What automation does not fix

12 minutesLesson 3 of 9
Femi, the HustlCraft strategy mascot

This is the sentence I want you to carry out of the whole module. A program cannot give a losing plan an edge. It can only carry that plan out faster, more often, and without one moment of doubt to slow it down.

The one thing it cannot touch

An edge is a set of rules that makes money on average after costs, over a large enough number of trades to mean anything. That definition, from what an edge is, does not mention who presses the button. Automation changes who executes and how consistently. It does not change whether the rules were worth executing.

So the question a program answers is never "do these rules work". It is "what happens when these rules are followed exactly", and if the honest answer to the first is no, the second arrives very quickly.

A program has no opinion about whether your rules work. It has never seen your account, it does not know what the money is for, and it cannot be talked out of anything. It executes.

The arithmetic of losing faster

Negative expectancy plus more trades is not a smaller problem. It is the same problem delivered sooner.

One losing rule, two speeds

  1. Say a rule loses 1,200 naira per trade on average once spread and commission are counted. That is not a disaster on any one trade, which is exactly why nobody notices it.
  2. By hand you manage four trades a week, so roughly sixteen a month. That is about 19,200 naira a month.
  3. Automated across five symbols it finds twenty trades a week, so eighty a month. That is about 96,000 naira a month.
  4. Five months of your own trading now happens in one.

Nothing about the rule changed. The clock changed. Automation is a multiplier on whatever expectancy you already had, and a multiplier works just as well on a negative number.

Every one of those extra trades also pays the same three costs, and they are rarely as good as the trades a person would have picked: the setups you would have skipped get taken too, and fast entries often fill worse. The costs arrive on every trade; the edge only arrives on the good ones. That is how a rule with a thin edge by hand can become a losing rule once a program takes four times as many trades.

Trading carries risk and you can lose money.

The list of things people hope it will fix

What people hope automation fixesWhat actually happens
A strategy that was never counted honestlyIt gets counted, quickly, at full speed and full size.
An account too small for the risk being takenThe same risk is taken, with no hesitation left to slow it down.
A broker with wide spreads and slow fillsEvery extra trade pays that spread again.
Not really knowing why a trade was takenNow a machine takes trades you cannot explain, which is worse, not better.
Discipline in generalIt enforces the rules it was given, including the bad ones, with no exceptions.

Read that right hand column again. Not one line is a malfunction. Every line is the program working exactly as designed.

Femi's warning on obedience

Femi, the HustlCraft strategy mascot
Femi's warning

Give the obedient driver a good route and he drives it perfectly, every single day, without getting tired. Give him a bad route and he drives off the same cliff every single day, perfectly, without getting tired. The obedience is the feature. It is also the danger, and nothing in the settings will ever tell you which route you handed him.

There is only one defence, and it is not a better program. It is knowing, before a single live order, whether the rules have an edge at all. That is what the testing sequence in lesson six is for, and what the over-fitting trap in lesson seven quietly destroys.

The picture to keep

The obedience cuts both ways: a good route driven perfectly every day, or a bad route driven off the same cliff every day. The driver cannot tell the two apart, and neither can the software.

PreviousNext: The honest pros