The split-rounds have rolled around again and 6 teams will be taking this weekend off. Luckily, we’re being treated to a particularly intriguing first bye round with 9 of the 10 teams that still lay genuine claims to a finals spot in action over the 6 games that are still going ahead.
We are sure to learn a thing or two about how the top of the table sits. Right now, the FFSS ratings on which these predictions are based are about as tight as you can get up the top.
The current 2nd highest rated side Hawthorn plays 6th highest North on tonight. While the the no.1 team Geelong faces 5th rated Western Bulldogs tomorrow. If the Hawks and Cats convincingly win these games then we will have more faith in our ratings and we might see some separation among the top sides for the first time in a while.
Of course, if the Dogs or Kangas win, we’ll most likely be back to square one regarding ratings. It really is a tight year up top.
My usual qualifiers about using model predictions for betting apply.1
The graphic above highlights all of FFSS’s probability estimates for the week. The prices next to each team are the minimum suggested price to take if you were to bet on them to win the match, assuming that FFSS predictions are perfect. In reality this is, of course, very likely not so.
The graphic opposite shows the FFSS team ratings at the start of the round. These, along with a concession for Home Ground Advantage are the direct inputs for the final probability estimates.
Rather then write previews for every match of the week, I will instead comment on interesting discrepancies between bookie odds and FFSS odds, why these may exist, and whether there is any value in betting at the available price.
North Melbourne v Hawthorn
Friday night seems as good a time as any to remind people of the major flaw the FFSS ratings and predictions currently still has; it considers teams as single entities rather than a sum of individual players. This means ins-and-outs each week are not yet factored into the prediction.
This should give us pause for thought when we see something like this on Thursday night:
4 out vs. 3 in really does challenge our assumption that teams remain essentially the same week-to-week.2
Due, I feel, mainly to these shortcomings, the model rates North as a 48% chance compared to the 28% implied chance being offered by some bookies.
Fremantle v Port Adelaide
The Model sees this one as almost a coinflip. Port are rated 127 FFSS rating points3 higher than Freo but the Dockers maintain a hefty home ground advantage that pulls things back in their favour somewhat.
The potential model shortcoming in this one is not injuries but whether you believe that teams having “something to play for” makes them more likely to win.
Port are challenging for the 8 while the Dockers don’t have much to look forward to. This possibly explains the difference between by 47% chance of a Freo win and the bookie’s implied 43%. If you don’t believe in the “something to play for” phenomenon, then this one deserves a bet on the Dockers.
Western Bulldogs v Geelong
Luke Dahlhaus will be a big out in this one, which as I talked about above, is not factored into the FFSS model calculations. The model has this one at a 46% chance of going to the Dogs while you can get implied chance of 43% at certain bookmakers. This may be worth a bet, but again I feel like the injury leaves too much to chance.
Sydney v Melbourne
I must say, I was a bit surprised to see the model’s predictions for this one quite so one-sided. But then again, that may be me suffering from a recency bias.
If this game were played two weeks ago, with Sydney coming off a win against the unbeaten North and Melbourne coming off a thumping by Port, I probably wouldn’t have raised an eyebrow. It may just be that last week’s results (which weren’t unexpected by any stretch) and the media play up afterwards have got me overrating Melbourne again and underrating Sydney at home.
Tippet is no doubt a loss, but the model’s 85% chance of a win against the bookies 78%, make this a bet for me.
Bet 2 Units on Sydney @ $1.29 – Bet365
- I have translated my calculated probabilities into inferred match odds and compared these to the current prices offered by some of the bigger bookmakers around the country, highlighting any major discrepancies. The reason I have done this is not to recommend or even advocate having a bet on any particular team (although I will certainly talk about “good bets” and “value”). But it is rather used as a way to explore the strengths and weaknesses of the model in greater detail.
Bookie prices can be seen as a general “public consensus” about what the true probabilities of a team winning a match are. When the model differs greatly from the public view it is good to know why. Is it seeing something else that the public are not valuing? Or, is it missing something entirely that others are taking into account? If it’s the latter, then there is clear improvement that can be made, if the former, then I guess we’re on to a winner. All betting amounts will be discussed as unit bets assuming you have 100 units to play with as your full bankroll. For example if you have $1000 that you’re willing to lose over the year if worst comes to worst, then 1 unit is $10. A higher unit bet shows more confidence in the models assessment and the value to be made.
- Writing this blog each weeks helps me assess the model and think about potential fixes. Playing lists are certainly something I am looking at working into the next major update to the ratings system. Hopefully I can get something together by the end of the season, if not, definitely next year they will be a more major focus.
- Or roughly 2 wins out of every 3 games between the two teams at a neutural venue.