Now all that is needed is someone to write an easy to use back-testing platform to use this data. It is surprising that there is nothing that back-tests against bid/ask and last prices to give more accurate results. With quote size information you could even make a good guess at partial fills.
squeeze, You post raises an interesting question. That is: could bid/ask data with size info be used for backtesting? I would worry about the games that market makers and others play to "manipulate" the market. Much of what you see in bid/ask is more of what someone wants you to see, rather than a snapshot of reality. Norm
TS2000i has all the data field build-in already for data,time,trade type, best ask/bid, size ask/bid, volume, .....available With TS8.1 they are not build-in and you have to assign price fields to some values through EL coding. Norm, you are missing a point here, price fields for best bid/ask is the top tier price in level II window, not second, third,...tiers Top tier price are the prices that executable, market makers can manipulate lower tiers only. HH:MM:SS where time gap between SS and size can give you better picture about probability for price fill at certain levels. For Scalping you need something like X-trader pro with "Autotrader" for $550/month (extra $250 a simultaneous market) and very fast internet and low commission through exchange membership, it's not everyone cup of tea
I have actually written an application that uses all this information to back-test and it produces very accurate results. Using all this information it is possible to get a good idea of slippage on stops and partial fills on limit orders as liquidity is taken into account. It generally produces results that are much poorer looking that Tradestation but a lot more realistic. It's draw back is that is is very difficult to use and spending all the time thinking about coding with it just gets in the way of trading. Writing a generic and easy to use application to do the same thing is a big job and something that I am not willing to take on at the moment. Perhaps an opportunity for a talented programmer with some time on their hands. Tradestation was designed as a bar based back tester and just doesn't have the right structure or the performance for running sims on vast ammounts of time-sales data. Tick based back-testing needs an application designed specifically for the job.