How to Create a Team Rating Model

The majority of bookmakers and professional punters use ratings systems. They’re most commonly applied to horse racing. However, ratings are just as relevant to sporting models, including AFL.

There are different ways to create a rating, and certainly different variables to consider. For example, a player rating model gives every single player a value and their team of 22 makes up a collective score. Alternatively, you can give each team a rating and use player changes as one of your variables. In this article, we’re going to look at the latter.


What Is The Purpose Of A Rating System?

Essentially, you’re trying to determine how many points better one team is from another. If two teams met on a mutual venue what would the margin be? We determine this by giving every team a rating and then we match them against each other to create our line (e.g. +13.5) for the match.

We then include extra variables like team news and weather, and that gives us our final line.

For example, Richmond might be 28+ and North Melbourne 44.5 so our line would be Richmond -16.5.

If the markets are betting Richmond -4.5 then depending on our own rules, we’ll be backing Richmond in that market. The more accurate our ratings and the more discipline we show, the more successful our betting.


So Where Do We Start?

We assign every team a number. Remember, that these numbers evolve after every game or with every significant injury. Don’t get hung up on perfection. Just start your own ratings system and improve it with experience and additional variables.

A good starting reference is the Betfair premiership market. The premiership favourite, or your premiership favourite, should be zero. Every team after that is x points worse. And by points, we mean actual score. Remember, we’re trying to create a number that can be matched up when the teams play, and the difference is the line.

2017 Betfair Premiership Market, after round 9, May 22nd

At this point in time, the Adelaide Crows are marginal favourites, ahead of the Giants. Let’s start them on 0. The Brisbane Lions are last in the premiership market. Let’s give them a score and then work out all the numbers in between.

Subjectively, on a mutual venue with no weather, how many points head start would the Lions need? At what point would you take the Brisbane Lions at the +?

-40: You’d take Crows

-50: Crows

-60: Crows

-70: Nearly 12 goals is too much, Lions

-65.5: No bet, that’s my line

Therefore, your ratings so far are:

  • Crows 0
  • Lions 65.5

As you can see, this is subjective opinion using one futures market and eight rounds of results. It’s not perfect, but it’s a start.

-65.5 looks like a huge line, however, the Crows and Lions just played at the Gabba. The Crows won by 80 points. there was no weather and Tom Rockliff, as expected, didn’t play. The line during the week moved from -50.5 to 60.5, validating out -65.5.

Using similar thought processes, examinations of the draw and the Betfair market, your complete ratings system might look like:

  • Adelaide Crows: 0
  • GWS: 3.5
  • Western Bulldogs: 8.5
  • Port Adelaide Power: 10
  • West Coast Eagles: 12
  • Geelong Cats: 13
  • Sydney Swans: 17.5
  • St Kilda Saints: 19.5
  • Fremantle Dockers: 22
  • Essendon Bombers: 24.5
  • Melbourne Demons: 25
  • Richmond Tigers: 26.5
  • Collingwood Magpies: 28.5
  • North Melbourne Kangaroos: 32
  • Hawthorn Hawks: 35
  • Gold Coast Suns: 40
  • Carlton Blues: 42
  • Brisbane Lions: 65.5

These numbers are based on teams being full strength and includes season long injuries. Short to medium term injuries are factored in when you do your weekly head to head market. Venue data, another influence, is also fluid.


Venue Data

Once you have your team rating you need to assign a number to the home ground advantage. Let’s look at some examples so you can see how these values have to vary.

West Coast hosting St Kilda at Domain Stadium might be a 12 point advantage. However, West Coast hosting Fremantle in a derby might only be worth 2-3 points because of the crowd being predominately West Coast.

Sydney hosting Adelaide at the SCG might be worth 10 points. Sydney hosting Collingwood might be worth 0 points, given the Pies have won their last 8 of 12 games across all venues in Sydney.

St Kilda are going to have a stronger value at Etihad than the MCG. Geelong are even stronger at Simonds.

Recent history between the two teams, found on FinalSiren, will tell you how pronounced the venue advantage is. The number most punters use will sit between 0 and 12 points.


Week To Week Injuries

The last factor to consider on a team rating model is short term injuries. The amount of points you subtract from a particular team is essentially the difference of value between the two players. If Patrick Dangerfield misses two games and is replaced by Sam Menegola and your rating model makes Geelong 5 points weaker, then you’re concluding that Dangerfield is a 5 point better player than Menegola.

Given the AFL is 22 players it can be dangerous to overvalue one player’s influence, even if we’re talking about Lance Franklin, Dustin Martin or Nat Fyfe.


Let’s Put It All Together

Now that you’ve got an understanding of how to create a team rating, let’s do a practical example.

St Kilda vs Fremantle at Etihad Stadium, and Jack Steven is out for four weeks with a hamstring:

  • St Kilda’s current rating is 19.5
  • Fremantle’s is 22

You would think that Etihad is a huge advantage for St Kilda, given their reputation at the venue. However, both team’s form at Etihad matches how their season went.

  • 2014 Fremantle 2 wins – 1 loss / 2014 St Kilda 8 wins – 4 losses
  • 2015 Fremantle 2 wins – 1 loss / 2015 St Kilda 3 wins – 8 losses – 1 draw
  • 2016 Fremantle 0 wins – 3 losses / 2016 St Kilda 9 wins – 5 losses
  • 2017 Fremantle haven’t played / 2017 St Kilda 4 wins – 3 losses

The market is going to put a lot of value on that long flight and St Kilda’s reputation at Etihad. However, recent history says that the team’s ability and form is much more important. Fremantle were 0-3 last year but they weren’t winning at any venue in 2016. St Kilda are 4-3 at Etihad this year, it’s hardly a dominant home ground, especially after last week’s 50 point loss to Sydney there.

We’ll give St Kilda 5 points for the venue.

Our line is therefore:

  • 19.5-22
  • -5 (Etihad influence is a St Kilda -)
  • +2.5 (Steven injury is a Fremantle +)
  • =.5
  • We mark St Kilda -5 or 5 point favourites

If the market is significantly different, say 8 or more points off our line, we could say it is worth a 1 unit bet.


Ok, But How Do I Get More Accurate Ratings?

Our ratings are researched but they are subjective. The more accurate models are automated and weighted. They find the most influential statistics and prioritise them in order of significance. Determining what they are might involve conversations with Champion Data officials, AFL coaching staff (past or present) or using chat rooms like Big Footy.

Another option is building a spreadsheet or model, using recent games. That’s how the Betfair Model determines ratings and weekly odds.


Conclusion

Rating models are a great way to learn how to price games. Bookmakers and professional punters use them in almost every sport. They’re not overly complicated to start and with initiative, you can dramatically improve your accuracy. If you prioritise the right values you’ll find mistakes in the market and turn those opportunities into profit.


Related Articles

How to Find Value in AFL Futures: Premiership

If futures betting isn’t your forte, it soon will be. The opportunities are undeniable and the nature of Betfair’s ...

Read More

How to Build a Bot Without Knowing Code

Make money while you sleep. It isn’t as far-fetched as you’d think. There’s thousands of Betfair members whose primary ...

Read More

Trends that Predict a Team’s Performance

When you’re trying to determine a team’s chances of winning a particular match, there is an overwhelming amount of ...

Read More