A Question For HFT Insiders

Discussion in 'Automated Trading' started by clearinghouse, Dec 10, 2010.

  1. This is more of a rough poll, but how much money would it take to create a successful equity market making operation from scratch?

    $30mil, $50mil, etc? I'm talking salaries, equipment, etc.

    And do you need the fast connection from Chicago to NY, or can you stand-alone make markets and still make 5-figures or more a day?

    And finally, are the players at the top currently more or less going to be the same ones there 10 years later? Has the game become that tight?
  2. I know someone who knows someone who works at an HFT firm. They are located right next to the exchange. Speed is a must. You might be able to co-locate, but the server making the trades needs the fastest access (shortest distance to the exchange servers) possible, no question about it.
  3. You can make markets from a netbook on a tropical island.

    How competitive do you want to be and how many markets/securities do you want to trade? You need BIG capital to be a MM, not in terms of salaries, more in terms of being on the wrong side when the market goes against you. If you want to start small you can make markets for under $25k at a good prop firm.
  4. This is not what I meant to ask. What I wanted to know is essentially what it takes financially to compete against the big boys in equity MM, not what it takes for some guy to throw out a few passive orders a day and make a living.
  5. Why not think about the problem differently. Instead of asking about competing with them directly in terms of $ and time, why not think - What innovative trading strategy or trading technology would it take to compete against the big players in HFT?

    The only way to ever compete against something big like that is to develop some disruptive trading technique .... It would be much faster, cost very little and erode their business model immediately.

    What is it? I don't know.
  6. $10mil is enough to start. Maybe another 3 mil for hiring and infrastructure.

    You can churn 100 lots on 100 different stocks with no leverage, thousands of times. Thats what pros do anyways.
  7. Sure, but as a hypothetical, what's the dollar value needed for entry? I'm just interested in the barriers to entry.
  8. Sure, but you aren't going to run SPY vs ES without an expensive connection between NY and Chicago. At some point along the way, the speed assists you in snatching up fleeting inefficiencies in the equity market. The less speed you have, the more quant-magic you need. The ideal situation is to have speed and quant-magic, and I imagine most of the big boys have a nice blend of both, and that those things cost some fairly sizable amount of money.

    100 lots on 100 different stocks sounds good, but just how much growth can you pick up running intraday stuff if you aren't running towards the bleeding edge?

    FWIW, I'm just theorizing and digging for opinions.
  9. rosy2


    you dont need anything near 10mil
  10. A good dedicated OC (Optical Circuit) is about $1k/month, maybe $1,200. A locker/rack in a decent datacenter close to crossing engines at NYSE/ARCA/NASDAQ will run you ~$15k/month. Fill that with $100k worth of hardware, hire a few $100k/year programmers and you'll need about $10M in capital. I have no idea how much exchange seats/leases are or Broker Dealer licenses, etc.

    Hardware related startup costs of about $100-$150k with monthly costs of about $20k, $10M capital.

    Thats it... Plus exchange fees and memberships and data feeds.
    #10     Dec 11, 2010