Thanks for your suggestions. Great stuff.
Agree that being able to buy backtest units separately would be useful. We kind of see the packages as offering either live connections (with backtesting chucked in), or backtesting (with live connections chucked in), but will look to see if we can reduce the cost for those only interested in one or the other.
We had set the minimum for a backtest to one unit as it takes a load of server resource for set up and tear down regardless of how short the backtest (though, yes, the refund mechanism is as generous a possible for failing tests, so not too hard to grab your unit back!). Have decided to scrap that for now, so will do a release soon that will let you run short backtests at no unit cost (will work out at approx. anything under 2 weeks at minute level). May also look into some form of 'quick run' feature directly from the edit screen.
Note that the resolution feature on the new backtest screen allows you to do a quick test on your script by, for example, set to "1 day" and set the date range to a single day (though it will still use a single unit at the moment).
Am going to revisit performance on the optimization screen (and backtest screen as well, actually). We went for the live updating approach for real time feedback as your backtests are running, though this is pretty demanding on the browser. May end up replacing this with just a progress bar with results on complete, or at least provide an option.
Modification of stop loss / take profit is in the pipeline (trailing stops also).
Will be advancing the editor for smarter color coding, auto complete / suggestive text and to more intelligently detect potential runtime errors in advance as well as just syntax.
Have added full screen editing to backlog (though fairly low priority for now).
Will look into your issue with riskPercentage. Sounds like a bug.
Very appreciative of your feedback - some really good suggestions there so thanks very much!