I hit the wall too many times with backtesting. A few weeks ago I gave it up for live trading. The thing about not knowing what is going on inside a bar, which extreme came first, has tripped me up too often. The software people guide you in the direction of backtesting on bars of data. That way they don't have to supply much tick data and you can still test on stuff from previous decades and their software is that much simpler. I'll never run another backtest with any unnecessary uncertainty. I found that I can build the bars in arrays during the backtest from tick data and backtest on those psudo bars and never have the problem of guessing which extreme came first. A spin off from that work is the ability to have bar interval as a variable during a test. A way to avoid tick data and lengthy tests is to only design systems with time stops, ie, exit on your target during a bar or at the close of the bar. With those systems there is one less thing to worry about, the tests run much, much faster, and stops based on time have proven to be effective by some. System development is going better for me now just by trading during the day, noting ideas, and looking at charts after the market. Developing systems via backtest is not very productive, fine tuning them with backtesting, with no uncertainties, is probably very productive, I don't know for sure yet because I got so tired of backtesting I'm still on open-ended hiatus from it.
What framework are you using if you don't mind me asking? Does it allow you to use the same exact data for real trading as for backtesting? In other words, can you save the exact tick data you used for trading and "replay" the session repeatedly to tweak your strategy?