pointers to readings on automatic market making?

Discussion in 'Order Execution' started by mizhael, Jul 11, 2010.

  1. You can start a solo operation now, but it is MUCH harder than it has ever been.

    True HFT is almost totally out of reach. The top tier (including Chi<>NJ networking) going to be $500k/mo in infrastructure with a term commitment. Then you have to write the logic, get the correct access in place, put $5m up at a prime to get portfolio margin, etc.

    Look at the results of the successful, active players of late: 2Q10 was down 50-90% YoY pretty much everywhere. Sometimes the markets are dead, and there isn't anything you about it--but your costs are fixed.

    What is still completely possible (and in my opinion is by far the best approach at this point) is getting yourself some good 1min bar data, dumping it into Matlab, building yourself a GUI, and wiring it into IB. Your costs will be about $500/mo and you can definitely play in the 5-15 min space. Not a lot of low-hanging fruit, though.
     
    #11     Jul 27, 2010
  2. To clarify, by "true HFT" I mean AMM/arbitrage.
     
    #12     Jul 27, 2010
  3. i'd have to agree with rufus. i'm a one man shop and have come up and grown nicely in the amm/arb space and i can tell you i'm definitely not spending 500k/mo in overhead, and not trading 5min bars at IB. those extremes are a bit hyperbolic.

    strong comp sci skills are a requirement though obviously. throw in a little stat and basic business knowledge, it shouldn't take a good above average intelligence hard worker 6 months to a year to get at least _some_ decent 7-11 money going.
     
    #13     Jul 28, 2010
  4. We can't be comparing apples to apples.

    When I refer to AMM/HFT I'm talking about the ability to trade equities. options, FX, and futures, and at rates and speeds where buying on the bid and selling on the offer has meaning in the sense that it is STILL the bid or offer when you've gotten your hedge off elsewhere (or something along those lines).

    If you're talking something like equities-only, I agree... you can do some real damage with your brain, $100k, and an account at Lime.
     
    #14     Jul 28, 2010
  5. Yo gotta hand it to the OP. His naive questions occasionally sparks off some interesting discussions - like this one.

    My personal experience is that if you can find a niche and evolve from that, then HFT is not that unreachable.
     
    #15     Jul 28, 2010
  6. dloyer

    dloyer

    brownegg: Why lime rather than IB?
     
    #16     Jul 28, 2010
  7. rufus, do you know what kind of volume they were moving getting those numbers?

    my margins have definitely taken similar hits, but have been able to scale well to maintain some small growth in profitability. i can definitely attest to it being quite crowded now.
     
    #17     Jul 28, 2010
  8. IB throttles data delivery and all orders go through a risk check in Greenwich. They are unbeatable for anything that isn't extremely time sensitive, but you can't do (what I keep calling) HFT. For many people, and probably rightly so (it's a matter of perspective), HFT would be 1-minute holding periods, which you could certainly do (and I would) via IB.

    Lime is as fast as agency delivery can get, from what I can tell and have heard, although they do have some network issues during spikes (again, I've heard that--I do not currently execute via them and haven't seen it personally).
     
    #18     Jul 28, 2010
  9. dloyer

    dloyer

    Thanks for the reply, brown.

    I will take another look at lime.
     
    #19     Jul 28, 2010