Is it feasible to design an algorithm to beat VWAP in 2 min?

Discussion in 'Order Execution' started by mizhael, Jun 30, 2010.

  1. Given any 2 minutes time window,

    lets say open, lets say close, or noon, etc.,

    is it easy to design an algorithm to beat the VWAP in those 2 minutes?

    Any pointers?

  2. you are increasenow and traderzhangshan. ET is fake no one here is a real person it's all robots AHHHHHH
  3. If you want to buy and get better than VWAP in two minutes, here's what you do:

    - Buy some shares at the beginning
    - If it goes up, congratulations, you have beaten VWAP for the period
    - if it goes down, buy some more (repeat this step)

  4. Seriously, think about what you're asking for. If I had such an algorithm, I could set up two boxes - one tasked to sell N contracts above VWAP and one to buy N contracts below VWAP. Net result, zero contracts and at least one pip per contract. There are a lot of 2 minute periods in a trading day.

    Were I to have such an algorithm, you can bet I'd be on a beach in Tahiti with two of the best hookers surgery can build by Friday. So there's two possibilities: one is that someone will get bored of their hookers and post from Tahiti to help you. The other possibility is that you asked a dumb question.
  5. i had the code 4 such a system but the cat ate it.
  6. nbates


    "The Big D" and I think a lot alike :)
  7. Not sure if this is what you are after but you can consistently win if there is an uptick or down tick inside of this 2 minute range.

    If there is not an uptick or down tick and you picked the wrong direction you could lose big.

    just average down and exit on the next counter tick.

    1,2,4,8,16 up to your limits.

    The odds of 10 ticks in a row without a counter tick are about 1 in 1000.

    obviously set your tick increments to cover your trading costs. ie. add every 2 ticks and exit on 2 ticks.
  8. Okay, then what about the current practice of algo trading:

    people use VWAP as a benchmark to measure algo's performance.


    I guess we should modify the term "beat" VWAP to be "on average", or "statistically"?