I hope you don't mind me chipping in but, you're changing your parameters way too often. That's very unlilely to be effective because all you're doing is chasing recent variations. What sample size you need is a whole topic on its own but fyi I trade about 1000 selections a day and still have losing days sometimes, 6 this year already. I wouldn't treat a sample of less than 500 markets as any sort of indication at all as to how you're doing. This is the problem with using trial and error in a live environment, it just takes too long. So if you're serious about automation you'd be better off using these early months to refine your process and methodology so you can rerun multiple variations on the same data. If you don't you'll just be running variation A for a month and then variation B on a different month, and it'll tell you nothng about the difference between A & B because they're running on different races.
People might advise differently but they're often thinking within the contraints of what BA allows you to do. If you're serious about this, think about what you need rather than what you have, and then go and do what's necessary to build it. At the very least learn how to use stats properly and you'll be able to see how statistically relevant your results are, and what sort of deviations would be considered normal for any given trend. Stepping away and doing a few weeks or months of proper study, ideally following a course, will pay dividends and is a trivial amount of time compared to how long you hope to be profitable for. TBH you can't automate without the basic skills to do it, and there aren't any serious automated traders who only use proprietary software or haven't got at least a decent grasp of data analysis and statistics. There's more to it than firing up Guardian and randomly changing things, the quant realm is brutal and margins are small, and even if you're lucky enough to find something by chance you'll probably not be able to repeat it as the market evolves. Also with enough work up front you should have a pretty good idea of how a strategy is going to perform when it goes live, and you'll realise the idea of paying to test and refine (by way of losing) is unnecessary, if you've made that initial investment in time and effort.
You seem really keen and motivated goat68, tbh I wouldn't have bothered commenting if you weren't, so give yourself a fighting chance and I'm sure in a year or two the money will start flowing and hopefully snowball.
Something general: Really try to break away from the mindset that you have an 'open' and a 'close'. You don't, you're just building an inventory of backs and lays on selections as the situation dictates and any sort of 'close' is merely tidying up any excess liability you don't want to carry through to in-play. In fact the whole idea of "opening" and "closing" and calling that a 'trade' is one of the most unhelpful and misleading things that new people get in their heads. Remember too that a major factor in the size of your profit is turnover, and a single "open" and "close" will only ever turnover a fraction of what a strategy that's making multiple bets either side can do. Ditto potentially trading all runners rather than just the favourites, and all sports in all countries, win and place. A market is a market is a market imo, they all behave in roughly the same way. So as I suggested above, get a process and a methodology in place, collect data and build an accurate simulation, then you can run 100s of strategies and variations on large datasets, and you might make some decent money and be able to sustain it.
