Read on Twitter
#HFT Here's an example of an HFT strategy. Keep in mind this is 1 in a billion examples. This isn't investment advice. Don't trade it. I'm in a unique position where I can share strategies without getting sued cuz it's my company & IP.
#HFT Part 1: Important terminology. It's hard to jump into an HFT strategy without explaining some terminology first.
A “feature” or “predictor” is any kind of basic independent variable that we think has some predictive value. Some call it a “feature” if they’re from the machine learning world, others call it a “predictor” or “regressor” if they come from a statistics or econometric background.
A trading signal is what you get when you take one or more predictors, and feed it to any kind of mathematical model to create a single predictor or trading decision. You can also “stack” multiple of these models...
i.e. feed multiple models into another model that aggregates them into a single predictor or trading decision. In this case, you’d sometimes refer to the intermediate model outputs as “meta-features”.
A trading rule is a hardcoded trading decision. For example, “if there is only 1 order left at the best offer, then lift the offer; if there is only 1 order left at the best bid, then hit that bid”.
A strategy puts together a bunch of trading decisions, by some combination of trading rules and mappings of signals to trading decisions.
A strategy that is only based on trading rules is a rule-based strategy.
A strategy that is based on signals or models is called a model-based strategy.
A strategy that is based on signals or models is called a model-based strategy.
It's common to combine the two, but if I had to choose one, we leaned more towards model-based strategies because every rule you can possibly imagine has probably been done before.
That's enough for today. This is a multi-part series on building an HFT strategy. Next week I'll talk about examples of features.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
