Automated trading
A program stops making mistakes out of emotion, but makes technical ones instead. It helps you stay disciplined, but doesn't solve the problem that you first need a strategy that works.
Learning objective: After this lesson you can judge which mistakes automated trading really solves and which it only shifts.
A program sticks to rules, even at night and even after three losses in a row. That's its real advantage over a person, and it's not a small one.
Limits of Automation
What it can't do: make a bad strategy good. A program only carries out a negative expected value more reliably. So the question of the strategy still comes before everything else.
New Sources of Error
New sources of error come with it. The connection drops. A typo turns 10 shares into 100. The broker's interface changes overnight. The program crashes and buys everything again after restarting. Mistakes like that happen fast and on a large scale.
Caution With Purchased Bots
And a warning about bought bots: if someone sells you a program that supposedly makes money reliably, ask yourself why they're selling it instead of using it themselves. Nobody has ever given a convincing answer to that.
The benefit of a program is that the strategy is carried out exactly as you wrote it down. A large part of private investors' return gap comes from precisely such deviations: skipped purchases, winners sold too early, stops that got moved. So this benefit is real, regardless of how good the strategy itself is.
Guarding Against Technical Risks
The technical risks need their own safeguards. Set in the system how big an order may be at most and how many there may be per day. Regularly check whether the actual positions match those the program expects. Decide what happens if the connection drops. And build in a kill switch that lets you stop everything immediately. Without these safeguards, a single error can cost more than the strategy earned in many months.
Sellers' Incentives
With bought systems it's worth looking at the incentives. If a provider really had a strategy that makes money over time, they'd rather manage other people's money with it than sell software. The profits of a working strategy grow with the capital used. Software sales only grow with the number of customers. So the fact that they sell tells you where they actually make their money.
Summary
- A program solves the problem of discipline, not of strategy.
- Technical errors happen fast and on a large scale.
- Anyone with a working strategy wouldn't sell it.
Did you get it?
What's the real benefit of a program?
That the strategy is carried out exactly as it's written down.
What safeguards does an automated system need?
Limits on order size and number of orders, a position check, defined behavior if the connection drops and a kill switch.
Why is selling a trading system a warning sign?
Because the profits of a working strategy grow with capital, while software sales only grow with the number of customers.
Check your understanding
Your task
Imagine a program trading a $5,000 account. Write down a maximum order size in dollars and a maximum number of orders per day. Also add one sentence on what the program should do if the connection drops.
Sources and further reading
- Morningstar's annual "Mind the Gap" study shows that the dollar-weighted returns investors actually earn regularly lag reported fund returns, because purchases and sales are poorly timed. View source ↗
Related
- Gurus and signal groupsFraud Protection
- The classic beginner mistakesStage 1
- Spread, slippage, and liquidityStage 2
- Matching questions for this stageQuestions