A Weekend Tweak
If you have read the previous article in the series, you will be aware that I turned a profit in week one. But before I start celebrating, it’s time to see whether ‘StatBot’ was just the welcome recipient of beginners luck or if I’ve actually built something that works.
To be honest the latter of the two options is not necessarily a conclusion I can jump to this early, what I’ll actually need to do is test this process for somewhere in the region of six weeks and from there I’ll get a good idea of just what exactly needs tweaking.
However, one quite important tweak has been made and it is one that might suggest that it was a sense of luck that enabled us to turn a profit. I’ve mentioned the form index before and how the combinations of six-game results were ranked from 1 to 729 and this is the bedrock of the model.
The overall process hasn’t changed, but before it was ranking the last of the six games first and then in the direction of the most recent. Whereas the index should actually be looking at the most recent game first and then working backwards from there.
LET ME EXPLAIN FURTHER
I hope I haven’t lost you, so let me explain in actual match details. Let’s say CSKA Moscow have this six game run of form (from most to least recent)
WWLWDD
Looking at this, it would suggest that they are a club going into their next fixture with relatively decent form and especially when you consider that they would have won their previous two outings against two clubs of your choosing.
However, in last week’s setup the form index was looking at the successive draws as the first elements to rank and therefore the guide was actually back to front in a sense. That said, somehow it didn’t seem to matter all that much last weekend (especially as we turned a profit)
But this weekend, it has kicked up some outcomes that even with gut instinct alone looked ‘wrong’ and that is what led me to deep dive into what exactly was the problem.
Now this has been identified, it has provided me with this set of picks for this weekend:
Now the only thing that sits a little uneasy with me, is the fact that ‘StatBot’ has decided that there are to be no away wins. Something that could be possible in a eight fixture setup, but also something that would rarely happen on a weekly basis.
While also we should not forget that this is not splitting the form between six home games that CSKA Moscow have played and six away games that FC Rostov have been played – it is just any of their last six, something that may also have to be tweaked in the coming weeks.
Although, if this is what the model is telling me for this weekend, then who am I to argue with ‘StatBot’. Ultimately test and learn is the only way we are going to be able to refine this and hopefully we can once again turn a profit from our 8 unit outlay. Next week, we’ll find out if that is the case.
Happy punting and thanks for reading, Dan.
If this has grabbed your interest and you would like to discuss/feedback then please feel free to drop me a message at dan@realfootballman.com. While I am always looking for new football/data projects to work on and if you feel that my skills would be of use, I can be contacted at the same address.