You ask this, as the sole content of your post, as if you imagine it's in appostion to the title you gave the thread ("Can algorithmic trading systems beat human traders?"), Lloyd, but they're two very different questions. Anyway, the answer to each of your two questions is "yes", in my opinion.
I know of one situation in which the algos that were trading the markets went totally crazy with all sorts of weird data that were confusing the systems. It takes a human being to realize that the reason everything was going crazy was that an earthquake had just hit Japan and the markets were reacting.
They were probably badly designed algos... (I saw exactly the same thing at the same time). Does that mean all algos are bad? Every day human traders do stupid things. Does that mean all human traders are bad? GAT
i developed my own algo system for trading ES, NQ, IWM, QQQ, NG, CL and ETFs. they easily beat human traders hands down. while some are stopped out the majority hit targets and sometimes more. eg >>> ES had a sell showing at 2097 half. stop was 2109 and target was 2007. target hit and then some. also had 2 sell signals invalidate during october. the stops on both of these got hit with a 14 point loss on both. ive been trading these signals for over 2 years and now place trades everytime they show up on es. i can say without a shadow of a doubt they destroy the returns i had previous to developing the system
Can algorithmic trading systems beat human traders? Some can and some can not. Thus, depends on the algorithm system and depends on the trader and depends on the trading instrument and depends on the trading style. Another general question thread that can not get a Yes or No answer.
That's an algorithm that doesn't like volatile price action. There's a lot of different types of algorithms out there and they compete against each other due to those differences.
I've just written a blog on this topic http://qoppac.blogspot.co.uk/2016/01/computers-vs-humans-considering-median.html (ok it's a cold thread, but it's not like the question ever went away...) GAT
In many cases, you add a layer of coding to detect these situations. However this does increase the complexity of the algo and a different way to handle this is just be sure to monitor your systems in realtime and shut them down when these situations arise with a "close all positions" button.
“If you are able to extract rules, and your understanding of how you create rules…how you adapt those systems, you could potentially build AIs that are capable of also creating other trading systems,” said Netherwood. “(These) will be even more efficient at extracting the inefficiencies of the markets, and will be able to do it at such speeds that humans are not able to compete.” http://www.financemagnates.com/inst...er-threat-from-market-technology-singularity/