Carlton Supporters Club

You know what I'm getting from all these stats is this...
Hit out stats are totally useless.
HTA stats are also pretty useless given their low frequency.
Even clearance stats aren't that reliable unless the clearance clearly advantages your side.

The thing that really matters is the connection between your ruckman and his midfielders in terms of positioning, both attacking and defensive, and the second efforts of all players in the vicinity.

That's not something you can measure with a statistic.
It's something you observe.

If you're in any doubt about who was the most effective ruckman in our game last weekend just watch the game again.

Remember our chat about sample size of data....thats what is important.

The most important stat from the Essendon game (and all games for that matter) is the Points For v the Points Against. Carlton 96 to Essendon 70 is what I care most about. Everything else can be read in many ways to suit anyone and many times will just muddy the water.

Jinx

The most important stat from the Essendon game (and all games for that matter) is the Points For v the Points Against. Carlton 96 to Essendon 70 is what I care most about. Everything else can be read in many ways to suit anyone and many times will just muddy the water.

Sample size is important, no doubt about it.

But no matter how big the sample size, the data is garbage if the definitions used in the collection methods are flawed, and ultimately that is the problem with many of the AFL stats, the problem is in the methods.

The R&D saying, "Garbage In = Garbage Out, you can keep piling it higher but it's still garbage!"

Remember our chat about sample size of data....thats what is important.

Any individual game can be an anomoly.
Look at the same information, over weeks and years and you will get trends that 'absorb' these anomolies.

For example, look at the results this week. Look who won, who lost and the margin. Use that information to tell me who the best team is.
Now, if that is the only data you have, what information could you get out of that in terms of overall strength of any one team relative to another? Relative to last week? Last year? Last decade?

Our first 5 games of R13 were this....
North Melbourne won.
Richmond won.
Hawthorn won
Brisbane won
St. Kilda won

Based on that and that alone, you would think that those teams should be in the top half of the league (they won, same amount of teams lost, so they are top half).
Based on that, you would probably look at the margin each team won by....which would rank them in this order....
Brisbane
North Melbourne
Richmond
Hawthorn
St. Kilda

Extrapolate that out and you would assume that Brisbane and North Melbourne are the most successful in recent years as they had the best margin.

In isolation, thats all you have.

Then you look at the ladder...
12th - Hawthorn
13th - Brisbane
14th - St. Kilda
17th - Richmond
18th - North Melbourne

That paints a completely different picture.

Then go back and look at where they finished last year.
Look at average finishing position over the past 10, 20, 50, 150 years.
Look at finals made, flags won...
Eventually you will get a better understanding of who is actually good now, last year, over the past decade or 2, 50 or 150 years.

You see the point?
More data gives you a better understanding.

Don't throw out any particular stat because of their low frequency. Just get more data before using it.
There is more hitout to advantage then there are goals.
Do we deduce that Elijah Hollands is the best goal kicker in the league because he kicked 3 goals this week?
Is Charlie and Harry just average forwards because they didn't kick as many?
Is McGovern as good a forward as Harry given they kicked the same amount of goals?
No, get more data. It will show you trends and give you better information.

There are more stats the AFL used, which i've eluded to previously, that the general public does not have access to but a got a glimpse of last year.
Talks about ruck hitouts sharked etc on top of that.

Nobody is saying its perfect.
But its not black and white either.
That information is better than no information. It paints a better picture than no information.

You've argued about how the midfield does half the work for the ruckman in the past......to which i've responded with relativity in that Pittonet was consistently above TDK despite using the same midfield and same rotations.

So, don't go better your house on these stats. But if its a choice between betting your house on these stats, or 1 persons individual opinion on any given battle, i'm putting my money on the stats because the opinions vary so wildly.

Someone bagged out TDKs game this week, i think in the in-game thread, then he got BOG from both coaches.

Try this little exercise Kruds.

Go back over the last 50 years and look at our season goal tallies

I suspect what the 'extended data' will give you is a wide range of scores,

They'll be affected by

Our ability at the time
Pesonnel available
Coaches
Game plans
The way the games is played.
Rule changes.... and a host of other factors.

Each of these variables also affect the statistics of teams and individuals.
You often see a team change tack mid season and a weakness in the first half of the season becomes a strength in the second half..or at least not the same issue.

Watch Melbourne in the second half of the year....
With Petracca missing some may step up and increase their individual stats, others will find an extra burden and attention and their stats may actually drop...and that's just the effect of one player.

Football stats are more likely to have a very limited life span as variables affect the way the game is played.

And that's the flaw in your argument.  The stats that you use to determine whether Pitto is above the King are unreliable because they are poorly chosen/defined, are too simplistic, are dependent on the actions/abilities of more than one player, and are subject to the fickle bounce of the footy. 

Let's say that I spot you on the half forward flank and fire a 35m pass to your advantage.  You chose to wear short stops and lose your footing as you sprint towards the ball.  LP dashes past you, gathers the ball and boots it back over my head.  My kick goes down as an ineffective kick because you fell over and LP took advantage, not because my kick was poor or ill-directed.  LP's kick goes off the side of his boot but travels 45m to where EB and the Prof are wrestling for position.  The ball bounces off their hands and rolls out of bounds and LP is credited with an effective kick to go with his intercept. 

Not suggesting otherwise lods. But that is the point. Looking at the data over a large period of time will show you these changes. As it will not just be club to club that is varying but league averages as a whole. Year to year, decade to decade.

The thing is though it will all be relative.  In the 70s all the teams were scoring more than any of the teams now. But there will be an average (afl/vfl wide) over that period and looking at the teams above that vs the teams below that will tell you a story of relative strengths to each other.

You can also analyse that to see how much better a team is vs the average at the time and that would be comparable to today.

In simplistic terms, looking at how many wins a team has will vary year to year with teams output but also amount of games in a season.
So instead look at % of game they have won.
A 18-2 team is the same as a 9-1 team, 90% win record. That eliminates the season length variable.
There are other tricks to use as well.

All data will fit on a bell curve with 99%(?) Of data fitting within 3 STD deviations of the mean. The further away from that the rarer it is....the outliers.

You compare year to year based off of that.
You could do it over a decade or 3 weeks or whatever time period you want.
But you 100% can compare and in an unbiased way.

I don't want to get bogged down in a maths lecture, suffice to say, low numbers does not make it negligible.

We should probably agree to disagree Kruds.
Neither of us are going to change our opinion

I really don't discount statistics completely.
I often use them myself...but I give them less value than I do observation, even observation that contains a bit of the occasional bias.

I guess the way I see it is you're trying to apply the "logic of mathematics" to the most illogical and unpredictable of games.
A game where things that shouldn't happen often do.
A game where things such as 'momentum' in a match can change in minutes just by a side lifting their effort.

The complexity changes in the observation and the complexity of the mathematics change along with it. Everything can be taken into consideration if you have enough time.

If mathematics can be used to help model and predict the weather, a game of afl is nothing. 😉

Yep and it's equally as likely that the opponent slips over too.

Simply put for everything that can go your way there is an equivalent that can go against you and work in your favour.

At the end of the day the sheer weight of numbers balances everything out.