Build your own Brownlow Model G’day all, your friendly neighbourhood PhD student here. I am here to help you all build your own models. That is very different from what you might be used to which could consist of any one of the following scenarios. Here is a model ...

So watching footyclassified and Matthew Lloyd made a comment that it would seem as though the key to beating Richmond is controlling the ball via having more kicks to handballs. Got me thinking, wouldn’t that be cool if a plot showed the same insight that an industry professional had? ...

library(fitzRoy)
library(tidyverse)

## -- Attaching packages --------------------------------------------------- tidyverse 1.2.1 --

## v ggplot2 2.2.1     v purrr   0.2.5
## v tibble  1.4.2     v dplyr   0.7.5
## v tidyr   0.8.1     v stringr 1.3.1
## v readr   1.1.1     v forcats 0.3.0

## -- Conflicts ------------------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

library(lubridate)

## 
## Attaching package: 'lubridate'

## The following object is masked from 'package:base':
## 
##     date

library(mgcv)

## Loading required package: nlme

## 
## Attaching package: 'nlme'

## The following object is masked from 'package:dplyr':
## 
##     collapse

## This is mgcv 1.8-23. For overview type 'help("mgcv-package")'.

afltables<-fitzRoy::get_match_results()
tips <- get_squiggle_data("tips")

## Getting data from https://api.squiggle.com.au/?q=tips

afltables<-afltables%>%mutate(Home.Team = str_replace(Home.Team, "GWS", "Greater Western Sydney"))

afltables<-afltables %>%mutate(Home.Team = str_replace(Home.Team, "Footscray", "Western Bulldogs"))

unique(afltables$Home.Team)

##  [1] "Fitzroy"                "Collingwood"           
##  [3] "Geelong"                "Sydney"                
##  [5] "Essendon"               "St Kilda"              
##  [7] "Melbourne"              "Carlton"               
##  [9] "Richmond"               "University"            
## [11] "Hawthorn"               "North Melbourne"       
## [13] "Western Bulldogs"       "West Coast"            
## [15] "Brisbane Lions"         "Adelaide"              
## [17] "Fremantle"              "Port Adelaide"         
## [19] "Gold Coast"             "Greater Western Sydney"

names(afltables)

##  [1] "Game"         "Date"         "Round"        "Home.Team"   
##  [5] "Home.Goals"   "Home.Behinds" "Home.Points"  "Away.Team"   
##  [9] "Away.Goals"   "Away.Behinds" "Away.Points"  "Venue"       
## [13] "Margin"       "Season"       "Round.Type"   "Round.Number"

names(tips)

##  [1] "venue"       "hteamid"     "tip"         "correct"     "date"       
##  [6] "round"       "ateam"       "bits"        "year"        "confidence" 
## [11] "updated"     "tipteamid"   "gameid"      "ateamid"     "err"        
## [16] "sourceid"    "margin"      "source"      "hconfidence" "hteam"

tips$date<-ymd_hms(tips$date)

tips$date<-as.Date(tips$date)

afltables$Date<-ymd(afltables$Date)
joined_dataset<-left_join(tips, afltables, by=c("hteam"="Home.Team", "date"="Date"))

df<-joined_dataset%>%
  select(hteam, ateam,tip,correct, hconfidence, round, date,
         source, margin, Home.Points, Away.Points, year)%>%
  mutate(squigglehomemargin=if_else(hteam==tip, margin, -margin), 
         actualhomemargin=Home.Points-Away.Points, 
         hconfidence=hconfidence/100)%>%
  filter(source=="PlusSixOne")%>%
    select(round, hteam, ateam, hconfidence, squigglehomemargin, actualhomemargin, correct)
df<-df[complete.cases(df),]

df$hteam<-as.factor(df$hteam)
df$ateam<-as.factor(df$ateam)
ft=gam(I(actualhomemargin>0)~s(hconfidence),data=df,family="binomial")

df$logitChance = log(df$hconfidence)/log(100-df$hconfidence)


ft=gam(I(actualhomemargin>0)~s(logitChance),data=df,family="binomial")


preds = predict(ft,type="response",se.fit=TRUE)
predSort=sort(preds$fit,index.return=TRUE)

plot(predSort$x~df$hconfidence[predSort$ix],col="red",type="l")

abline(h=0.5,col="blue")
abline(v=50,col="blue")
abline(c(0,1),col="purple")
lines(df$hconfidence[predSort$ix],predSort$x+2*preds$se.fit[predSort$ix])
lines(df$hconfidence[predSort$ix],predSort$x-2*preds$se.fit[predSort$ix])

# predicting winners
ft=gam(I(actualhomemargin>0)~s(hconfidence),data=df,family="binomial",sp=0.05)
# the 0.05 was to make it a bit wiggly but not too silly (the default was not monotonically increasing, which is silly)
plot(ft,rug=FALSE,trans=binomial()$linkinv)
abline(h=0.5,col="blue")
abline(v=0.5,col="blue")
abline(c(0,1),col="purple")

# predicting margins
ft=gam(actualhomemargin~s(hconfidence),data=df)
plot(ft,rug=FALSE,residual=TRUE,pch=1,cex=0.4)
abline(h=0.5,col="blue")
abline(v=0.5,col="blue")

# add squiggle margins to the plot
confSort = sort(df$hconfidence,index.return=TRUE)
lines(confSort$x,df$squigglehomemargin[confSort$ix],col="purple")

Each week triplemfooty tweets out a free kick ladder that is based only on total free kick differential throughout the season. But what if we did a ladder in the same style of the home and away ladder. 4 points for a win (winning free kick count in game) 2 points for ...

I have just finished reading the book Beautiful Evidence by Edward Tufte and in it he talks about the fundamental principles of analytical design. Henri Matisse I do not paint things, I paint only the differences between things Show Comparisons, Contrasts, Differences When looking at a graph or reading a ...

We also see this in sports, Hollinger refers to it as the fluke rule. Bill James calls it the plexiglas principle, you might call it the Sports Illustrated Jinx. They all are different names from different people who have observed the same thing mean reversion or regression to the mean. ...

Linear regression is an amazingly powerful concept in statistical modelling. It forms a major part of lots of first year statistics cources and how to deal with problems with linear regression can form second year units by themselves! This post goal is to through the use of AFL data provide ...

This week in AFL Nathan Fyfe one of the brownlow favourites got rubbed out for an ear tickler to Levi Greenwood When looking at the Brownlow odds, its a bit surprising to see Max Gawn listed as the second favourite. While he is having a great season, historically Ruckman have ...

Apologies for the rushed nature of post Recently I saw this post on twitter and it got me thinking is there another way? So lets run through the rough problems as outlined in this tweet. Numbers have context (suprising I know!) What time period is best when looking at fantasy ...

One of the best things about the online R community is there are lots and lots of great people to follow online especially. One person I follow dataandme consistently links great how to do things on R tutorials. Recently I saw this post where she linked a post by drsimonj ...

Using fitzRoy and the tidyverse

library(fitzRoy)
library(tidyverse)

## -- Attaching packages ----------------------- tidyverse 1.2.1 --

## v ggplot2 2.2.1     v purrr   0.2.5
## v tibble  1.4.2     v dplyr   0.7.6
## v tidyr   0.8.1     v stringr 1.3.1
## v readr   1.1.1     v forcats 0.3.0

## -- Conflicts -------------------------- tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

So what we did before, was we created 3 variables from afltables about Tony Lockett. They were Year (afl season), GM (totals games played in afl season) and GL (total goals kicked in season). These are summary statistics i.e. the year ...

One of the fun things when doing graphs, is that moment when you identify something that sticks out. A great example of this is done by ethan_meldrum In it we can see how much Paul Seedsman 2018 has stood out so far. Lets give it the ggplot/fitzRoy treatment. Step ...

This is the start of the tutorial series, where we will cover visualising and manipulating data in R. There will be a series of mini checkpoints that should be used as a guide to check understanding. To check the most basic functionality of R (you can use it as a ...