX and the rest — in futures, options and ETFs. It was a sell-side seat at a futures broker, and it taught me things a backtest never could.

Chief among them: my own judgement, minute to minute, was the least reliable part of the process. I remember very well what it is like to stare at charts all day, and to let a loss or a drawdown talk me out of a plan I had been right to make. Those losses were the real teachers. They are also the reason I eventually wanted the decisions taken out of my hands and put into something I could measure.

The tool that changed everything

I had been using StrategyQuant since 2014, but there was a specific breaking point — a version of the platform that finally let me automate the part of my workflow I had been grinding through by hand. That was the moment the whole thing tipped from “software I use” to “the way I work.”

Two other things clicked around the same time. I started using StrategyQuant alongside Python — a simple enough language that I could write quick scripts for data analysis whenever I needed one — and I stopped working entirely alone. I teamed up with another StrategyQuant user in a small, high-trust partnership. I have never believed in big, inefficient collaborations, but two people with a large amount of data and a lot of ideas move a great deal faster than one.

What I actually love about it

People assume algo traders are in it for the automation. For me it comes down to three words: verifiability, modelability, objectivity. I like being able to invent something, model it properly, and then check whether it is real — rather than argue about whether it is. And after years of watching charts all day, having strategies that trade without me signing off on every position has a value that is genuinely hard to overstate.

The philosophy, in one breath

If I had to compress everything I have learned into a few lines: the more I learn, the less I know. Simple strategies, and the more data the better. Almost every serious mistake I see — my own included — traces back to overfitting, so most of my process is built to fight it: statistical hypothesis testing, large samples, and a healthy distrust of any backtest that looks too good. If you want the long version of that argument, it is the whole point of what 1.2 million FX strategies taught me about overfitting.

I have also found, testing it directly, that strategies tend to start losing their edge after about two years. That single fact reshapes everything: the work is never finished, and a good process is a pipeline for producing and retiring strategies, not a trophy cabinet for one. On the portfolio side, the ideal I keep chasing is a set of uncorrelated strategies across as many markets as possible — a pool that has passed every test, decorrelated with our own Python scripts, and assembled into something that does not all fall over at once.

Where it went next

When StrategyQuant interviewed me, I was writing indicators by hand and building portfolios with Python scripts. I never stopped doing that — I have since published 135 free tools to the StrategyQuant Codebase — but the work has moved up a level. Today I am StrategyQuant’s AI Solutions Architect, and most of my energy goes into an agentic AI layer that turns a plain-English idea into a finished, verified strategy. You can see that on my strategy lab page.

The tools change. The obsession — building things you can actually trust with real money — does not. That thread runs unbroken from the misspelled search, through the desk, to whatever I am building this week.

If you want the shorter career version, it is on my about page. If you want to build on the same platform I do — and pick up my private toolkit while you are at it — here is how that works.