Here's what I'd like to see available: A prop firm that has space at Genesis or Lime (could be others but these 2 seem more practical, please PM me if you know of comparable shops) offers to startup HFT projects a 6-month deal: Trader/Programmer pays $3000 upfront + $5000 risk deposit, 85/15 split per month and receives: - A 1U space for a server - High quality UDP data feed - 30:1 leverage - $0.50/1000+passthrough commission - Acces to many routes - S7 registration (+ fees if new registration) If the startup makes money, the firm can do well with the %15 of profits. If they don't make money, the $3000 fee covers data/hosting/hassle costs. Think something like this would work?
Here are a couple of HFT incubators: http://www.thesystech.com/ http://www.txcus.com/ Here's what I think the main drawback of these shops is: 1) You need significant scale to make it worth their while 2) You often need to give up a significant amount (in fees or equity stake) - mainly because they use expensive providers 3) They want a track record anyway - so there is a chicken-or-the-egg issue. What I think may make sense is a kind of technology cooperative. 10 guys who want to do HFT could band together to share technology resources. It's an organizational headache but could someone with the org skills make something like that work?
The comms seem steep for HFT type trading. I can't imagine too many HFT edges that won't be eaten away by paying .0005
Well you may be right. I tried to outline something that could be within the realm of possibility - where a firm gets their nut and might be willing to incur the overhead. The last time I traded with a prop firm I paid $4/1000 w/ 100% payout - mostly because I wasn't doing high volume. I tried to use that structure to do some automated trading, but spent most of 3 months debugging my code and doing some live tests. In the end, the strategy didn't work out. I have more ideas but it's hard to find a structure that would make sense for HF bootstrapping. 50 mils may be high for true HF but I'm looking for a structure where I can develop/test and have 6 months to get some volume going.
is .20/1000sh enough for HFT? i don't imagine fees being any lower, I think the lowest i have heard is .18/1000sh.
I think you always have to be acutely aware of where you are in the food chain. One way to look at it is that the Getcos of the world have made it so that nobody who isn't using FPGAs at the exchange is incapable of making any money. The instant some quirk in the market is present, super geniuses have locked it up. Another way to look at it is that some quirks may just not be worth the time and expense of exploring and implementing. If you have a company with 200-some employees that makes $400M+ per year, you aren't going to have an employee slave for 6months on a specialized algo that won't make much (relative) money.
you made one point, I didn't say that one should try a startup with that type. But indeed, there is a list of electronic market markers out there which are in the same level of Getco, like knight cap?