Back To Normal
Back To Normal
If you read my previous Bundesliga article, you will be fully aware that the pursuit of data driven excellence once again took me to the home of Bayern Munich, Borussia Dortmund, Audi and Bratwurst – you get the idea.
Yes it is time to now see how I fared regarding my German data driven football predictions and after week one returned a staggering seven from nine, the challenge was on to go one better, or perhaps even two better and win big.
To win big, I would ‘BundesBot’ to return nine correct picks from last weekend’s Bundesliga action and if that proved to be the case, victory with odds of 402/1 would be coming my way, lets see what would have been needed for this to be the case:
ICH BIN EIN BERLINER
Unfortunately, any dreams of a nine-fold were blown out of the water as Hertha Berlin got the better of the crosstown rivlas Union and with the hosts now under the stewardship of Bruno Labaddia, they are definitely dealing with the new manager bounce.
A bounce that I failed to get the better of and unfortunately, that cost me the ultimate prize. However, there was at least scope to still improve on last week – that is if all the other eight outcomes were correct.
DAS TOP FOUR
In terms of a comparison, there was not as much success this time around and only the wins from the quartet of clubs that now find themselves in the top four of the Bundesliga table, got over the line last weekend. So after nailing seven last time, I’m down to just four our of nine this time around.
Although the overall performance was average, a hat must be tipped for Bayer Leverkusen’s win over Borussia Monchengladbach. This was the result that carried the biggest risk of all, but credit to the visitors and Kai Havertz in particular, they were too much for the hosts cardboard fans.
The fact that there were no draws from ‘BundesBot’ could be considered a little risky from the get go, although when you consider that there were only two draws last weekend, it does seem as if the winner takes it all in Germany.
While another thing to consider, is the fact that home advantage has been all but nullified since the restart. From the first 18 games since the restart, just three have resulted in home wins, this equates to 16.6%
Now some of this will come down to stronger teams playing away from home, but even so, that is an incredibly small return from these past two weeks and something to keep an eye on over the final seven game weeks of the season (Away weight perhaps?)
In comparison, the first two weeks of ‘BundesBot’s operation has returned thus:
Week 0 – Control Week: 7/9 (77.7%)
Week 1: 4/9 (44.4%)
Total: 11/18 (61.1%)
Not a bad return overall, but it must be said a very small sample indeed – but ultimately the more we test, the more we’ll have a better idea of what will go on.
Of course, that’s not the only thing I looked at last week and it is also time to review that as wel:
TAKE A (DOUBLE) CHANCE
With their being no draws offered up by ‘BundesBot’ it made perfect sense to go for a double chance nine-fold acca. Obviously the odds will be far less in terms of landing nine correct data driven football predictions, however the chances of winning would be far greater.
Also unsurprisingly with another layer of insurance, this double chance variant of ‘BundesBot’ fared better, as a respectable six out of nine results come back correct – with the three offenders being:
These three wins were not covered in any of the respective double chance bets and therefore any hopes of a 7/1 win were left short. With that said, it does perhaps add some logic to bigger bets, smalling odds rather than chasing for the ultimate dream.
Then again, going for the ultimate dream is why were doing it, although next season I may start to look at this from more of a profitbaility angle instead, as this whole campaign was more about test and learn than anything else.
With a rare German midweek program, there is a chance to do this all over again very quickly and I think I’ll leave ‘BundesBot’ unchanged. In doing so, I get a better idea of what was the good week and what was the bad week and what direction this model needs to go.
The action points are:
Get closer to seven rather than four correct predictions
Track home win performance
Test Double Chance again
Also look at goal markets (6 out of 9 last weekend ended Over 2.5 goals)
Happy punting and thanks for reading. Dan
(THESE ARE NOT TIPS PER SE, THESE ARE JUST DATA DRIVEN FOOTBALL PREDICTIONS IN A PURE TEST AND LEARN MODELLING CAPACITY)
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.