Hacker News new | ask | show | jobs
by ig1 5595 days ago
More likely is things like front running, where you use buy/sell something upfront when you know someone else is about to move the market price

Layering is a completely different practice which involves creating an impression of volume when there isn't one (which is strictly regulated in the equities market, less so in others).

It's not uncommon for firms to try and exploit flaws in each others trading strategies, for example a rooky mistake in pricing algorithms is to use the last traded price as the basis of your price, so a common strategy in infrequently traded items is causing a price spike causing the rooky algorithm to go way off market allowing them to be taken to the cleaners.

I've seen firms manipulate prices in weird monotonic ways in order to see how other firms algorithms work in order to reverse engineer them.

I've done similar things myself in order to figure out the internals of exchanges, in one case I was even able to identify an exchange had incorrectly configured it's kernel tcp settings causing an irregular latency spike in their internal systems.

Trading algorithms to some extent have to be designed to protect against vulnerabilities, not because of code theft, but because people are getting very good at doing remote reverse engineering based upon observed inputs/outputs.

1 comments

Front running would be tricky - even if you have Goldman's code, you'll still need to build an infrastructure faster than theirs. Good luck with that.

In contrast, if you can figure out how to implement layering in such a way that their system acts on it, or use price spikes, or something of that nature, you don't need to beat them on speed. You manipulate them, they trade against you as fast as they want to, and you win.

Only if you're talking HFT, most quant trading isn't HFT.
That is what JSig was discussing, no?

"...somebody who knew how to use this program could use it to manipulate markets in unfair ways," said Assistant U.S. Attorney Joseph Facciponti [in regards to the code taken by Sergey Aleynikov]

(Aleynikov is the guy who allegedly stole code from Goldman.)

Yes the GS code was HFT, but the article is discussing the MS case where there's no indication the code was HFT, the article mistakenly assumes all quant trading code is HFT.