Wednesday, 10 August 2011

Value bets this weekend (just for fun remember!)

As discussed on my last post, attached are the bets thrown up for this weekend which are classed as value by TFA ratings. This is bets thrown up by any of the 3 rating algorithms.

I don't want to go into too much detail around what my cut-off value % rates are for each system as it's different for the 3 algorithms and also different for the Similar Games Model I've developed. I can't use the SMG model even if I wanted to as I don't have the time or inclination to look at this at the moment.

One thing I am pondering is how I'm going to manage to get all the system bets out from the 3 rating algorithms each week at a decent time and how I'm going to find the time to do the necessary work for the SGM. The SMG model is fairly complex in terms of Excel work and once I've updated all the rating algorithms, I don't have much spare time for the SGM work. This shouldn't affect the blog too much as I'll be pasting the bets on here each Friday night late on, so I'll have the work done by then I would think. Anyway, something for me to ponder.

Looking at the bets for this weekend, it's hard for me to look at the bets and comment as there is a mixture of 3 algorithms there and I've not put them into the correct systems. What I would say is that you can see the difference in the type of bets thrown up by my new DNB rating algorithm compared to the other bets. Anyone following the blog from last season will notice this.

Teams like Swansea, Stoke, West Brom and Newcastle can only ever be thrown up on my DNB rating algorithm. They are playing Man City, Chelsea, Man Utd and Arsenal!

My first two rating algorithms have the 4 top teams as obvious winners but with no value at all by backing them. The smaller sides are poor value. However, on my new DNB rating algorithm, these 4 smaller teams are thrown up as decent value.

Looking at it logically, Man Utd only won 5 of 19 games away from home last season. Chelsea won 7 and Arsenal won 8. I'm not sure I'd be backing these teams odds on away to teams who finished 11th,12th and 13th last season and who only lost 5 games each at home all season.

I quite like the fact the new DNB rating algorithm is going to give us lots of different angles to hopefully exploit this season as the bets aren't the obvious value bets that are usually thrown up on the other rating algorithms. As you'd expect, there is a much higher variance by backing these teams to win but with DNB, you can actually have a fairly smooth P&L as I've shown using the historical results. Let's hope the same is true this season!

The Swansea one is a tricky one as I've no idea how much you adjust the ratings by to account for teams moving up to the Premiership. I've looked at the last few seasons with decreasing weights on each season and looked at how newly promoted teams perform. Obviously, Blackpool carry a little bit of weight from last season and they started the season very well. Looking through the data, it does appear that newly promoted teams do tend to easily outperform their expected finishing position (if I assume they should finish bottom 3) early on in the season.

Splitting the data between Home and Away, I can see that newly promoted teams at Home are particularly very strong and easily outperform their expected finishing position.

Interestingly, QPR are at home this weekend and I have Bolton as a less negative value bet than QPR are! That's more to do with how QPR ended the season I think and how strong Bolton were last season. All interesting though.

Incidentally, not for a second do I think Swansea will beat Man City this weekend but at odds of 10/1, I can see why my DNB rating algorithm thinks Swansea represent a bit of value in that game.

Anyway, seeing these games thrown up and trying to analyse them and why the ratings pick them is all part of the fun for me. It's just rows and rows of numbers in Excel with different weightings and I just concentrate on the output page which tells me the % chances on a home/draw/away and of course, based on the odds, whice bets are value.

Incidentally, my ratings rarely find value in the draw. The way my ratings work mean that the draw % tends to always be lower than it should be, so it doesn't tend to get many value draws.

For the record, the only 2 games with draw value this weekend are the Swansea and Dunfermline games. Both are away picks by the DNB ratings although Dunfermline also appear on one other algorithm if I remember correctly.

I'll do a post after this weekend on how the ratings did and I'll also post the midweek bets too for next week.

Date Home Team % Value Win Odds DNB Odds
14/08/2011 Kilmarnock 29.38% 2.50 1.78 DNB Value
13/08/2011 Bournemouth 22.85% 2.60 1.84
14/08/2011 West Brom 21.68% 7.04 5.38 DNB Value
13/08/2011 Newcastle 15.83% 4.35 3.14 DNB Value
13/08/2011 Watford 7.91% 2.10 1.52
14/08/2011 Stoke 3.68% 6.03 4.44 DNB Value
13/08/2011 Plymouth 3.28% 2.52 1.78
13/08/2011 Blackburn 2.78% 2.16 1.53
13/08/2011 Accrington 0.35% 2.45 1.73 DNB Value

Date Away Team % Value Win Odds DNB Odds
15/08/2011 Swansea 109.25% 13.20 11.00 DNB Value Wouldn't be a bet as it's Prem Away
13/08/2011 Reading 43.41% 5.00 3.67 DNB Value
13/08/2011 Torquay 28.93% 4.20 3.04 DNB Value
13/08/2011 Cambridge 26.78% 3.25 2.29 DNB Value
13/08/2011 Dunfermline 19.88% 4.50 3.20 DNB Value
13/08/2011 Southampton 11.83% 2.45 1.74
13/08/2011 Walsall 11.15% 3.60 2.58 DNB Value
13/08/2011 Everton 11.06% 4.50 3.24 DNB Value Wouldn't be a bet as it's Prem Away
13/08/2011 Preston 10.97% 3.02 2.16 DNB Value
13/08/2011 Aberdeen 3.69% 5.26 3.89 DNB Value
13/08/2011 Tranmere 1.75% 4.02 2.86 DNB Value
13/08/2011 Swindon 0.39% 2.20 1.58


  1. Not based on this post Graeme, so apologies - but how do you get the fink tank ratings for the scottish leagues?


  2. One other question. When you are posting (for example) reading as 43.41% value, how is that calculated?

    Does it mean your estimated price was about 3.486 (5/1.4341) or is it something else?

    Just for interest, it would be handy to know for any followers, since they could consider staking according to kelly criterion rather than level stakes (as I think we talked about before)

    Apologies if you have answered elsewhere!

  3. Hi Mouldhouse.

    Nice to see you back on here mate. :)

    I don't know much about Fink Tank ratings although I'm trying to find out a little more for my own curiousity. I've asked across at the SBC if there is any way of getting historical data for me to analyse. Dan across there is a fan of them and from looking at my bets this weekend and what Dan posted, we agreed on a few games today. I'd like to investigate further.

    As for Scottish leagues, I'm not sure. I just had a look there and I can only see the top 4 English leagues. Not sure if they do the Bsq Prem too though but I'd be keen to know.

    I'll answer your other comment in a another comment.


  4. In your example which was Reading today (not a bad pick if I say so myself!), I had Reading down as having 28.682% chance of winning that game today. The bookies priced them up at 4/1 which indicated they had a 20% chance of winning the game.

    For my value calcs, I just do (28.682/20)-1. That gives me 1.434-1=0.434 or 43.4%.

    I think that's the same as what you've said as if I work back, the 28.682% chance of them winning is the same as 3.486/1 I think.

    Hope this makes sense mate.


  5. Thanks for that Graeme.

    Is there a mailing list to go on for when the blog is updated, or is it a matter of checking SBC forum?


  6. Hi Mouldhouse.

    If you drop me an email mate, I'll get back to you.


  7. hi i recently found youre blog and was wondering if you do these dnb bets as singles or doubles ect.

  8. Hi Anonymous.

    All my bets this season will be single bets. I looked last season at multiple bets (not on DNB admittedly) but it just introduced an extra amount of variance I can do without in the P&L. I think sticking to single bets is safer although I can't say for sure that DNB multiples wouldn't work.



  9. thanks for replying

    out of curiosity which bookmakers do you use as i can only find select games with dnb on ladbrokes , and how much do you usually stake on these bets if you dont mind me asking?


  10. Hi Anonymous.

    Firstly, DNB is the same as AH 0, so that maybe explains why you can't find odds for them!

    Secondly, if I was following these systems using DNB (or AH 0), I wouldn't be taking the odds on offer by the bookmakers. I'd be taking the best odds for the team to win, the best odds for the draw and dutching them. This usually gives you better odds in 95% of the cases. My odds I'm quoting for DNB are the dutched odds.

    Lastly, I only follow some of the systems using DNB and my stake varies depending on which portfolio it is in and what system it is on. The stake could be anything from £20 to £200 on average I guess with the odd game being higher in stake.

    By dutching the draw and the win incidentally, you should be able to get near enough any stake you want on the games without having to worry too much about restrictions. You'll be placing 2 bets with 2 different bookmakers in most cases and bookies tend to like to take high volume of bets on different leagues on different match outcomes as it looks quite random, so they'll never know you're following a system with an edge hopefully until it's too late.

    Hope this helps.