Does automated trading work?

Automation is a powerful tool.

And robots are being used more and more in an age of social distancing, doing jobs like cleaning and disinfecting, taking temperatures and delivering food.

And, as someone who’s spent the last few years building an automated trading system, I believe they have their place in the markets.

But they have their limits. Think of the performance of certain algorithms we’ve seen in recent months to deliver exam grades, or modelling of Covid cases and election results … and spare a thought for the security robot in Washington that drowned itself in a shopping mall fountain a few years back.)

Yet we keep on putting faith in these systems.

Software failures have lost millions of pounds … had people falsely arrested … caused plane crashes … yet we still follow what the computer tells us to do.

 

‘Overtrust’ is a serious issue in robotics.

Scientists have worked hard to engage their systems with humans – and to build this trust. Even when we say we’re skeptical of an automated system – we still follow it. In fact, when robots fail, we continue to trust them.

In a fatal crash of a semi-automated Tesla, the driver (who should be actively driving) was playing a game on his iPhone.

And one of my favourite robot studies, had human subjects being led by a robot out of a smoke-filled building. Even when the robot (which was deliberately made to look as tin-can-like as possible) repeatedly led them the wrong way, into dark store rooms, they kept following it.

When questioned why they’d trusted the robot, people said, ‘I thought the robot knew something I didn’t know’ … ‘I thought it would be connected to the Internet, so would know everything.’

It sounds very similar to the things we traders say when a Forex robot has just lost us a load of money.

We assume – because a robot potentially could manage more data that we can – that it is handling all that data.

The shocking truth about a lot of Forex robots is that they are ridiculously simple.

Simple moving average crossovers, but with exciting-looking arrows flashing on our charts … Expert advisors that charge a lot of money really can be that basic.

With a Forex robot, we rarely know what’s going on under the hood. So, we give them powers they really shouldn’t be trusted with.

Does that mean that automated trading systems don’t have any use?

There’s a quote I like, usually attributed to the British statistician George Box,

All models are wrong, but some are useful.

It’s true of manual trading systems and automated ones.

And it all comes down to recognizing the weaknesses of a system, and not expecting it to do stuff that is just beyond its abilities.

Automated trading systems are brilliant at doing the leg work to save us time – they can watch for levels to be hit, they can tell us when an indicator reaches a certain level, they can open trades for us, close out parts of trades, move up stop levels.

These are great tools that can enable us to step away from our screens and get on with the fun stuff in life.

But they aren’t great at the nuance, and need a human hand on their shoulder to check they’re still working as they should as market conditions change.

It wasn’t until I began work on developing my own automated trading system that I uncovered just how flawed many of these systems are. Unfortunately for my developer, my pedantry and nit-picking means that it’s taken years to get to a point I’m happy with.

And even then, I insist that I always press the button on whether to take a trade or not.

I know some Traders Bulletin members have had bad experiences with robots emptying their trading accounts while they looked the other way – I never want my automated trading to have that kind of power.

I hope to be able to share this system with you later this year, so please watch out for updates over the coming weeks.

 

To enjoy more content and get it faster Follow @TradersBulletin