Not how it works

Goldman Sachs and UBS both seem to have had algorithmic trading code stolen and/or leaked recently. I haven’t seen an official statement from UBS, but a PR person from Goldman said something to the effect of, “Since the algorithms integrate with a large, proprietary system, we aren’t worried about this.” I realise that this person has the job of reassuring jittery investors that everything’s OK, but the statement shows a chronic lack of understanding that anyone should be able to see through. Sure, I probably couldn’t just compile the code, drop the binary into my system and start trading. But that’s not the point — there are two, very significant things I can achieve just by studying the code.

Firstly, I could determine an algorithm’s behaviour and reproduce it. If Goldman’s algorithms are as good as they claim, the actual implementation detail should guarded closely. If anyone could clone them, Goldman would lose their competitive advantage in algorithmic trading.

Secondly, I could possible study an algorithm’s behaviour and find a way to identify it and game it. That way, if I reasonably suspect that a Goldman algorithm is trading a security, I could use my knowledge of its behaviour to my advantage. If enough people with enough capital can game your algorithms, it can have a big impact on performance.

But it doesn’t really matter what I think of the announcement. I just build systems — I’m not the one deciding where to send the money.

This entry was posted on Friday, 31 July, 2009 at 8:53 pm and is filed under Development, Technology. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply