Competitive balance in the AFL

[autocol]

Hey Switftdog, good post. I already understand normal distribution. but you did a pretty good job of explaining it so hopefully others will follow along.

My question is, how is the "ideal" standard deviation formula arrived at? I realise you divide the average number of wins by the square root of the number of matches played, but why is that the ideal standard deviation? Is that an arbitrarily selected formula based on observation (by whoever thought it up), or is there a more concrete fundamental reason for that to be the case?

 

[Gibbke]

So? Have you seen any statistics that suggest the population growth is happening in "non-AFL" territories? Because I seem to recall seeing some stats which said that the percentage of the population participating in Aussie rules is steadily increasing, and has been for many years. "Participants", by the way, are the group of people from which we draw the AFL talent pool.

It's alright though, if you'd like to ignore reality in order to cling to your preconceived idea, you're welcome to.

Instead of bagging the guy who has basically brought up a good point regardless, can you show us some exact figures of this population growth relating to Australian Football and not just a general Australian population rise...? Clearly, an increase in Western Sydney v an increase in Melbourne will bring about different results in teenage footy recruitment, the point where scouts and clubs get serious...

 

[mianfei]

My thoughts is that you don't really need competitive balance, what you need is to maintain hope for every club that success is possible. Competitive balance may not even be desirable and having Collingwood win 3 times as many premierships as Melbourne, say, might even be a good thing for the competition. As long as Melbourne is not constantly at the bottom of the table.

That's a sensible point on economic as well as cultural grounds.

However, Stefan Szymanski said that in fact it is not a bad thing if some teams actually have no hope of success - as was the case with:

1. Hawthorn between 1925 and 1953 (de facto profit-maximising club with limited zoned area, no wealthy patrons, plus little interest in competitive team sports in catchment)
2. North Melbourne from 1925 to 1943 (limited zoning and no patrons)
3. St. Kilda from 1941 to 1955 (no wealthy patrons and little interest in competitive team sports in catchment)
4. South Melbourne, North Melbourne, Footscray and Fitzroy in the 1960s (old suburbs dominated by supporters of soccer rather than football, no historical patrons to recruit top players and keep fan interest)
5. Melbourne and South Melbourne in 1970s and 1980s (weak zone from gerrymandered country zoning system, no corporate patrons of any significance, small supporter base affected by southern European migrants)

In all these cases, attendances (which were then the measuring stick) remained high to very high and what issue existed was the cost of supporting these poor clubs.

 

[Kwality]

A conclusion you arrived at before the discussion had even begun, and which you are clinging to despite logical commentary to the contrary.

Not so, I wasnt across your thinking, sadly flawed that it is.

 

[Gibbke]

That's a sensible point on economic as well as cultural grounds.

However, Stefan Szymanski said that in fact it is not a bad thing if some teams actually have no hope of success - as was the case with:

1. Hawthorn between 1925 and 1953 (de facto profit-maximising club with limited zoned area, no wealthy patrons, plus little interest in competitive team sports in catchment)
2. North Melbourne from 1925 to 1943 (limited zoning and no patrons)
3. St. Kilda from 1941 to 1955 (no wealthy patrons and little interest in competitive team sports in catchment)
4. South Melbourne, North Melbourne, Footscray and Fitzroy in the 1960s (old suburbs dominated by supporters of soccer rather than football, no historical patrons to recruit top players and keep fan interest)
5. Melbourne and South Melbourne in 1970s and 1980s (weak zone from gerrymandered country zoning system, no corporate patrons of any significance, small supporter base affected by southern European migrants)

In all these cases, attendances (which were then the measuring stick) remained high to very high and what issue existed was the cost of supporting these poor clubs.

This isn't true. Hawthorn endured regular calls for exclusion due to non-competitiveness. Outside South and Melbourne, all of these teams endured lengthy spells where their attendances were abysmal, contributing excessively to the list of sub-10000 attendances over time. And at the end of the day, all of these teams have been under the hammer at times fighting for their very survival, with only two able to say that they are safe in 2013 (one of those with heavy AFL subsidisation after a forced relocation). In the mid-1980's, only Hawthorn and Melbourne in this group were able to entertain the notion of membership in a breakaway comp.

 

[swiftdog]

Hey Switftdog, good post. I already understand normal distribution. but you did a pretty good job of explaining it so hopefully others will follow along.

My question is, how is the "ideal" standard deviation formula arrived at? I realise you divide the average number of wins by the square root of the number of matches played, but why is that the ideal standard deviation? Is that an arbitrarily selected formula based on observation (by whoever thought it up), or is there a more concrete fundamental reason for that to be the case?

Good question and one that I can't say that I really know the answer to. I've done a little bit of research and found a paper (Humphreys, 2002) that presented a formula for the standard deviation of winning percentages in an idealized league. An idealized league would be one where every team has a 50% chance of winning on any given day. We know that mathematically this is always the case but in reality (considering form, injury, talent avaliable, home-ground advantage etc) this isn't actually the case. The formula from Humphreys was:

Standard deviation = 0.5/sqroot(G).

The 0.5 refers to a team's 50% chance of winning a game on any given day and G indicates the amount of games played per season. I assume that Noll-Scully just expanded on this formula. In an idealized, completely mathematical, world where a team has a 50/50 shot of winning a game on any given day you would expect the average team to win 11 out of 22 games. As such I think this is how we arrived at the stdev = 11/sqroot(22) formula and a standard deviation of 2.35 wins.

 

[Sprout]

When the EPL is listed as having a more competitive balance than the AFL, you know these stats are basically rubbish.

This, and I say that as an EPL fan.

To say that La Liga has twice the competitive balance of the NBA is an absurd notion. Real Madrid and Barcelona have absolutely dominated the title race in the period 1976 onwards, with only four other teams achieving success outside of those two in that entire time.

 

[RogersResults]

Something to look at may be that the nature of scoring in Australian Football means that better teams will tend to win more often with less opportunity for weaker teams to 'steal' matches? The average winning score 1997-2012 has been 16 goals to 11 - does this make for a difference from a competition were the average winning score is more likely to be 2 goals to 1* that Noll-Scully doesn't account for?

The main reason for abolishing free kicks for out of bounds ( in force from 1925) and re-establishing the boundary throw-in in 1939 was a concern that that these rules had resulted in scoring 'blow-outs' leading to a more uneven competition.

* Manchester City's current average to the nearerst goal.

 

[autocol]

Gibbke, you at least have a point (unlike my mate Kwality, it seems), and one that's worth addressing. I don't have the figures at hand (and searching the ABS site on an iPad sucks, so I'm not having any luck locating it just now), but I recall seeing data which said that the participation rates for Aussie Rules football increased nationally by about 5% between 2005 and roughly 2010. If the national participation rate increases, and so does the population, then the total pool of participants has increased in size. Granted, perhaps more of the population growth happened in western Sydney, and more of the participation growth happened in South Australia, but that doesn't change the overall growth in the participant pool.

National Population x National Participation Rate = Total Number of Participants.

If both of the first numbers have increased (and they have), then the talent pool is bigger.

THAT, Kwality, is the real world. If there's a sad flaw in my argument, at least do me the favour of spelling it out.

 

[flamingEMBERS]

NFL probably has one of the fairest draws in the world of sport, especially when not every team plays each other.

 

[Gibbke]

NFL probably has one of the fairest draws in the world of sport, especially when not every team plays each other.

Not really. It's one step away from a knock out comp. Two years ago I watched my Rams play the last Sunday night of the sason against the Seahawks, the winner to be crowned NFC West champs, and if the Rams won they would have gone into the playoffs with a 7-9 record...as it was Seattle got there with 8-8, with a home ground advantage against a better performed side. When the conferences are imbalanced, you can sit there happy in the knowledge that every team has a similar road to the playoffs, but the rankings of the teams based upon season performances can be very skewed...

The only part of the NFL draw that resembles a strategic plan of equalisation is the two games you get against teams that finished in the same divisional position as you the previous season, and when you consider the difficulty NFL teams have in maintaining consecutive winning seasons, you can see it's equalisation in name only. You should know - many opposition team fans going into 2011 thought they'd be getting an easy game against the Eagles! The other fourteen games are completely set in stone for what looks like all time to come - six against div opponents and the other eight against a regularly rotated sequence of NFC and AFC opponents. This scenario isn't that far removed from the current AFL situation - regular games against rivals, and the only thing missing is the separated ladders for actual divisions.

 

[RogersResults]

Does Australian Fooball give better teams more time to eventually impose their superiority in a game - leading to a higher percentage of wins for such teams than in American Football?

11 Minutes of Action

"According to a Wall Street Journal study of four recent broadcasts, and similar estimates by researchers, the average amount of time the ball is in play on the field during an NFL game is about 11 minutes." Wall Street Journal 15 Jan 2010
http://online.wsj.com/article/SB10001424052748704281204575002852055561406.html

 

[autocol]

I'm not sure that's a relevant statistic, Roger, because the potential for a decisive, game winning play to occur in only a few seconds is vastly higher than in Aussie Rules.

 

[RogersResults]

I'm not sure that's a relevant statistic, Roger, because the potential for a decisive, game winning play to occur in only a few seconds is vastly higher than in Aussie Rules.

Exactly my point - much harder in American Football than in Australian Football with the time available for better teams to build up a lead where a play occuring in only a few seconds can be decisive in turning the match the other way. The minimal scoring in soccer may have a similiar effect.

12.5% of AFL matches 1897-2012 have had results where one more score of a goal would have changed the winner. (1 behind would have changed the result of the 1.06% of matches that ended as draws and made draws of of the 2.2% of matches that were won by 1 point.) Anyone have the stats for the NFL or EPL?

 

[flamingEMBERS]

Not really. It's one step away from a knock out comp. Two years ago I watched my Rams play the last Sunday night of the sason against the Seahawks, the winner to be crowned NFC West champs, and if the Rams won they would have gone into the playoffs with a 7-9 record...as it was Seattle got there with 8-8, with a home ground advantage against a better performed side. When the conferences are imbalanced, you can sit there happy in the knowledge that every team has a similar road to the playoffs, but the rankings of the teams based upon season performances can be very skewed...

The only part of the NFL draw that resembles a strategic plan of equalisation is the two games you get against teams that finished in the same divisional position as you the previous season, and when you consider the difficulty NFL teams have in maintaining consecutive winning seasons, you can see it's equalisation in name only. You should know - many opposition team fans going into 2011 thought they'd be getting an easy game against the Eagles! The other fourteen games are completely set in stone for what looks like all time to come - six against div opponents and the other eight against a regularly rotated sequence of NFC and AFC opponents. This scenario isn't that far removed from the current AFL situation - regular games against rivals, and the only thing missing is the separated ladders for actual divisions.

Yes and no. It is extremely tough when you dont play everyone twice and its all about your division. However I think they have done it pretty fairly. More to the point at least you know what your getting into each year. Theres complete transparency. I could sit down right now and tell you who everyone is playing next season. You dont really get that with AFL at all.

 

[Slax]

Exactly my point - much harder in American Football than in Australian Football with the time available for better teams to build up a lead where a play occuring in only a few seconds can be decisive in turning the match the other way. The minimal scoring in soccer may have a similiar effect.

12.5% of AFL matches 1897-2012 have had results where one more score of a goal would have changed the winner. (1 behind would have changed the result of the 1.06% of matches that ended as draws and made draws of of the 2.2% of matches that were won by 1 point.) Anyone have the stats for the NFL or EPL?

Don't have a figure for the EPL but willing to bet that if you worked on the basis that 1 goal would change the result then it would be running at over 50% when you consider that automatically all draws and 1 goal wins are counted.

 

[Matt_BL]

Exactly my point - much harder in American Football than in Australian Football with the time available for better teams to build up a lead where a play occuring in only a few seconds can be decisive in turning the match the other way. The minimal scoring in soccer may have a similiar effect.

12.5% of AFL matches 1897-2012 have had results where one more score of a goal would have changed the winner. (1 behind would have changed the result of the 1.06% of matches that ended as draws and made draws of of the 2.2% of matches that were won by 1 point.) Anyone have the stats for the NFL or EPL?

12.5% is actually fairly high. 1/8 basically means that during the season there should be one close match per week, it'll be a little less as there is 9 matches played but thats not bad. There's probably a difference between a thriller and a close match though but that can't really be measured.

 

[mianfei]

This isn't true. Hawthorn endured regular calls for exclusion due to non-competitiveness. Outside South and Melbourne, all of these teams endured lengthy spells where their attendances were abysmal, contributing excessively to the list of sub-10000 attendances over time. And at the end of the day, all of these teams have been under the hammer at times fighting for their very survival, with only two able to say that they are safe in 2013 (one of those with heavy AFL subsidisation after a forced relocation). In the mid-1980's, only Hawthorn and Melbourne in this group were able to entertain the notion of membership in a breakaway comp.

Actually, I was referring not to the attendances of these hopeless teams, but to the general attendances of the League! If you look at the record attendances during the very wet season of 1981, where Melbourne were a lucky 1-21 with a win after the siren and Footscray 2-20, it does seem as if the cost of supporting teams with no hope of winning is worthwhile.

I know Hawthorn did have calls for exclusion from the press (correct me if I’m wrong and it was from somewhere else) during the early 1950s, but if Szymanśki is correct, then the strong and rich clubs knew it was worth supporting them, or they would have voted Hawthorn out no doubt for a replacement from the VFA (unless they voted St. Kilda out too).

 

[Monkey King]

The AFL has a number of measures in place to ensure competitive balance between its clubs ...

One issue I have with your analysis is the time frame. It was only during the 90's that the measures really started to take effect. Prior to this the salary cap wasn't heavily enforced and their were a number remnants player wise from the days before the measures took effect.

An interesting point also is that the the power teams in this era were aided by influences anomalous to competitive measure; such as the Lions merger & early concessions and the F&S rule pre-bidding days.

 

[calyam]

12.5% is actually fairly high. 1/8 basically means that during the season there should be one close match per week, it'll be a little less as there is 9 matches played but thats not bad. There's probably a difference between a thriller and a close match though but that can't really be measured.

Although I suspect that would change if you excluded the first few decades of the competition. Average lower scores means that the likelihood of a game being decided by under a kick should be higher. I suspect the number of games decided by under a kick would be much lower in recent decades.

Does Australian Fooball give better teams more time to eventually impose their superiority in a game - leading to a higher percentage of wins for such teams than in American Football?

Another contributing factor for the lack of competitive balance in the AFL is that no one player is particularly important. In an NBA game, if LeBron James has a poor game then the likelihood of beating his team increases dramatically. When he left Cleveland his team won 42 fewer games the following season (in an 82 game season). One player can have a huge effect on a game's outcome.

Similarly in the NFL, a gun quarterback having an off game can have a massive influence on the outcome on the game. Or similarly a gun quarterback missing a series of games can completely derail a teams season: see, Peyton Manning.

In the AFL no one player is hugely important because there are so many players on the field and a lot of them do similar jobs. Consider this, when Gary Ablett Jnr left Geelong the Cats actually became better the next season. They actually improved significantly from the season before. Losing the best player in the league led to more wins and a premiership, which is in stark contrast to the best players in the NBA and NFL missing games.

To paraphrase: for an upset to happen in the AFL a lot more has to go wrong (or right for the team causing the upset). A great player missing the game or playing poorly may not be enough because they are not that important to the outcome of the game. For an upset to occur in the AFL you may need a number of players to be missing or to be completely off their game and that is less likely to occur than simply having one great player having an off day (or one opposition player having the game of their life).

I'd also add that with the increasing professionalism in the league the difference between a players best and worst is probably less than it has ever been, which should also contribute to a reduced number of upsets.

 

[RogersResults]

Although I suspect that would change if you excluded the first few decades of the competition. Average lower scores means that the likelihood of a game being decided by under a kick should be higher. I suspect the number of games decided by under a kick would be much lower in recent decades.

Percentage of matches where 1 more goal would have resulted in a win for the team scoring that last goal (draws and margins of 1 to 5 points).

1920-2013 R19 - 11.3% of matches
1994-2013 R19 - 10.6% of matches
2004-2013 R19 - 11.2% of matches

 

[calyam]

Percentage of matches where 1 more goal would have resulted in a win for the team scoring that last goal (draws and margins of 1 to 5 points).

1920-2013 R19 - 11.3% of matches
1994-2013 R19 - 10.6% of matches
2004-2013 R19 - 11.2% of matches

Thanks Roger. Not quite as large as I had anticipated.

 

[Gus Poyet]

When the EPL is listed as having a more competitive balance than the AFL, you know these stats are basically rubbish.

When the EPL takes in 1976 onwards despite not starting til 1992 you know it's rubbish.

 

[Damon_3388]

You'd see more competitive balance with less teams IMO. Like it or not, having 18 teams dilutes the product severely, and means that blokes who are otherwise probably not AFL standard (now or in the future) are getting games, either under the guise of "developement" or "helping out the kids".

 

[clogged]

I love that the measure on whether a league has a lot of beltings is called Scully.