by amber_fluid » Tue Jul 11, 2017 6:52 pm
morell wrote:Are you saying you want a 69 with me?
by morell » Tue Jul 11, 2017 7:04 pm
by bennymacca » Tue Jul 11, 2017 7:08 pm
morell wrote:bennymacca wrote:morell wrote:I just explained how I did!
Why is coaches votes 50% and other things less? Other than it will bias the results?
Coaches votes are
Seasonal
Non Biased
Aggregated from individual games
Anecdotal
I took the Coaches down to 20%. Ryder is still easily #1.
by morell » Tue Jul 11, 2017 7:14 pm
morell wrote:Interesting results. Even without the weightings Ryder wins pretty easily.
by morell » Tue Jul 11, 2017 7:26 pm
by morell » Tue Jul 11, 2017 7:44 pm
by bennymacca » Tue Jul 11, 2017 8:15 pm
by amber_fluid » Tue Jul 11, 2017 8:21 pm
Wedgie wrote:I bow down to you!
by morell » Tue Jul 11, 2017 10:29 pm
z scores are a way of measuring the variance of an individual value within a dataset. Think of it like putting an actual number on "how much" better a player is for a given statistic, instead of just saying better or worse, yes or no, 1 or 0.bennymacca wrote:It still doesn't make any sense why a z score is a meaningful metric based on the stats you have selected.
Again, because disposal efficiency was raised by a poster (not me) to point out a differentiator between the players.bennymacca wrote:Why is disposal efficiency and not hitouts to advantage used?
Again, because a poster raised them. Again, happy to include 2017 player ratings if you can point me to a source. I think that would kinda doubling up as those ratings are similar to what we're doing here anyway - aggregating other stats into one score.bennymacca wrote:Why are goals used as a measure of a ruckman and not afl player ratings points? (Which can be done just for this year too)
No. I would have just used the Coaches awards. Other posters came up with these metrics.bennymacca wrote:You have picked 6 stats and called them the definitive guide for measuring the impact of a ruckman whilst simultaneously explaining away other stats.
Sure thing. As I said, you either didn't understand what I was doing (tick) or were being deliberately obstinate (havent decided yet)bennymacca wrote:If that's what you do for a job then you must be a dodgy accountant or something
by carey » Tue Jul 11, 2017 10:33 pm
by Trader » Tue Jul 11, 2017 11:23 pm
morell wrote:Again, because a poster raised them. Again, happy to include 2017 player ratings if you can point me to a source. I think that would kinda doubling up as those ratings are similar to what we're doing here anyway - aggregating other stats into one score.bennymacca wrote:Why are goals used as a measure of a ruckman and not afl player ratings points? (Which can be done just for this year too)
by bennymacca » Tue Jul 11, 2017 11:35 pm
morell wrote:z scores are a way of measuring the variance of an individual value within a dataset. Think of it like putting an actual number on "how much" better a player is for a given statistic, instead of just saying better or worse, yes or no, 1 or 0.
Its a very common statistical measurement technique used in a variety of methodologies. I havent just made this up.
Its calculated using
z = (x - average) / standard deviation
by morell » Wed Jul 12, 2017 9:50 am
No, not really. Coaches votes are anecdotal. It's their opinion. They might factor in kicks, marks and handballs in their votes but they would also factor in their own more role specific thoughts too. That's what I like about it. Its a series of experts opinion on individual games which is then summised. Not a difficult to interpret or not relevant statistic.Trader wrote:morell wrote:Again, because a poster raised them. Again, happy to include 2017 player ratings if you can point me to a source. I think that would kinda doubling up as those ratings are similar to what we're doing here anyway - aggregating other stats into one score.bennymacca wrote:Why are goals used as a measure of a ruckman and not afl player ratings points? (Which can be done just for this year too)
But doesn't that exact same logic apply to using the raw stats (disposals, goals, hitouts, etc), and then doubling up and also including the coaches votes, which are essentially an aggregate of the other stats into a score?
by Dogwatcher » Wed Jul 12, 2017 9:58 am
by morell » Wed Jul 12, 2017 10:06 am
Rightio, fair enough. I am not trying to say Ryder is 4x better than x or y, just putting across the point that the statistics people have raised ITT have him in front of others. As do the anecdotal measure, like coaches votes. Happy to run it with a smaller deviation using more rucks, but all that will do is condense the scores a bit closer. Not change the order.bennymacca wrote:morell wrote:z scores are a way of measuring the variance of an individual value within a dataset. Think of it like putting an actual number on "how much" better a player is for a given statistic, instead of just saying better or worse, yes or no, 1 or 0.
Its a very common statistical measurement technique used in a variety of methodologies. I havent just made this up.
Its calculated using
z = (x - average) / standard deviation
i know what a z score is, i am not arguing about its value as a metric, i am arguing about the way you have used it.
The standard deviations should include every ruck in the league, not just the players mentioned - you have an incomplete sample which would skew the standard deviations massively
I provided a methodology.bennymacca wrote:I also think your attempt to weight each statistic is a correct one, but there needs to a methodology about how they are weighted, not just plucked out of thin air - your original weightings and also equal weightings are both not correct, but i dont know what the right answer would be
Well of course, goals are correlated to kicks. Kicks are correlated to disposals. The point of spreading it out over multiple measures and weighting them is to extract the differential out of those measures.bennymacca wrote:thirdly, trader is right - coaches votes is a biased statistic as it is correlated to the others in some way
by morell » Wed Jul 12, 2017 10:27 am
Competitions SANFL Official Site | Country Footy SA | Southern Football League | VFL Footy
Club Forums Snouts Louts | The Roost | Redlegs Forum |