Any pointers about backtesting high frequency strategies?

Discussion in 'Strategy Development' started by mizhael, May 26, 2010.

  1. Bob. Dumb question. Given the accuracy of your strategy and the poor fills from limit orders, why do you not use market orders? My stupidity is that it sounds to me like the lost profits with limit orders should exceed the losses with market order spread and slippage. Thanks.
    #11     May 27, 2010
  2. promagma


    Stacking 100-share orders is what worked for one of my strategies, and also allowed for an accurate simulation.
    #12     May 27, 2010
  3. Bob111


    in this particular case using market orders is suicidal. just like with fill ratio calculation I've done some studies on spreads,sizes and so way i'm using market order. "flash crash" is a good example of how deep "they" can screw you with your market order. stacking 100 shares orders may work, but commissions will be "substantial". they(commissions) already eat 30% of my profits.
    #13     May 27, 2010
  4. HFT to you is like clicking the mouse 3X a minute
    #14     May 27, 2010
  5. Thanks for explaining, Bob. Trade well!
    #15     May 27, 2010
  6. byteme


    No, that doesn't explain why only the winners are partially filled and the losers are fully filled according to promagma.

    His statement implies that the market knows at the time of his placing the order whether it was going to be a winner or loser. If that's true it would be an exploitable edge.
    #16     May 28, 2010
  7. I think you run into a nightmare pretty quick trying to add liquidity no matter what the matching engine just from a data management stand point even if not conceptually. IMO a one man shop, not close to the exchange on a standard internet connection is dancing with a devil in a dance that is better to sit out.
    Optimal Trading Strategies by Kissell and buying up all the microstructure literature can probably get you down the right path but in the end it will just make you realize its a path not worth taking.
    #17     May 28, 2010
  8. Bob111


    how it can be exploitable edge, if you can't get into a trade?
    i have number of systems in my drawer that are useless only for this particular reason. i can't get into a winning trade and losers are off course all systems for short only. traded for year. real results-zero. on paper average PnL =3% per trade. backtested on thousands of trades.for some magic reason stock that about to fall hard almost always not on short list.
    "they" will do everything they can to prohibit you from getting into. it's not a "market", it's market's other participants. sub penny on front of you,hide your order-they will trade trough,even if your orders are perfect match, they will post bigger size and will get better priority or route their orders to another ECN. the list of those tricks can be pretty long and all of them are not in your favor. and there if you try to dispute the "rules"-there is always excuse,exemption or whatever reason..i see this bs every day from 8am to 8pm 10 years in row.
    #18     May 28, 2010
  9. byteme


    So you agree with promagma: at the time of placing your order "they" know whether it's going to be a winner or a loser. "They" know whether your trade is to open a position or to close a position. "They" know how long your are planning to hold the position for. "They" are monitoring the positions of all of the people trading at any given time.

    Like I said, if that's true, it's an exploitable edge: If "they" only give you full fills on your losers then you know it is going to be a loser as soon as the order is filled! You can exit appropriately. Conversely, if "they" only give you a partial fill you know it will be a winner and you can let it run or add more.
    #19     May 28, 2010
  10. Bob111


    well..there is obviously other players,with their own systems and it's shouldn't be surprise for anyone if your systems are similar. not to mention the fact that market(and your orders) are constantly exploited for any edge by other players. what i'm trying to says that there always somebody else, who believe that the odds of winning here,at this particular price are greater, than 50/50. the one probably can find some edge by analyzing certain things, but for retailer it would be really hard. i do have few observations about certain price\size behaviors that can be applied to the market as a whole,but right now i'm too lazy to do anything with it. maybe later,when i'm hungry(long loosing strike or few slow months) :p
    #20     May 28, 2010