BRAB
Biggest Victim on this Site
- Oct 20, 2014
- 32,988
- 63,489
- AFL Club
- Brisbane Lions
Does Match Thread Posting have a Correlation with Qooty Results?
By BR AB
Much has been said over the true randomness of the sim. The exact nature as to how the sim outputs results is a mystery to many, with league legend Mobbenfuhrer coding the program himself using a microwave.
In theory, the results of the sim should be truly random, in a 2-way contest, that means on average teams should lose as many games as they win. Naturally, this does not occur season to season, as is the nature of a coin toss. Just because something should happen 50% of the time doesn’t mean it will. The small sample size between Sweet FA seasons only amplifies this effect season to season. Even over the course of the entire history of the competition we still don’t yield extremely the high sample sizes to get towards that expected value. Nevertheless, a team like the Demons is probably most suited for this, having played 414 games in their history if you exclude finals. It is pertinent to consider that the expected value of an outcome with two possible results is 0.5, consider also that the experimental value may differ a bit from that. (Oh * me in the dick draws exist too, for the point of this exercise they do not count) Fundamentally, as more games are simmed in the Sweet FA, we should get closer to that magical 0.5 number. Outliers take us away from that number, specifically, the fact that I doubt Mobbs was capable of designing a truly random sim. Any outliers 2 or more standard deviations outside the norm get more interesting though. These could be driven by factors not yet considered by the league.
Today I am here to analyse whether posting quality or quantity is correlated with the results of the sim.
This is based on a few premises. Firstly, that the ratio of season results (Wins/Games played) should be close to 0.5. This means a team, over its history, should be pretty close to winning as many games as it loses. Anything 2 standard deviation outside that should be investigated, as that forms that last 2.1% of results on either side of a bell curve. We are going to perform this analysis on my club, the demons, in an attempt to find any correlation between posting stats and sim results. Firstly, the raw data.
Visually inspecting this data shows us a few values that we could expect to be outliers. 0.77 in Season 3 and 0.25 in season 21 for instance. However, we can confirm these are outliers through standard deviation analysis. By “Standard Deviation analysis” I mean using two commands in excel, and by “analysis” I mean bending the stats to suit my preconceived views.
Anyway, the average ratio for a normal demons season is 0.50. Astonishing really, it means over the Demons history they have won about as many games as they have lost. As in, Mobbs has actually created a random sim. The standard deviation (Number by which any result can deviate from the mean) is 0.12. Using a bell curve, we can determine which results would be considered extreme outliers, and analyse those seasons further.
I’m just going to assume the dataset is normally distributed. I doubt anyone will care. Basically, any season 2 standard deviations outside the mean has about a 2.1% chance of happening, (4.2% for both directions of deviation)
In real terms. Any number below 0.26 or above 0.74 can be considered a relatively extreme outlier.
A regular season takes into account natural variation in the sim. If something else influences the sim, say, posting quantity and quality of a team in a specific season, then we can perhaps account for the outliers in this way.
Season 21 from the demons is a pertinent example. The club was in crisis, with posting numbers at their worst in the clubs history relative to other teams at the same time. As a result, the club had the worst ratio of wins to losses in the clubs history.
To me, this starts a compelling case to look into other teams, and determine whether posting quality/quantity has a tangible effect on the results of the sim. From there, we can perform correlation analysis on each specific instance where we suspect an outlier has occurred. My main question is has the Demons example unearthed a smoking gun, whereby for the entire league's history, a Sweet FA sides posting stats have secretly been influencing results and consequently taking teams to glory. Are the best posting sides in the comp potentially hogging wins based on their constant output of relevant, enjoyable posts in match threads?
In my search for another example to support my hypothesis, I ventured into the final matchup of season 23, the Season 23 Grand Final.
I have now come to a firm conclusion that posting quality/quantity is in no way correlated with on field Qooty results. What a waste of time.
Enjoy grand final week.
By BR AB
Much has been said over the true randomness of the sim. The exact nature as to how the sim outputs results is a mystery to many, with league legend Mobbenfuhrer coding the program himself using a microwave.
In theory, the results of the sim should be truly random, in a 2-way contest, that means on average teams should lose as many games as they win. Naturally, this does not occur season to season, as is the nature of a coin toss. Just because something should happen 50% of the time doesn’t mean it will. The small sample size between Sweet FA seasons only amplifies this effect season to season. Even over the course of the entire history of the competition we still don’t yield extremely the high sample sizes to get towards that expected value. Nevertheless, a team like the Demons is probably most suited for this, having played 414 games in their history if you exclude finals. It is pertinent to consider that the expected value of an outcome with two possible results is 0.5, consider also that the experimental value may differ a bit from that. (Oh * me in the dick draws exist too, for the point of this exercise they do not count) Fundamentally, as more games are simmed in the Sweet FA, we should get closer to that magical 0.5 number. Outliers take us away from that number, specifically, the fact that I doubt Mobbs was capable of designing a truly random sim. Any outliers 2 or more standard deviations outside the norm get more interesting though. These could be driven by factors not yet considered by the league.
Today I am here to analyse whether posting quality or quantity is correlated with the results of the sim.
This is based on a few premises. Firstly, that the ratio of season results (Wins/Games played) should be close to 0.5. This means a team, over its history, should be pretty close to winning as many games as it loses. Anything 2 standard deviation outside that should be investigated, as that forms that last 2.1% of results on either side of a bell curve. We are going to perform this analysis on my club, the demons, in an attempt to find any correlation between posting stats and sim results. Firstly, the raw data.
Visually inspecting this data shows us a few values that we could expect to be outliers. 0.77 in Season 3 and 0.25 in season 21 for instance. However, we can confirm these are outliers through standard deviation analysis. By “Standard Deviation analysis” I mean using two commands in excel, and by “analysis” I mean bending the stats to suit my preconceived views.
Anyway, the average ratio for a normal demons season is 0.50. Astonishing really, it means over the Demons history they have won about as many games as they have lost. As in, Mobbs has actually created a random sim. The standard deviation (Number by which any result can deviate from the mean) is 0.12. Using a bell curve, we can determine which results would be considered extreme outliers, and analyse those seasons further.
I’m just going to assume the dataset is normally distributed. I doubt anyone will care. Basically, any season 2 standard deviations outside the mean has about a 2.1% chance of happening, (4.2% for both directions of deviation)
In real terms. Any number below 0.26 or above 0.74 can be considered a relatively extreme outlier.
A regular season takes into account natural variation in the sim. If something else influences the sim, say, posting quantity and quality of a team in a specific season, then we can perhaps account for the outliers in this way.
Season 21 from the demons is a pertinent example. The club was in crisis, with posting numbers at their worst in the clubs history relative to other teams at the same time. As a result, the club had the worst ratio of wins to losses in the clubs history.
To me, this starts a compelling case to look into other teams, and determine whether posting quality/quantity has a tangible effect on the results of the sim. From there, we can perform correlation analysis on each specific instance where we suspect an outlier has occurred. My main question is has the Demons example unearthed a smoking gun, whereby for the entire league's history, a Sweet FA sides posting stats have secretly been influencing results and consequently taking teams to glory. Are the best posting sides in the comp potentially hogging wins based on their constant output of relevant, enjoyable posts in match threads?
In my search for another example to support my hypothesis, I ventured into the final matchup of season 23, the Season 23 Grand Final.
I have now come to a firm conclusion that posting quality/quantity is in no way correlated with on field Qooty results. What a waste of time.
Enjoy grand final week.
