Back In Black
Back In Black
If you read my previous article on trying to build a predictive model for the Russian Premier League, you will have been aware that it was the setup for a weekend where changes were to be made to ‘StatBot’
Changes that we would not know if they were successful until the other end of the weekend and now we are that at point, it is time to see just how effective the fundamental changes were – which to summarise really quickly were:
Adding home bias to the predictive model
Changing the form index and the way that the six-match combinations were changed.
Ultimately I won’t go into too much detail, as they have been covered in the link above. However, what I will do is remind you of the picks that ‘StatBot offered up last Friday:
As you can see, there were no away wins that were predicted last weekend, which in itself was always going to be something of an issue. Especially when three out of the four games on the Sunday (the bottom quartet) ended in that exact fashion.
Although in the previous article, I did also mention the fact that the two Moscow derbies had the potential to really mess things up and as they say when a derby fixture comes around, the form book really goes out the window.
Something that when you consider how poor both Dinamo and Spartak Moscow’s form going into their matches had been, suggets that they are two shock results rather than poor modelling or logic beforehand.
Therefore, if the games that were expected to go my way didn’t then what did and thankfully it was a couple of longer shots that not only got me out of the ditch but also returned back into a profit (now 3 weeks out of 4 have turned a profit)
If we look at where it did go right:
Rubin Kazan vs Ural – Draw (23/10)
FC Rostov vs PFC Sochi – FC Rostov Win (4/7)
FC Krasnador vs FC Orenburg – Draw (16/5)
We can see that of the three correct picks, only one was the same way that the bookmakers suggested and it was once again correctly backing the draws, that was the difference between profit and loss.
This meant that of the 8 unit outlay (£4), I returned a £4.54 profit for the week and that also means that overall I have recorded a profit of 87p – which works out as a ROI of 5.4% after four weeks of operation.
So although we’ve got back into profit, it does raise another couple of questions:
HOW CAN WE BUILD AGAINST SHOCKS?
To be honest, this is going to be the almost impossible question to answer because ultimately we don’t want ‘StatBot’ to predict bad results for the sake of it, especially as that would turn out to be counter-intuitive.
However, we do know that shock results happen from time to time and it was clearly evident last weekend, so is this something that needs to be absorbed from week to week and if so how?
DO WE JUST BET MORE ON DRAWS?
Is there perhaps a logic of just betting on where ‘StatBot’ says differently to what the likes of Betfair are saying each week. Because if we were to find the differences and spread are units accordingly, it has the potential to sniff out even more value.
If I bet on all eight matches each week and the favourites come in, it is not going to bring back a huge amount of profit (although that said, profit is profit) and I’m going to have work a little harder to constantly avoid a weekly loss.
Therefore if I was to just hunt out the games that look like there is an edge, it means I can divide the 8 units via less matches and if/when I do get a winner it would then return more of a profit. This is an idea that certainly provides food for thought.
When we consider that the last three weeks of predictions (24 matches) has only returned 33.3% correct results in total (8) it means that the model is not quite at its optimum, but if it’s still returning a profit more often than not, is it more that the results are too volatile/ the model is too rigid.
Ultimately more and more questions are cropping off with each week that passes and I guess this is the beauty of the test and learn format, because we can answer them along the way and once again it gives me more thinking to do over the coming days.
However, for now I am going to celebrate a return to winning ways and plan ahead for the next round of Russian Premier League fixtures.
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 firstname.lastname@example.org. 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.
Follow me on Facebook at Dan The Stat Man
You can also check out my Premier League Podcast on Soundcloud