In total, there were 20 games highlighted as value by the TFA ratings. Backing 1pt on each game to win returned a profit of 5.3pts (26.4% ROI) and backing 1pt DNB on every game returned a profit of 3.1pts (15.6% ROI).
The DNB return is a little unfair though as my ratings also now have the ability to tell us when there is value in the games for DNB and when there isn't value for DNB purposes. In addition, the DNB ratings deliberately don't take into account Premiership Away games as they always throw up high value bets which seem to make a loss every season. Swansea were the example last weekend which is a good indication of how crazy the ratings can be!
There were 13 games where the ratings determined that there was DNB value if we exclude Swansea. These 13 games returned a profit of 6.55pts and an ROI of 50.4%. If people think I'm trying to make it seem better than it was even though I said Swansea wouldn't qualify as a bet, then Swansea included gives 14 games, a 5.55pts profit and an ROI of 39.6%. However, Swansea wouldn't ever be a bet as I said when I posted the ratings!
I think it's important this season to make the distinction between when my ratings determine DNB is value and simply, using DNB on all games where the ratings suggest backing a team to win is value. Looking at historical results, backing short priced home teams DNB isn't value for example, so these will mainly be excluded from the bets given out by my DNB ratings you'll find. Hence, although the 15.6% ROI for DNB above seems decent enough, I doubt it will beat the DNB returns on the DNB algorithm this season. This new DNB algorithm is totally unproven though, so no one should be following these new bets anyway!
Overall, whatever way you look at the first real set of results from the ratings this season (even if it's only a trial!), the results were very good.
Now, anyone who understands my systems will know that the results above aren't really meaningful. Yeah, it's good to see backing every team made a profit but long-term, this isn't how my systems will be judged. The systems will be judged on how good they are at identifying the higher value bets from the lower value bets. As everyone knows, the higher the value of the team, the more systems these bets will appear on. I deliberately don't give out the value ratings relating to each system as I think I'm best placed to decide which teams should go on which systems.
This is a subtle point but it's very important. I could just give out the bets each weekend with the value % attached and leave it to people to decide on their own staking plan and people could build their own systems. However, I know most would just back every value team and therefore, they'd be missing the whole point of my ratings. The reason my ratings were so good last season and the reason the results look so good historically is because I play on the fact the ratings are very accurate at determining the % of value in each game. Hence, the higher the value, the higher the stake you should be playing. In simple terms, my systems are trying to recreate some sort of basic Kelly staking criterion here which was my intention when I started out with this project.
However, I don't want to tell people the value% cut-off marks for each system (it's different for Home and Away teams on each system incidentally) as I don't want people trying to do something smart with the ratings and trying to use Kelly staking on my bets. The systems are doing this for everyone already, so no one needs to try to be smarter than the systems.
A few people have asked me how the systems would have performed at the weekend. Well, if we look at the results of the games by % value, this tells you all you need to know about Saturday's results.
As you can see, 6 of the top 9 value bets won. An exceptional performance and if every weekend was like this one, I wouldn't need to work I suspect!
At a high level, there were 67 points staked on the systems with a profit of 62.2pts. That's an ROI of 92.8%. By backing all the teams DNB, there was a profit of 44pts and an ROI of 65.7%.
As I said on my last post on here, you really can't read too much into one set of results. Yeah, the results were great last weekend but if you read my blog last season, you'll know that these sort of returns happened fairly frequently but so did losses as large as this happen nearly as frequently! Hence, you can't read too much into these results.
I'll round up the midweek results later or tomorrow. I now need to go and work on the weekend ratings as I'm a bit behind this week!
Hi there
ReplyDeleteI am a new reader of your blog. I find this sort of analysis very interesting. I was wondering if you could clarify what the difference is between the win odds and the DNB odds in the diagram above.
Many thanks
Chris
Hi Chris.
ReplyDeleteWin odds relates to the odds of backing the team to win. This is the odds available at the time I posted the bets. The value % I calculate for each game relates to these win odds.
The DNB odds stands for Draw No Bet odds. It is also known as AH0 by bookmakers who allow you Asian handicap odds. Simply, it is backing the selection to win and also backing the draw to recover your stake if the game finishes a draw.
In most cases, it is better to back the selection to win and then work out the stake to place on the draw to recover your stake. This will allow you to achieve the best possible DNB price for each game. Of course, it takes a little bit of extra work when placing the bet but a small sacrifice to make to ensure you maximise your profits.
Excel can do all the work for you anyway if you input your stake, win odds and draw odds into a simple DNB calculator.
When the system bets start in a few weeks time, it will be easier to follow on the blog as I'll be showing the DNB split stakes for example for each game.
Hope this makes sense.
Graeme