Talking about the quality of the data in testing, my experience so far shows it matters little in forward testing. I am running 5 EA’s on 4 Demo account for a three months testing period. All five have been tested on good data and I was happy to move forward to the ademo testing stage. Now three are performing well but two are just not performing as I hoped. The main reason is the change in market types from the testing stage to the demo stage. My point is even though those two tested with good profits, there are flaws in both EA’s which testing can never show. I believe we shouldn’t get too stressed on testing quality over real world trading conditions of changing market conditions. My idea last year was when I started building EA’s with tradeview is to build over optimistation EA’s because I feel the results shouldn’t matter too much if we retest and alter a Algo every three months. I now am starting to see this as a bad idea mainly because the market can change more quicker than I can confirm the market type and therefore change my settings. I feel I am chasing the market instead of being on top of the market. So if I had better data, it wouldn’t of changed the outcome. I am now looking at building even more robust models with less external functions. I tend to be a slow learner.
