Backing The Draw
-
I’ve been doing a lot of work over the past few years on modelling draws in football. With the tendency for the casual punter to favour action, goals and or to back a favoured team, the draw can often be overlooked. This can result in some good value being missed.
I’ve been doing regression analysis with a few years worth of data but can’t present a filter here because I’ve been doing slightly different filters for each league looked at, there’s not a one size fits all.
The majority of the filters are based on variations of the ELO difference between the two teams, the average goals scored and conceded and the Betfair odds. Some also take in to account the home team being the Asian favourite.
It’s pretty much a set and forget strategy although I will be trading out at 80 mins if the game is drawing, to avoid those late goals (although hopefully I’ll benefit from some when we’re not drawing at 80!)
This is still a work in progress and I’m using small stakes to test.
I’m going to use this thread as a journal for me to record progress. It’s also worth noting that I’ll be doing further pre-match research before placing trades (recent form, team news, motivation etc) so not everything that the filters throw out will be traded, although I will keep a log of everything that the filters present, alongside a record of the trades I chose to proceed with.
The strike rate is likely to be between 30% - 35% and with average odds of around 3.30 we should return a profit.
-
@dave-adams sounds like its a good learning curve! And how was the weekend?
-
@daniel-cooper Three losers, one came close but the other two were duds. Think I’m starting to draw some conclusions on suitable leagues so it’s all a learning curve!
-
Friday 23rd October Results
Suwon vs Seongnam @3.25 -1
Kortrijk vs Anderlecht @3.80 -1
Lyngby vs Odense @4.00 -1Well now it’s getting interesting! 77th minute OG left it 3 mins from success in the early Korean game. The Lyngby game was never competitive and ended 0-3, The Anderlecht game was a one sided 1-3 with the consolation being a late OG.
-
How did Friday go?
-
Saturday 24th October
Bristol vs Swansea @3.35
Coventry vs Blackburn @3.55
Altach vs Admira @3.90
Ried vs Sturm Graz @3.95
Brondby vs Midtjylland @3.70
Vaduz vs Zurich @3.95
Lugano vs St Gallen @3.65 -
@dave-adams wonder if you could adapt it into an overall one and see if the home for home and away for away stats effect results positively or negatively, would be interesting to know!
-
@john-hurst Hi John, sorry for the delay, haven't been online so much last few days.
I'm more than happy to share all the filters, I've been trying really hard to get them in to BTC format so that I can share. Problem is I've used the Predictology database to create them and not all of the variables map over.
For example on one I used 'Average Goals Home Team Matches' which means the average amount of goals witnessed in all of the matches, home and away and scored by either team. The BTC filters don't break down quite that granular.
It's not to say the BTC ones aren't as good - they're better and I like them much more, it's just annoying that I have these ones in the other system!
-
Friday 23rd October
Suwon vs Seongnam @3.25
Kortrijk vs Anderlecht @3.80
Lynby vs Odense @4.00 -
Wednesday 21st October Results
QPR vs Preston @3.30 -1
Only the one game for the 21st and that ended 0-2 to Preston. Having seen them play against my team earlier in the season I was pretty surprised to be honest but QPR didn't look like getting back in to it.
Running Total +7.76
-
@dave-adams said in Backing The Draw:
Tuesday 20th October Results
Bristol City vs Middlesbrough @3.35 -1
Coventry vs Swansea @3.50 +1.473rd minute winner for Middlesbrough stopped both coming in but happy again with the selections and added a bit of green! The Coventry vs Swansea ended 0-0 so once I’ve updated the spreadsheet and shared it that’ll show a nice gain in support of leaving the trades to run.
Running Total +8.76
very good!
-
@dave-adams Nice! That's a nice running total so far as well.
-
Tuesday 20th October Results
Bristol City vs Middlesbrough @3.35 -1
Coventry vs Swansea @3.50 +1.473rd minute winner for Middlesbrough stopped both coming in but happy again with the selections and added a bit of green! The Coventry vs Swansea ended 0-0 so once I’ve updated the spreadsheet and shared it that’ll show a nice gain in support of leaving the trades to run.
Running Total +8.76
-
@dave-adams would love to know how you get these picks, mainly so i can try and do it myself. I know you use your own info mate but is there a way any of us could do it? i made a filter which got a few of your picks the other day using just the goals scored/conceded aand a couple of other things so that was good but is there anything else i could do. Thanks
-
Tuesday 20th October
Bristol City vs Middlesbrough @3.35
Coventry vs Swansea @3.50 -
@daniel-cooper The strike rates vary from
league to league but I guess they’d average out at about 35% if I added them all together (I’ll do this tomorrow, I’m planning on sharing a google sheet of all the selections)I know that the best one is 39% but it’s not a huge sample for that league, Portugal I think.
-
@daniel-cooper No need to apologise, my ramblings often go off on a tangent so I may not have explained.
I’ve been studying draws for a few years but just using my own light touch research, soccerway, soccerstats & intuition etc and I never really took it seriously.
Recently I started a really delving deeper and using historical data to backtest some filters. So the filters I use have been modelled on 3-5 years data. They are broken down in to separate leagues, so I have a lot of filters each slightly different to suit the league. For the leagues that have shown a consistent upward trend over that period I’ve started placing money on.
Hope that makes sense! Happy to do in to more detail, drop me a DM if you like and I can share some screenshots and graphs.
-
Do you have a strikerate?
-
@dave-adams said in Backing The Draw:
@daniel-cooper Thanks! Yep I’ll keep posting daily on here, hopefully all the research and testing will hold up!
Sorry if I’m asking a repeat of something you already mentioned - I think you said this had been something you modelled over a long period of time - how many games were tested and what were the results prior to you beginning to place money on the games?
-
@daniel-cooper Thanks! Yep I’ll keep posting daily on here, hopefully all the research and testing will hold up!
