While the CS ticket itself didn't profit (due to the specific combination of legs hitting a 30% strike rate on exact scores suggests the model's underlying algorithms for calculating team strength are very calibrated,)
You should really give betting singles a chance to build a profitable model. With combos, everything becomes way more a game of chance, and it's nearly impossible to get a bet sample that is big enough to remove all the variance.
I get it though, low stakes risk and potential high reward is far more attractive than higher stakes with no chance to hit a big jackpot but more boring grinding it out.
At the same time, you are up at the moment so well done so far!
I know I should bet on the singles. But I will give the model some more time to prove itself. And when I do bet on the singles, I should rule out all bet below probably odds of 1.6 and so.
During the short winterbreak in the Premier League, I will run some queries on the predictions and see where I do have an edge.
You know, I do love the accas. And yes, I do know they are a bookies dream. But I just don't want to be tracking too much. For me it is more about the prediction module then the betting on itself.