Bernieslavenschin
Well-known member
Frustrated at the ongoing debate about whether or not form matters going into the playoffs (and constant need to wash and dry my wet sheets)
I decided to have a look at some previous results and answer this question.
Let the long thread commence….
(Feel free to skip to the end)
I will apologise in advance to any mathematicians out there with regards to my statistical analysis. I realise I may not have chosen the best way to analyse results, and am very open to criticism or suggestion as to alternatives.
Many will be familiar with this website: Do form and position make the EFL Championship Playoff winner? Northwestern Sports Analytics Group
For those that aren't, it seems to be one of the only sources of information regarding this topic. The frustrating thing is it was published in 2018 and hasn't been updated since.
On the plus side It does include data from 1989-2017, and looks at chances of promotion via playoffs based on league position (across multiple English leagues), and chances of promotion via playoffs based on form in the 2nd tier.
The author’s conclusions were that:
1.League position
-Finishing higher in the league (across multiple tiers) is 'likely but not certain to be a factor in a team's playoff run'-
-The results for all leagues were almost (but not) statistically significant. The results for the Championship were even less so.
2. Form
-In the Championship, on average the team that won the playoffs scored 1.14 more points than the other playoff candidates in the 7 games leading up to the playoffs, however, this was also not statistically significant.
-The author suggested as much, but then went on to conclude that 'form probably correlates to winning the playoffs but we can't reach a conclusion'
(Form in this case was measured by taking the team with the best form over the last 7 games, and comparing this to the other 3 team's averages in the playoffs)
Overall they suggested teams should ideally enter the competition in good form and in a good finishing position.
Makes sense but I'm not sure it was quite backed up by the data.
I figured it was time to look at the years 2017-2022 as well.
Method
I started off gathering data with the intent of providing some sort of new insight or analysis on form specifically, given that Boro are in a relatively poor run of form and I wanted to see if that correlates to poor performance in playoffs or not
I looked at various websites and ultimately used a site called sportsmole:
Sportsmole example
I collected data on the previous 18 seasons, I picked this number because the website I used only recorded back to ’04-’05.
Records are available back to 89-90 (and these were used in the previous analysis)- but the websites I found that recorded the results made it a lot more difficult to read through.
I could do it but it would be more time consuming. - I am aware this automatically means I have less data than the website above. This thread isn't here with the intention of being the definitive/end product, more of a gentle probe.
I arranged the data into tables for each season, looking at final league position, points picked up in the last 5 games, last 10 games, and then used these points to rank teams based on form also.
Eg:
I did this for each of the 18 seasons, then compiled the data to make some charts.
I've shamelessly copied the display format of the data in these next two charts from the original website listed above.
RESULTS
A suggestion of a trend towards final league position and promotion, more on this later.
For the next chart I used a similar method from the original website. I took the amount of points earned by the promoted team in their final 5 and final 10 games (I called these 'form points'). I then subtracted the average points of the other 3 runner up teams for that season for each set.
IE: In the above table Norwich got 10 points in last 5, and 21 in last 10
That’s 0.67 points better than the average over 5 games and 3.67 better than the average over 10.
Plotting this on a graph, you can see on that in the 14-15 season, Norwich appear just over the 0 line for the 5 game form, and higher up for the 10 game form.
(The idea being that if form makes a big difference to the team that gets promoted, the teams that get promoted should persistently have more form points than the other 3 average score, and if that was the case the data points would mostly be above the line marked 0)
As you can see, it's a mixed bag, and certainly not a clear trend.
Honourable mentions for Palace (2012-2013)- Who got promoted despite having only 8 points from their final 10 games, and Huddersfield (2016-17)- Promoted wth only 4 points in their last 5, and 10 in their last 10.
Maybe there's hope for us yet....
Indeed, at least 5 teams have had similar (or worse) form points to us currently and still got promoted.
I did run T tests for the above data, and found no statistical significance between form points difference and promotion. (I have this data available if anyone wants)
Finally, I wanted to look closer at the relation between league position and form, and how this correlated with promotion or not (rather than just ripping off someone else's work)
I used a binomial logistic regression, which in this case should help predict whether a team is likely to get promotion or not based on different variables.
I put in data for all 72 teams over the 18 seasons, and looked at team league position, form across the final 5 games, and form across 10 games.
I started off analysing these variables in the same set of data, but then realised that the variables weren't necessarily independent of one another, so then analysed them separately. (Teams with good form in 5 would likely have good form in 10, and likely have a high league position- so it would be difficult to decide which one of these variables was related to promotion on it's own)
EG:
My findings from this were
1. Final league position did seem to correlate with chance of promotion (P value 0.034)
2. Form over the final 5 games (both in terms of points and in relation to other teams), did not significantly predict promotion (P value 0.710 and 0.379 respectively)
3. Form over the final 10 games (both in terms of points and in relation to other teams), also did not appear to significantly predict promotion
(P value 0.395 and 0.278 respectively)
The bottom line
In this study, between the years 2004 and 2022 in the Championship, the probability of promotion was significantly related to final league position of a team, but not form in the final 5 or 10 games of the season.
Suggestions for improvement
As above, it would be useful to stretch this back to '89 to get more numbers.
I like the idea of cross league analysis as well, it’s just arduous.
I'm sure there's better ways to analyse the data than what I've done, and I'm prepared to stand corrected, but the results are interesting IMO
FAO fans of other clubs, this isn't some way of attempting to cope with Boro's poor form, we all know the score and I'll be back here next year come rain or shine.
I was genuinely interested in having a go at answering the question. I wish all teams the best of luck this year (though obviously I wish Boro more luck than the rest)
UTB
Disclaimer
There’s obviously a chance I may have made a typo here or there, but I’ve triple checked the data, any mistakes (both in terms of data collection and analysis) are accidental so apologies if someone finds one.
I decided to have a look at some previous results and answer this question.
Let the long thread commence….
(Feel free to skip to the end)
I will apologise in advance to any mathematicians out there with regards to my statistical analysis. I realise I may not have chosen the best way to analyse results, and am very open to criticism or suggestion as to alternatives.
Many will be familiar with this website: Do form and position make the EFL Championship Playoff winner? Northwestern Sports Analytics Group
For those that aren't, it seems to be one of the only sources of information regarding this topic. The frustrating thing is it was published in 2018 and hasn't been updated since.
On the plus side It does include data from 1989-2017, and looks at chances of promotion via playoffs based on league position (across multiple English leagues), and chances of promotion via playoffs based on form in the 2nd tier.
The author’s conclusions were that:
1.League position
-Finishing higher in the league (across multiple tiers) is 'likely but not certain to be a factor in a team's playoff run'-
-The results for all leagues were almost (but not) statistically significant. The results for the Championship were even less so.
2. Form
-In the Championship, on average the team that won the playoffs scored 1.14 more points than the other playoff candidates in the 7 games leading up to the playoffs, however, this was also not statistically significant.
-The author suggested as much, but then went on to conclude that 'form probably correlates to winning the playoffs but we can't reach a conclusion'
(Form in this case was measured by taking the team with the best form over the last 7 games, and comparing this to the other 3 team's averages in the playoffs)
Overall they suggested teams should ideally enter the competition in good form and in a good finishing position.
Makes sense but I'm not sure it was quite backed up by the data.
I figured it was time to look at the years 2017-2022 as well.
Method
I started off gathering data with the intent of providing some sort of new insight or analysis on form specifically, given that Boro are in a relatively poor run of form and I wanted to see if that correlates to poor performance in playoffs or not
I looked at various websites and ultimately used a site called sportsmole:
Sportsmole example
I collected data on the previous 18 seasons, I picked this number because the website I used only recorded back to ’04-’05.
Records are available back to 89-90 (and these were used in the previous analysis)- but the websites I found that recorded the results made it a lot more difficult to read through.
I could do it but it would be more time consuming. - I am aware this automatically means I have less data than the website above. This thread isn't here with the intention of being the definitive/end product, more of a gentle probe.
I arranged the data into tables for each season, looking at final league position, points picked up in the last 5 games, last 10 games, and then used these points to rank teams based on form also.
Eg:
I did this for each of the 18 seasons, then compiled the data to make some charts.
I've shamelessly copied the display format of the data in these next two charts from the original website listed above.
RESULTS
A suggestion of a trend towards final league position and promotion, more on this later.
For the next chart I used a similar method from the original website. I took the amount of points earned by the promoted team in their final 5 and final 10 games (I called these 'form points'). I then subtracted the average points of the other 3 runner up teams for that season for each set.
IE: In the above table Norwich got 10 points in last 5, and 21 in last 10
That’s 0.67 points better than the average over 5 games and 3.67 better than the average over 10.
Plotting this on a graph, you can see on that in the 14-15 season, Norwich appear just over the 0 line for the 5 game form, and higher up for the 10 game form.
(The idea being that if form makes a big difference to the team that gets promoted, the teams that get promoted should persistently have more form points than the other 3 average score, and if that was the case the data points would mostly be above the line marked 0)
As you can see, it's a mixed bag, and certainly not a clear trend.
Honourable mentions for Palace (2012-2013)- Who got promoted despite having only 8 points from their final 10 games, and Huddersfield (2016-17)- Promoted wth only 4 points in their last 5, and 10 in their last 10.
Maybe there's hope for us yet....
Indeed, at least 5 teams have had similar (or worse) form points to us currently and still got promoted.
I did run T tests for the above data, and found no statistical significance between form points difference and promotion. (I have this data available if anyone wants)
Finally, I wanted to look closer at the relation between league position and form, and how this correlated with promotion or not (rather than just ripping off someone else's work)
I used a binomial logistic regression, which in this case should help predict whether a team is likely to get promotion or not based on different variables.
I put in data for all 72 teams over the 18 seasons, and looked at team league position, form across the final 5 games, and form across 10 games.
I started off analysing these variables in the same set of data, but then realised that the variables weren't necessarily independent of one another, so then analysed them separately. (Teams with good form in 5 would likely have good form in 10, and likely have a high league position- so it would be difficult to decide which one of these variables was related to promotion on it's own)
EG:
My findings from this were
1. Final league position did seem to correlate with chance of promotion (P value 0.034)
2. Form over the final 5 games (both in terms of points and in relation to other teams), did not significantly predict promotion (P value 0.710 and 0.379 respectively)
3. Form over the final 10 games (both in terms of points and in relation to other teams), also did not appear to significantly predict promotion
(P value 0.395 and 0.278 respectively)
The bottom line
In this study, between the years 2004 and 2022 in the Championship, the probability of promotion was significantly related to final league position of a team, but not form in the final 5 or 10 games of the season.
Suggestions for improvement
As above, it would be useful to stretch this back to '89 to get more numbers.
I like the idea of cross league analysis as well, it’s just arduous.
I'm sure there's better ways to analyse the data than what I've done, and I'm prepared to stand corrected, but the results are interesting IMO
FAO fans of other clubs, this isn't some way of attempting to cope with Boro's poor form, we all know the score and I'll be back here next year come rain or shine.
I was genuinely interested in having a go at answering the question. I wish all teams the best of luck this year (though obviously I wish Boro more luck than the rest)
UTB
Disclaimer
There’s obviously a chance I may have made a typo here or there, but I’ve triple checked the data, any mistakes (both in terms of data collection and analysis) are accidental so apologies if someone finds one.
)
