improve my fills using ninja trader?

Discussion in 'Order Execution' started by youngstownsteel, Jan 26, 2013.

  1. i am a trader at a small algo shop. i am not a programmer, but supply the trading logic to the programmers. most of the code is written in c++ and traded directly via apis. most of what we do is with equities. most of the trading is very scalp-type trading.

    a few months ago i started using ninja trader to backtest some ideas using futures. we came up with some ideas we like, and have started trading them in a small account. Our problem is probably not Ninja specific, but we are using Ninja for this, so am posting it here. Our model is a mean-reverting quick model that attempts to buy on the bid and sell on the offer. not hft, we get maybe 1-5 trades per day per instrument traded. all of our trades hit out perfectly compared to our backtest,except for trades where we get 'left at the altar'- we are bidding or offering and we dont get filled. our $ per trade are not great enough where we can change to a market order.

    recently we added a 'front run' functionality, so we can enter an order for the next bar while the current bar is still forming, as we will know, using a what -if scenario if we will have an order at the end of this bar. we were hoping that by getting to the front of the que earlier we would have more success. we also have a 'chase' function that will cancel our original order and go to the market if the market moves x ticks away from our entry. but this is an expensive alternative.

    of course, setting ninja to liberal for backtesting, if something trades on the offer, ninja will take a fill. using default will not. our results using liberal are very appealing, on default not so. but even using our front run functionality, we still miss enough fills that it makes trading the model much less tradeable. the impact is much less in stocks where there is much more granularity. relatively speaking, misses in bonds are very costly.

    aside from our 'front run ' functionality, are there any other ideas i should be trying? of course, many things test out great, but can't be realized in real life. this may be one of them. any and all thoughts are appreciated.

  2. dom993


    From the description of it, IMHO there isn't much hope.

    You would need to be colocated at the exchange, analyze the order book at ultra-fast speed, to get a chance at this game.
  3. Yes. Definitely colocate. Not that expensive - per Exchange.
  4. thank you for the replies. i dont think colocating would do a lot, as we are already placing our orders ahead of time using a what if logic while the current bar is open. and of course, there may only be a few ticking on the bid or offer as well.