That need to be necessarily done with orders created realtime (robotically). Cannot be realistically done with prespecified lmt orders, which woud be skipped most of the time, resulting in a massacre (not to talk about mkt orders)... We are talking of scalps a little bigger than spread here I believe (which would not, anyway, yield anything good, i think, other than making happy the broker...). Tom
Why would it be skipped? Setting the MIT order one tick outside of your target would require price to go through and the level II data is used to validate sufficient size to fill for larger sizes. Were talking back testing... If your trading small lots you can bypass Level II validation and just record a fill at your target once price goes through. Obviously small lots would be more realistic to real life and shouldn't move the market. Fairly common for HF traders to use MIT orders or LIT orders in lieu of having orders resting in the queue.
Thank you everyone for your responses. Just to clarify my testing is based on 5 minute charts and targets between 4-10 tics for profit. I am using Ninjatrader to test the strategy and use the default fill (which requires price to trade through on both sides to cover the spread) along with slippage set to 1tic. I also bind the order itself to the range 1 chart to provide me with a closer view of intrabar price action. This is being tested on several futures contracts and provides profitable results if I set the trade to base its fills on 1min, 5min or range 1. Thanks again. You have all helped to put some of this into perspective.