4. High frequency and the bridge to the tech track

Theory 6: A survey of real strategies

There are firms that pay to move their servers a few meters closer to the exchange's. Not to go a little faster: to gain microseconds.

We close this level, and with it the complete theory course, with the family of strategies that most captures the popular imagination: high-frequency trading. And with an honest look ahead: what it actually takes to go from understanding these concepts to building something that really executes them.

HFT is not a strategy, it is a scale

High-frequency trading (HFT) is not, strictly speaking, a different strategy from the ones we have already seen at this level. It is an extreme way of executing them.

It consists of applying statistical arbitrage, pure arbitrage or market making on a time scale of microseconds, or even nanoseconds. At that scale, the competitive edge depends less on the intelligence of the statistical model and more on the physical speed of the infrastructure.

That is why HFT firms invest considerable sums on three fronts:

• Colocation: placing their own servers physically as close as possible to the exchange's servers, to cut signal latency by microseconds.
• Dedicated fiber-optic, or even microwave, links between different financial centers.
• Specialized hardware, capable of processing and reacting to market data faster than any competitor.

At this scale, winning or losing a latency race by a few microseconds can be the difference between capturing an arbitrage opportunity and arriving systematically late to all of them.

As already noted in the previous module, it is a competition in which an individual investor simply cannot participate competitively. And one in which there is no sense at all in trying.

Where you can compete

This does not mean that the rest of what you have learned in this course is out of reach for an individual with modest resources.

The momentum, mean reversion and pairs trading strategies from the previous two modules operate on much longer time scales: from minutes to weeks.

There the edge does not depend on winning a microsecond race against a firm with colocated servers. It depends on three other things:

• Having a reasonable statistical hypothesis.
• Backtesting it rigorously, avoiding the traps of the previous level.
• Executing it with the systematic discipline that automation provides.

That is precisely the ground on which an individual with programming knowledge can build and run their own systems, with a realistic expectation of competing on reasonably level terms.

The bridge to the technical track

And this is where this theory course connects with the technical track on this same platform.

Everything you have learned across these six levels is the conceptual map you need before touching a single line of code. How price forms. How an order is executed. How to read a company. How to measure risk. How to design and evaluate a strategy. And what families of strategies exist.

What comes next, in the technical track, is learning the tools to build that map in real software:

• Python as the language. The technical track starts literally from zero in programming, assuming no prior experience.
• Libraries for downloading and manipulating market data.
• APIs for connecting to brokers and exchanges.
• And finally the implementation of backtests and real strategies on historical data.

They are exactly the same concepts of signal, entry, exit and position sizing we have covered here, now translated into executable code.

Programming is not compulsory

You do not need to program in order to understand markets in depth. Nor to make better discretionary investment decisions with informed judgment. That has been, precisely, the aim of this theory course from beginning to end.

But perhaps the natural next step is to genuinely automate one of the ideas we have walked through across these six levels. In that case, the Python technical track — reachable from the 'Courses' section of this same platform — is exactly where this story continues.

The last thing, and the most important

One last idea deserves to be stated plainly before we close. No concept in this course, on its own, constitutes a guaranteed profitable strategy.

Momentum, mean reversion, statistical arbitrage and the other families covered at this level are conceptual frameworks with reasonable empirical evidence behind them. They are not magic formulas that work identically in any market, any era or any particular implementation.

Turning any of these ideas into a real, working system always requires going through the complete process this course has traced:

• Defining a precise signal.
• Backtesting rigorously, avoiding look-ahead bias, survivorship bias and overfitting.
• Evaluating the result with a set of metrics, and not with a single shiny figure.
• Managing risk with disciplined position sizing from day one.

Anyone who arrives at the technical track with this conceptual map already internalized starts with a real advantage over someone who begins coding without having asked any of these questions first.

It is, precisely, the difference between automating an intuition and automating a strategy.