AFL Prediction & Analysis

Squiggle on Preliminary Finals & Grand Final Preview

Squiggle trades in cold, hard data, so there will be no place in this post for barracking or bandwagoning. Sure, outside of this thread, I may be clicking LIKE hard and often on every pro-Richmond post I see. Sure, this is the first time Richmond have made the Grand Final since I was 9 years old, and my parents took me to the Royal Melbourne Show on Grand Final Day because it was less crowded. “You can watch the next one, Max,” they told me, and you know what? I bloody well will. I have a ticket and I am going. Sure, there were times when it felt like this would never happen in my lifetime. But you won’t catch me mentioning any of that here.

Both favourites won their prelims comfortably. But Adelaide were more impressive in defeating the Cats 136-75, while in PF2, or as squiggle knows it, “Who Wants To Be Slaughtered By Adelaide,” the Tigers ended the Giants 103-67.

That result leaves the Crows approaching somewhere near the gravitational centre of the squiggle’s premiership cups, which means their 2017 results are similar to those of many premiership teams from the last two decades. They’re still somewhat modestly rated for a premier, but they’re not far off it.

The Tigers, on the other hand, are a more niche model, belonging to a distinct class of teams with strong defences but mediocre attacks. That does include a few premiers, most notably the reigning premier (for a few more days, at least) in the Western Bulldogs, but many more who fell short, including a swag of Ross Lyon-coached teams.

Adelaide’s year bears a striking similarity to 2016, when the Crows also opened with a series of blisteringly high-scoring games, then began to tighten up defensively as well:

The Crows were squiggle darlings for most of last year before wobbling late and falling to the Swans in a semi-final that psychologically scarred Adelaide-based squiggle fans for all time. They still don’t look enormously different to that team, but they’re better balanced and the opposition standing in their path to a flag has been weaker, with GWS in particular less dangerous than the 2016 version.

The 2016 semi-final loss is interesting in light of this year’s Grand Final, because the 2016 Swans and 2017 Tigers look to have a bit in common, with both enjoying low-scoring slugfests. So one might wonder whether the Crows have a weakness in that area that the Tigers are designed to exploit. But by moving into the more balanced area of the chart, Adelaide are showing they’re more well-rounded and less brittle, better able to deal with a variety of game plans.

Fun Fact: Both Port Adelaide and Essendon remain eligible for a classical 4-Strike, by losing an Elimination Final to a team that lost a Semi-Final to a team that lost a Preliminary Final to a team that loses a Grand Final.

Squiggle is unique in placing a particular emphasis on attack-defence balance. Computer models commonly produce a single “strength” number and tip whoever’s strongest. Squiggle does this for all games except the Grand Final, where it leans heavily toward the team with the stronger attack, due to the over-representation of that trait in past premiers.

Without this, Squiggle would still tip Adelaide, as the overall better team, but far less emphatically.

Not only is Adelade the better-rated team, and not only do they have the better attack, but they’re also more balanced. Most Grand Final upsets have happened to unbalanced teams, such as Sydney 2016 (defensive specialist) and Hawthorn 2012 (attack specialist), and there are quite a few near-misses as well, like a well-balanced Port Adelaide coming within 3 points of attack-first Hawthorn in a 2014 prelim, and Geelong doing almost the same thing against the Hawks a year earlier. So while it’s been more rewarding to be a team that engages in high-scoring games rather than low-scoring ones, it’s also been preferable not to be too far one way or the other.

There are exceptions: The Bulldogs should have lost to GWS in last year’s prelim, according to this theory, and Geelong shouldn’t have lost to anyone in 2008. But in general, it’s ideal to be the best overall, the best attack, and the most balanced – and compared to Richmond, Adelaide are all three.

As such, Squiggle is very heavily into Adelaide, tipping the Crows by 41 points.

But that doesn’t mean the Tigers have no hope. While most of the Squiggle stable of algorithms are tipping Adelaide, there’s a healthy contingent of counter-tips from venue-based models, many of which take special note of Richmond’s amazing record at the MCG this year. That stable contains 202 algorithms at last count, and of the 36 that favour the Tigers, all of them pay special attention to the venue, either in the form of general interstate home ground advantage or results at the ‘G in particular.

That’s because Richmond haven’t had a bad match at the MCG since Round 8, when they lost after the siren to Fremantle. There were really only two Tiger embarrassments this year: one at the hands of the Crows at Adelaide Oval and another courtesy of the Saints at Docklands. But at the ‘G, Richmond have gotten better almost with each passing week. Their record since the bye looks like this:

  • defeated by Sydney by 9
  • def Carlton by 26
  • def GWS by 19
  • def Hawthorn by 29
  • def St Kilda by 41
  • def Geelong by 51
  • def GWS by 36

That compares favorably to the Tigers’ record away from the MCG, which is still good but shakier, including the aforementioned St Kilda humiliation as well as a loss to at Kardinia to Geelong.

The Crows’ record at the MCG over the same period is:

  • def Carlton by 12
  • drew with Collingwood

Their only other game there this year was a 24-point defeat of the Hawks in Round 2. So if you believe that the venue makes a difference, there’s plenty of reason to think it will strongly favour the Tigers. And it’s been a good year for venue algorithms: they’ve actually been the most accurate of the stable this year, both in terms of tips and average margin predictions. They don’t have the same long-term record, but they have had the better read in 2017.

It’s also worth mentioning that Squiggle has underrated Richmond all year, as it did with the Bulldogs last year. When a team improves quickly over a short period, as the Tigers have in rising from a 13th-placed finish last year, Squiggle takes time to be convinced they’re the real deal, since that kind of radical change is unusual, and more commonly signals a team that’s temporarily overperforming and will soon revert to their natural level. If the Tigers really have become a lot better at football in a short period, they may still be underrated.

In terms of similar Grand Finals, 2017 bears the most resemblance to 2006, when West Coast (a well-balanced and modestly rated team) defeated Sydney (a defensive specialist) by 1 point:

That was, of course, one of the great all-time cracking Grand Finals, so we’d be blessed to get anything like a repeat.

In summary: Adelaide are deserved favourites, and have the ability to kick away and win by a significant margin. But any footy follower knows anything can happen in football, and there’s room between a 15% chance and a 0% chance to drive a premiership cup through. If the Crows don’t break free from the Tiger stranglehold and get locked down into a scrappy, high-pressure series of ground contests; if they kick inaccurately; if the venue plays a role; then it will be a Richmond premiership.

P.S. I just wanted to finish the post with the phrase “Richmond premiership.”

Discuss on BigFooty!

Squiggle on Finals Week 2


Geelong were so impressed by the 90 to 40 thing that Richmond did that they tried it on the Swans. And it worked!

This was a wild upset in the most obvious sense, because Sydney haven’t looked like losing a game for a long time, unless they were playing Hawthorn. But it was also something of a typical 2017 performance for Geelong, who have spent the year backing up weak performances with strong ones, and vice versa.

What that means, I’m not sure. But it was a real shocker for the Swans, who have been so good, they’re still ranked #1 on regular form even after that pasting. But it’s time to lower their flag:

The Giants looked a different team as well, after trialling a new game style where they actually look interested in football. This is a bit alarming because they were pretty good even before they got interested in football. The Eagles, however, retreated back to their squiggle nest where they’ve spent the entire season. This could really be the least squiggling from a starting point ever.

Still, both the Giants and Cats are positioned quite a way outside the premiership area, with only two games to make up ground. Richmond are closer, but are lower-scoring, which, as discussed above, has proven problematic for teams in the past.

With Sydney out of the way, it looks like Adelaide in a canter.

That said, it’s worth remembering how probabilities work, and that a 70% chance of a win followed by an 80% chance of a win sounds like something approaching a sure thing, but in truth it’s close to a coin flip (0.7 x 0.8 = 0.56) as to whether Adelaide takes out the flag or not.

Also, while Adelaide have been ranked #1 by flagpole non-stop since Round 1, they haven’t gotten to the heights reached by teams like Hawthorn, Geelong and Collingwood of years gone by. They’ve been the best team all year, but not by a whole lot. It would be a real surprise for the Crows to lose from here, but not a shock.

Adelaide Crows vs Geelong Cats

Richmond Tigers vs GWS Giants


Discuss on BigFooty!

Why Squiggle is down on Richmond

Question via BigFooty:

Looking at the Flagpole at the moment: why is Geelong ranked so much higher than Richmond despite Richmond being ranked ahead of them in both offence and defence?

Because Richmond is in an area of the chart where good teams go to not win premierships.

Over the last twenty years, only three teams have done it from there: Bulldogs 2016, Sydney 2012 and Sydney 2005. All were upsets. Two were huge upsets, part of a series of other upsets.

Meanwhile, it’s been a graveyard for otherwise strong teams who failed to go all the way: Sydney 2016, Sydney 2015, Richmond 2015, Fremantle 2015, Sydney 2014, Fremantle 2014, Fremantle 2013, Sydney 2013, Fremantle 2012, St Kilda 2011, St Kilda 2010, Adelaide 2010, St Kilda 2009, Adelaide 2009, Sydney 2006, Adelaide 2006, Adelaide 2005, Sydney 2004, Geelong 2004, and Carlton 2001. I suppose I should add Sydney 2017 now, too.

Flagpole is different to the regular squiggle algorithm because it’s slanted to favour teams that look like past premiership winners, and not like teams that win games but not flags. That basically means it prefers teams with a strong attack, and dislikes teams that look like they’re being coached by Ross Lyon.

This is all done without any special reasoning; i.e. squiggle offers no theory on why, exactly, defensive specialist teams have done poorly in finals after successful regular-season campaigns. It just notes that they have, and expects the trend to continue. The landscape today may look very different if St Kilda had snagged a couple of flags in 2009 and 2010 and if Fremantle had managed it in 2013. But they didn’t, so this is the historical reality, which Richmond need to overcome.

Squiggle on Finals Week 1

a.k.a. “Richmond goes wheeee!”

Throughout the season, the squiggle refused to rate Richmond any higher than a somewhat-above-average team. Now there are only six teams left, the Tigers are rated 3rd, i.e. still just on the better side of average. But they made a big move after holding Geelong to only 40 points, essentially pulling level with Adelaide in terms of raw form.

That’s still not enough for a flag, because squiggle respects the historical fact that attacking teams have outperformed defensive teams when it counts over the last two decades. So while the Tigers have a dream run to the Grand Final, avoiding both top teams and hosting an interstate opponent for a prelim, squiggle doubts they can contain the scoring power of Adelaide or Sydney if they get there.

Nevertheless, the Tigers are certainly close enough to give it a shake, which is just about the most exciting thing I’ve experienced in my entire football-following life. So that’s something.

There’s not a whole lot to say about Adelaide v GWS or Sydney v Essendon, both of which played out to expectation.

But Port v West Coast! Once again, the Power failed to knock off a well-rated opponent, although only just, in the cruelest possible way. Taking an 8 scoring shot advantage and turning it into 2 fewer goals and 10 more behinds is going to sink any team.

Despite all the evidence Port has offered to the contrary, there’s still reason to believe they’re a very good team, and if they can avoid self-immolating over the off-season, they could follow in the footsteps of Sydney 2015, who kicked themselves out of their first final against Fremantle (7.18 to 10.9), then rebounded the next year to win the minor premiership and contest the Grand Final. Port’s stumbles in 2017 have been absurd, dramatic and memorable, so it’s easy to forget they annihilated the bottom tier of the competition, beat most of the middle tier, and weren’t too far off the rest, while recording wins over Sydney and West Coast.

Meanwhile, the Eagles are lurking just close enough to be dangerous. All year, they’ve been thereabouts, competitive in practically every game, including those against everyone they have to beat for the flag. They are rightly underdogs, but having landed on the soft side of the draw, it’s not beyond the realm of possibility that the Eagles could defeat the Giants, who they only lost to by 8 points earlier this year, then Richmond, who they lost to by 11, and then take on Sydney or Adelaide, both of whom they’ve beaten.

An Eagles flag would be mind-boggling, given they were unlikely even to make finals a few rounds ago. In another year, I’d say it was fanciful, because normally we have at least one truly dominant team plus a tradition of rested Qualifying Final winners beating up Semi-Finalists. But this year, our top two in Adelaide and Sydney haven’t created as much separation from the field, you only have to beat one of them (unless you’re Geelong), and there’s the unknown quantity of the pre-finals bye in play. So almost anything is within the realms of possibility.

On flagpole, it’s bombs away and lights out!

Richmond squeak into 3rd even with the heavily defensive game plan that Flagpole abhors. And Geelong are still rated pretty well, even though in the real world everyone has concluded that they are the worst football team to ever play the game and who will win that Adelaide v Sydney prelim.

Geelong Cats v Sydney Swans

There’s a long history, of course, of people overrating Elimination Final winners and underrating Qualifying Final losers. But it’s still hard to see Sydney dropping this one: They’re a premiership favourite who finished much lower on the ladder than their form suggested, while the Cats made top 4 on the back of a few close wins and didn’t fire a shot against the Tigers.

GWS Giants vs West Coast Eagles

This one is harder to pick, with neither team having stood out over the season, but both also refusing to go away. In a neutral venue, squiggle would pick the Eagles, but only just; in NSW, it gives it to the Giants.

I’m just going to stare at that for a while now. I basically made that Squiggle Doors function just so I would have an excuse to imagine Richmond Grand Finals in alternate universes. Now there might be one in this one! That’s even better.

By the way, I would also like to take this moment to enjoy when hardly anybody has started hating Richmond crowds yet. I know that’s coming, because Richmond crowds when the Tigers are on the verge of success are the fucking worst. Not for me. They’re fine for me, because I’m wearing yellow and black. But for other teams’ supporters. I didn’t see anything bad at the Geelong game, but I’ve been to every final of the last five years, and they bring out plenty of mentally damaged individuals who have been psychologically broken by the last 35 years of heartache and now clearly believe they are the plaything of a malevolent deity who can only be banished by a spray of alcohol-fueled insults. If we make a Grand Final, it’s going to be so ugly. I mean win or lose, either way, there is going to be some shit on display. So it’s nice to get in before that.

Discuss on BigFooty!

Models Leaderboard After Home & Away

2017 has been a rough year for tipping, even tougher than 2016, which itself was a wake-up call after the lovely, predictable Hawthorn dynasty.

Sadly, none of the models managed to outperform the bookies in tipping winners head-to-head during the Home & Away season, nor in terms of probabilistic Bits.

Tips Bits MAE Correct

4 8 4 7 7 6 3 5 5 7 4 4 3 7 5 5 7 6 6 9 8 7 5

132 22.1970 66.7%

5 8 4 7 6 6 3 5 6 7 3 3 2 8 5 7 7 5 6 5 8 7 6

129 17.9055 28.89 65.2%

5 9 4 6 7 6 3 5 4 7 4 3 2 6 5 5 6 7 7 8 6 6 7

128 17.5804 64.6%

3 8 5 6 7 7 4 5 4 7 4 1 3 6 4 6 8 7 7 8 6 6 6

128 16.3202 28.87 64.6%

5 7 5 6 6 6 4 5 7 7 3 2 3 7 3 5 7 5 7 7 7 7 7

128 15.4157 30.11 64.6%

5 8 4 6 7 6 3 5 5 7 3 3 2 6 4 5 7 6 7 8 7 6 5

125 20.1305 28.92 63.1%

5 7 5 6 6 6 3 5 5 7 3 3 3 6 4 6 7 6 6 7 7 6 6

125 16.5200 29.03 63.1%

4 7 4 6 7 6 3 5 5 7 4 3 2 6 3 5 6 6 7 8 8 6 5

123 18.6520 28.73 62.1%

5 8 5 5 5 6 3 5 5 8 4 2 2 7 3 6 7 6 6 7 6 5 6

122 17.0455 29.79 61.6%

A particularly rough year for Squiggle’s in-house tipping algorithm, which came in dead last and didn’t deserve to be any higher. But very credible performances elsewhere, especially from Plus Six One (for the second year in a row), and it’s worth noting The Arc‘s chart-topping Mean Average Error in a year when the line-ball games didn’t fall their way.

I was curious this year to see what kind of performance the Aggregate would have, where it represents a simple average of everyone else’s tips. Would there be some kind of wisdom of the crowds effect, where it could outperform most of the individual models that provided its inputs? Well, kind of: It landed mid-table in terms of tips and MAE, but with more Bits than all models. So there could be something there.

Apologies for finals tipping being a little shaky at the moment: Apparently everyone (including Squiggle) uses different formatting for posting their finals tips, and squiggle-bot needs to learn how to parse it.