I can see the need for market makers in illiquid, thinly traded stocks.
But HFTs are by their nature trading in the most liquid stuff with super tight spreads. The only need for a market maker there is if someone has a huge order - and in taht situation isnt an HFT if it notices it just going to front run it to somewhere daft... Im not seeing how that would be beneficial over a human market maker?
I see the claim that HFTs “front run” orders all the time. I have no idea where this idea came from (well, maybe some idea). Further, I’ve never seen anyone explain exactly how they are supposed to be doing this - the nitty gritty technical details of it. It seems like it’s always just assumed to be true.
What makes some stocks thin and illiquid and others tight and liquid? The presence or absence market makers and pseudo-market makers. Take them out and everything would be illiquid. Add them back, and they become thick and liquid.
The fact is automated market makers are simply better at market making than humans. It’s not like the industry decided to arbitrarily switch to computerized market making one day. Over time, computer programs became more and more sophisticated to the point where they were simply able to out compete slower humans, and the humans were driven out of the business.
If any of you would bother to read the damn links, you would know that this has nothing to do with HFT. It has lots to do with automated trading, but there is no way that the market would work with present volumes without being computerized. And market makers predate computers.
The value of HFT is another discussion entirely.
really? Ever write a test harness? If you have, then you would know you don’t write it with all the bells and whistles and checks of production software, since if it fails the test fails and you fix it. No big deal.
They didn’t log each transaction for the same reason. You accumulate basic statistics to see how the test goes, without having to wade through hundreds of thousands of lines of reports.
You also don’t write code to check where your code is running - especially not test code.
I am probably overstating it when I call it a re-do but the article here shows they will probably only lose 2/3rds :rolleyes: of the $440M.
Well, a human could do it by for example seeing a huge order placed naively in the order book. Or perhaps they notice an iceberg that creeps up.
But the point is that an intelligent order worker could come up with more complicated strategies that wouldn’t be noticed by an average human market maker. This is where the HFT could come in with their data mining and so on. Let me make up an example and imagine they worked out that someone was buying 10000 shares every minute and furthermore they could come up with a statistical estimation of how many shares that person had bought in each minute. If they knew there was half a sceond and 5000 shares to go they can make easy free money by lifting 5000 shares and making the next offer. If the above numbers are unrepresentative then just scale them to more realistic examples! I don’t know anything at all about finance, I’m just assuming things from my own common sense.
The irony of course is that because of this fund managers are having to use their own algos to work orders in more sophisticated ways. So it’s an arms race which benefits no one at the end of the day.
No, the $440 million number is after accounting for the broken trades. “Clearly erroneus executions” happen and are broken nearly every day. Knight isn’t getting any special treatment here. You can read the NYSE’s and NASDAQ’s CEE trade break guidelines on their websites.