There is also a well known saying that ‘the best defence is a good offence’. This is an adage I firmly believe in.
I wanted to see if it holds weight in the greatest sport in the world – Aussie rules!
What I have done is taken statistical data from the past 25 AFL seasons to determine the strengths of each grand finalist. Once known, I could then work out who comes out on top and in particular whether defence or attack comes out on top.
Firstly, I ranked each of the two teams offence and defence using their percentages in the home and away period in relation to the rest of the competition for that year. If a team had a higher-ranking offence than defence then a classification of offensive was given with the opposite being true for a defensive classification.
With this method I was able to deduce the following from the 50 grand final teams over the past 25 years. There were 23 teams with an offensive classification, and 17 teams with a defensive classification.
Also, there were 11 teams who qualified as ‘balanced’.
Secondly, I took the total points scored in a given season and averaged it out across all teams. If a team had a higher points-for differential from this base line average compared to their points-against differential, then they were classified as offensive, and vice versa.
With this method 49 of the 50 teams were classified as either offensive or defensive. Amazingly there was only one year where one team was a statistical anomaly. This was unbelievable because the same team also ranked number one in both offence and defence and therefore is the most balanced team statistically to ever play the game.
The only way I could split this team was by going back to the raw average before rounding, resulting in an offensive classification by just 0.39 points.
It is also important to note that on some occasions one team statistically had a better defence, offence or both compared to their counterpart, yet was classified the opposite. This data merely shows whether a team is geared more towards offence or defence – not whether they are better at any given one compared to the other team in isolation.
When I combined the two lots of data, again the classifications held very true. There were only four occasions where a team ranked differently, three of which were bridesmaids and therefore not as relevant to this conversation. I used the more statistically significant data of the two methods to rate them.
The only premiership winning team to not match up perfectly was the 2013 Hawthorn team. As you can see from the data below they ranked first in offence and fifth in defence resulting in an offensive classification using method one. Using method two though the Hawks were classified as defensive.
Going back through the last four years of data an interesting pattern was seen. The amount of points for these given years had risen sharply, most likely due to the expansion teams being belted by 100 or more points by everyone.
Therefore the use of method two started to favour a defensive classification, especially if a team had a reasonable defence. It did not change Hawthorn’s classification of offensive for this year, because method one was the more statistically significant due to a 5/1 ratio of offence to defence ranking.
Where this was relevant though was in 2011 for both Collingwood and Geelong were classified as balanced using method one. Using method two they were both classified as defensive due to their measly defences.
However, they both also had exceptional offences and the expansion teams may have resulted in method two being out for the reasons stated above. I could easily have rated either of these teams as offensive, especially Geelong who won the premiership in this year.
Despite this, there were a couple of close calls on teams so this evens out my decision on the most balanced team ever being classified as offensive. But which team was it?
Overall, the most balanced team ever was the 2007 Geelong Cats, who beat the Power in one of the most dominant grand finals in recent memory. On the other side, the premiers with the worst statistical data was the 2005 Swans.
As for attack versus defence? It appears that being an offence-minded team is the way to go, with 15 premierships being taken out by these teams. Defensive teams took out 10 flags.
Here is the raw data.
1989 Hawks Premiers
• Hawks (Offensive) (PF) 2678 (2) – (PA) 1748 (2)
• Cats (Offensive) (PF) 2916 (1) – (PA) 1987 (6)
• Total Season Points 29123 @ 2080.21 average per team
1990 Pies Premiers
• Dons (Defensive) (PF) 2526 (1) – (PA) 1815 (1)
• Pies (Defensive) (PF) 2376 (4) – (PA) 1825 (2)
• Total Season Points 30816 @ 2201.14 average per team
1991 Hawks Premiers
• Eagles (Defensive) (PF) 2485 (4) – (PA) 1532 (1)
• Hawks (Offensive) (PF) 2793 (1) – (PA) 2055 (5)
• Total Season Points 33861 @ 2257.40 average per team
1992 Eagles Premiers
• Cats (Offensive) (PF) 3057 (1) – (PA) 2099 (6)
• Eagles (Defensive) (PF) 2206 (9) – (PA) 1752 (1)
• Total Season Points 34201 @ 2280.07 average per team
1993 Dons Premiers
• Dons (Offensive) (PF) 2333 (3) – (PA) 1959 (5)
• Blues (Offensive) (PF) 2315 (4) – (PA) 1968 (6)
• Total Season Points 31527 @ 2101.8 average per team
1994 Eagles Premiers
• Eagles (Defensive) (PF) 2078 (7) – (PA) 1572 (1)
• Cats (Offensive) (PF) 2403 (1) – (PA) 2104 (5)
• Total Season Points 31162 @ 2077.47 average per team
1995 Blues Premiers
• Blues (Defensive) (PF) 2357 (3) – (PA) 1711 (1)
• Cats (Offensive) (PF) 2558 (1) – (PA) 1939 (5)
• Total Season Points 33231 @ 2076.94 average per team
1996 Roos Premiers
• Swans (Defensive) (PF) 2152 (9) – (PA) 1737 (2)
• Roos (Offensive) (PF) 2526 (1) – (PA) 1982 (7)
• Total Season Points 33074 @ 2067.13 average per team
1997 Crows Premiers
• Saints (Offensive) (PF) 2294 (1) – (PA) 1918 (7)
• Crows (Defensive) (PF) 2151 (2) – (PA) 1769 (1)
• Total Season Points 31798 @ 1987.38 average per team
1998 Crows Premiers
• North (Offensive) (PF) 2486 (1) – (PA) 2117 (12)
• Crows (Defensive) (PF) 2172 (5) – (PA) 1763 (1)
• Total Season Points 32941 @ 2058.81 average per team
1999 Roos Premiers
• North (Offensive) (PF) 2463 (1) – (PA) 2129 (10)
• Blues (Defensive) (PF) 2088 (7) – (PA) 2028 (7)
• Total Season Points 33687 @ 2105.44 average per team
2000 Dons Premiers
• Dons (Offensive) (PF) 2816 (1) – (PA) 1770 (1)
• Daemons (Offensive) (PF) 2557 (4) – (PA) 2159 (3)
• Total Season Points 36393 @ 2274.56 average per team
2001 Lions Premiers
• Dons (Offensive) (PF) 2548 (1) – (PA) 1895 (3)
• Lions (Offensive) (PF) 2538 (2) – (PA) 1989 (6)
• Total Season Points 34530 @ 2158.13 average per team
2002 Lions Premiers
• Lions (Offensive) (PF) 2520 (1) – (PA) 1843 (2)
• Pies (Defensive) (PF) 2081 (9) – (PA) 1897 (4)
• Total Season Points 33397 @ 2087.31 average per team
2003 Lions Premiers
• Pies (Defensive) (PF) 2259 (3) – (PA) 1858 (3)
• Lions (Offensive) (PF) 2295 (2) – (PA1882 (5)
• Total Season Points 33351 @ 2084.44 average per team
2004 Power Premiers
• Power (Offensive) (PF) 2413 (3) – (PA) 1823 (4)
• Lions (Offensive) (PF) 2447 (1) – (PA) 1783 (2)
• Total Season Points 32796 @ 2049.75 average per team
2005 Swans Premiers
• Eagles (Defensive) (PF) 2261 (3) – (PA) 1824 (4)
• Swans (Defensive) (PF) 1974 (14) – (PA) 1696 (2)
• Total Season Points 33607 @ 2100.44 average per team
2006 Eagles Premiers
• Eagles (Offensive) (PF) 2257 (4) – (PA) 1874 (4)
• Swans (Defensive) (PF) 2098 (6) – (PA) 1630 (1)
• Total Season Points 32712 @ 2044.5 average per team
2007 Cats Premiers
• Cats (Offensive) (PF) 2542 (1) – (PA) 1664 (1)
• Power (Offensive) (PF) 2314 (2) – (PA) 2038 (10)
• Total Season Points 33645 @ 2102.81 average per team
2008 Hawks Premiers
• Cats (Offensive) (PF) 2672 (1) – (PA) 1651 (1)
• Hawks (Defensive) (PF) 2434 (3) – (PA) 1846 (3)
• Total Season Points 34374 @ 2148.38 average per team
2009 Cats Premiers
• Saints (Defensive) (PF) 2197 (4) – (PA) 1411 (1)
• Cats (Offensive) (PF) 2312 (2) – (PA) 1815 (3)
• Total Season Points 32172 @ 2010.75 average per team
2010 Pies Premiers
• Pies (Offensive) (PF) 2349 (2) – (PA) 1658 (2)
• Saints (Defensive) (PF) 1935 (8) – (PA) 1591 (1)
• Total Season Points 31845 @ 1990.31 average per team
2011 Cats Premiers
• Pies (Defensive) (PF) 2592 (1) – (PA) 1546 (1)
• Cats (Defensive) (PF) 2548 (2) – (PA) 1619 (2)
• Total Season Points 34768 @ 2173 average per team
2012 Swans Premiers
• Hawks (Offensive) (PF) 2679 (1) – (PA) 1733 (3)
• Swans (Defensive) (PF) 2290 (5) – (PA) 1629 (1)
• Total Season Points 36470 @ 2279.38 average per team
2013 Hawks Premiers
• Hawks (Offensive) (PF) 2523 (1) – (PA) 1859 (5)
• Freo (Defensive) (PF) 2035 (12) – (PA) 1518 (1)
• Total Season Points 36739 @ 2296.19 average per team
Geoff
Guest
Rick A bit delayed here with my reply but you are being petty. Geelong were up by 2 goals at half time. Enter matera. He kicked 5 goals playing on a wing. End of game. stats are clouding your ability to see the big picture. Every article summarising the grand final turns to materas exploits on the day. Of course he needed team mates to do stuff too as it's a team sport. But with the game in the balance matera stepped up and Geelong had no answer. All other factors were secondary. Maybe watch it and see what influence he had on the result of the match.
Doc Disnick
Roar Guru
I'm going to look into this today/tomorrow using my previous method on these years and maybe a few more. I'll see how this stacks up. Thanks for the info, I will get back to you on this.
Doc Disnick
Roar Guru
Maybe the rest of the competition shifted more defensively and tried to copy the premiers (like many teams do) and WC game DID stay the same resulting in them being more offensive relative to the competition. Remember I measuring ones offensive or defensive capability against the rest of the competition for the year, the previous year does not matter with this data. Maybe after losing the GF they decided to become slightly more offensive. After all this is EXACTLY what Ross Lyon said at the start of the year. He wanted to be a 2 goal a game better side, but I'm not sure that has worked out to well for him thus far. I understand what you are saying and they are all very valid points. In fact that's the entire point of this article to drive this debate. I don't know to be honest, I'm merely look at it from the data, how you interpret that data is the fun part for us :)
Dalgety Carrington
Guest
So you're happy with correlation and not concerned with causality then. Do you think you're getting a true indication of the offensive or defensive nature of the teams with this methodology? At times your definitions seem to get pretty arbitrary results. For example, in 2005 WC shows as defensive, while in 2006 they are defined as offensive. Yet in 2005 WC had a higher points for ranking than they did in 2006, but their points against ranking was the same in both years. It would also be fair to say that WC used the same style of play in both years.
Doc Disnick
Roar Guru
All fair point. The part about game plans doesn't really matter to me or the data. As mentioned above the data doesn't show how someone goes about being more offensive or defensive, it just does.
Dalgety Carrington
Guest
First I must say you've provided a nice thought provoking article Rick. The finals certainly did pop into my head after reading it, it being virtually a separate season in itself and the real place premierships are won. Hard to ignore. Finals success more than any other time is about pressure. The ability to apply or withstand pressure is the key to winning (of course that could be scoreboard pressure and we may see a shift to that in the regular season). But I reckon those lower scores are to be expected in the finals, especially in the modern game. It does seem that you've massaged the data a wee bit and I wonder about the validity of your method of ranking against the average, I'm not sure what this adds as opposed to plain points for vs against rankings. It'd also be fair to say intricate game plans weren't really thoroughly and universally applied before about 10-15 years ago and scoring patterns probably more influenced by personnel before then (and probably after too).
EddyJ
Guest
I've already done the data. Statistically, it's not the best attacking team DURING THE FINALS, that wins the cup. This is what I assessed in 2012 and correctly identified the winner of the Grand Final, on this page: http://www.theroar.com.au/2012/09/27/swans-set-to-defy-the-odds-yet-again/ "EddyJ said | September 27th 2012 @ 8:21am | Report comment This is a difficult grand final to predict but if we consider defence as the main pillar of a premiership, statistically, the signs are pointing to a Sydney win. There has been great talk about the MCG hoodoo for the Swans (one win and a draw from their past 15 matches at the MCG), but there’s another hoodoo coming into play for Hawthorn. The average score against Hawthorn during the finals is 95 points and since the final eight was established in 1994, no premier has leaked so much in defence. Also, the greatest defence points differentiation during a finals series for a premier has been 20 points (1998, Adelaide: average score against = 92, North Melbourne: average score against = 72). For the 2012 finals series, the defence points differentiation is 36 points (Hawthorn: average score against = 95, Sydney: average score against = 56). Other grand finalists to leak as much during a final series include Carlton (1999, average score against = 103), Melbourne (2000, average score against = 97), Geelong (1994, average score against = 92). Of course, this doesn’t guarantee a win for Sydney, but it’s a hard historical statistic for Hawthorn to overcome." The teams that have the loosest defence DURING THE FINALS, usually watch the team with the best defence DURING THE FINALS holding up the premiership cup.
Doc Disnick
Roar Guru
I don't take isolated incidents Geoff, I thought you would have understood that from me in the limited comments we have had with each other :) You will never convince me that any one person won anything for anyone in such a limited time space such as a final. The Eagles came into that match with a pre conceived game plan and that was to supress the Geelong attack. Look at Geelong's score for it was 50% less than their average for the year. It had very little to do with Matera (as much as I love him, not as cool as Turley though), more so the great West Coast Defence.
Doc Disnick
Roar Guru
I have looked at in to some extent, just not in detail. Even if that was the case, it does not necessarily mean a team is more defensively orientated. There are numerous examples of where a team had the best offence and defence, but are more offensively ranked. The same could be true in the finals, but I would need to troll through a lot of data to come to that conclusion. I could also take the more obvious route and say the team that wins the premiership most likely has the best offensive average also. I'd be VERY surprised if this was not the case. Finally you have laws of diminishing returns in the equation because each week you have less teams to compare averages, so it really only holds true if you compare the two finalists.
EddyJ
Guest
This analysis takes into the account the entire season, but if you break it down into just the finals matches, you'll see that since 1994, it's usually the team with the best average defensive record in the finals of that season. I realise that this ends up being a small sample – the grand final teams play at least three finals games, sometimes four (such as Brisbane in 2003, Sydney in 2005, West Coast in 2006), but it's inevitably the best defensive record DURING THE FINALS that wins the flag. Check it out for yourself.
Doc Disnick
Roar Guru
Thanks for the question Dal I used a T-Test However, of my two methods, method two is more statistically significant throughout an entire season vs method 1. The reason I used both is mainly because method 1 is easier to explain to average Joe Blow on this site and it looks good. The data though is to small, but I did use it for the 2013 season for one reason. Firstly, method 1 was statistically significant for teams with ratios of greater than 3/1 if you just took them in isolation, just not when across all teams. Hawthorn was a 5/1 ratio for this year and method 2 really didn't work only for the last 4 years, which skewed my data. After going back I noticed that the points scored was very high in these years and I made an assumption that this was due to the expansion teams. If I wanted to publish something like this then I would have to further explain this, which is beyond the scope of this article, so that indeed was the only real assumption I made that one could question. However, for the sake of this article I think most would agree Hawthorn would be considered offensive in this year. Again due to the length of the article (and for most this is very boring) I did not put in one key point at where my data breaks down due to one massive ASSUMPTION. I have taken data from the home and away season and made an assumption that during a finals campaign this is how the same team will play (that being offensive or defensive). Now I'm also assuming you know a little about stats, because this breaks down completely unless I can find a relation between the home and away season and finals campaign statistically to confirm my null-hypothesis. In order to do this I would have to spend weeks going through all the finals data (which I might do one day to see if it holds true), but I just don't have the time. I hope that answers your question.
Dalgety Carrington
Guest
Sorry Rick, just trying to get my head around this, what test did you use to establish "statistical significance"?
geoff
Guest
Rick please. it was peter materas 3rd quarter that won the eagles the 92 grand final Possibly the most attacking wingman eagles ever had.
Pope Paul vii
Guest
Just on North's 98 effort, they certainly attacked on the day but 2.11 in the second quarter blew their chance of a commanding half time lead. Adelaide, still in touch, stormed over the top. Not that I'm bitter. In 99 Carlton's epic victory over hot hot premiership favourite Essendon, gave North a less accomplished but nevertheless gutsy opponent. Just shows that anything can happen on prelim day let alone GF. St K should have been more attacking, particularly 2009. Same with Syd in 2005 and 2006.
Doc Disnick
Roar Guru
One other thing. Look at that offence in 1992. That is ridiculous, nothing short a extreme defence could have won the day against that Geelong team. Again I'm not saying defence is bad, because I love a good hard solid defence as much as anyone. That was pure Gold watching the Eagles and I truly hated every minute of it at the same time.
Doc Disnick
Roar Guru
Thanks Bosk, it actually took me fecking HOURS to do, but I thoroughly enjoyed doing it. Glad you liked it :)
Doc Disnick
Roar Guru
yep, I will have to check with the guy I paid $400 bucks to who did the stats on my thesis, lol. I'm pretty certain stast6ically that won't hold up, but like I said I'm not a statistician technically. However, despite this, I see what you are saying and would agree that when we take the extreme ranges, there doesn't appear to be that much difference. What is interesting though is who was favourite in all those games, because I checked. This is going to make you think a little. I had a lot more data on each of the two teams in each final, one of which showed the favourite. The team that is top on each years list is actually also the team that finished top. Quite often this also resulted in them being favourites with the bookies, not always but mostly this was the case. Now lets look at those teams that were favourites (with the bookies anyway) for those years. 92 - Cats 94 - Eagles 96 - Swans 98 - North 05 - Eagles 10 - Pies 13 - Hawks It's interesting because the defensive teams won 3 out of 4 times being the under dog in those examples. Like I said, defence keeps you in it. This is a small sample though and hard to come to any conclusion. However, when I retire in another 30 years I might have the time to go through 50 years of data and I'm betting my suspicion will be backed up with conclusive data showing this.
geoff
Guest
i think you know now but will elaborate 92 Eagles had - 8 (ie 1st defensive 9th offense) 8 difference. I made it -8 cause defense is negative. Cats had +5 ie their attack was +5 better than their defense. So the differential between the two is -8 to +5 = 13 Cause Eages won they get the spoils -13 D all the other years were single digit figures so i didnt bother adding them in.
Bosk
Roar Rookie
Congratulations on the excellent statistical analysis Rick. Clearly you've gone to a lot of trouble there.
Doc Disnick
Roar Guru
I kind of under stand what you are saying, but not sure how you have come up with the numbers for each team there 92 you have the eagles as 13 (D). Where did you get this number? I looked at the ratios initially because for the first few years it actually looked like the more unbalanced a team was, the more likely they were to win a premiership. In 1992 this is case in point because the Coasters have a 9/1 ratio with there offence ranked 9th and defence ranked 1st. In 94 the eagles had a 7/1 defensive ration. I think this is what you are saying, but I'm not following the numbers. Wait I see what you have done. You cant do this though (well you can but you cant draw a conclusion from it statistically anyway) because you are taking a ranking from across 14 teams in his case. You are then washing it with two teams, thus causing conflict. Secondly, this was just one method of coming up with a classification. Using my second method was actually more accurate. I see what you are saying, but I will have to think about it more because I used the ratio of the premier to determine if there was a relation between this and whether you are more likely to win as a defensive or offensive team. Interesting though, I will dig deeper. let me know if I did not understand exactly what you mean. Either way I agree with your assessment there and I'm glad someone is interested in it :) You have no idea how many things I picked out of interest and funnily enough this is one of them.