| Game 25 - Looking at the stats for survival 14:36 - Feb 1 with 10424 views | Jack_Kass |
Following on from the previous thread, looking at the stats for survival -
Bottom at Xmas/Game 19 - Looking at the stats for survival by Jack_Kass 23 Dec 2017 19:55A lot is made of the ‘bottom at Xmas tag’, and statistically it can be seen as the kiss of death, with only 12% of teams having avoided relegation from this position*, with the three survivors being (WBA (2005), Sunderland (2014), Leicester (2015).
But, the asterix next to that statistic, is Swansea City (2016), as we were effectively joint bottom, and level on points with Hull City on Xmas day last year, and as we all knew, survived! So to flip that statistic in a positive manner, the past 3 consecutive seasons have seen the team/s with the lowest points on Xmas day survive, here’s hoping for for a 4th!
But, what fails to be mentioned alongside the ‘xmas curse’, is that Xmas Day can fall at different points in the season, in terms of games played, e.g in 2016 it was Round 17, with 19 not coming until NYD, and this year Game 19 has already been played, a whole 9 days earlier than last year.
So let’s look at Game 19, statistically, as a benchmark for the season, rather than the 25th of December. Game 19 being the half way mark, and theoretically, meaning we have played every team in the league (but not always strictly true).
Below is a summary of 18th position (highest relegation spot) at Game 19, over the last 10 seasons, with their points total, and their points per game average, with the credentials of the team who were relegated that season in 18th position, to the right, There is also the difference between the two avg points totals.
http://i66.tinypic.com/2jd16hc.jpg
So with Game 19 already played in the 2017/2018 season, we now have the data available to make the prediction for the points total needed to survive relegation this year.
The avg points needed for the team in 18th to survive, from this point, is 19. It has been as high as 23 (2011), and as low as 12 (2010).
The avg rise in points per game for 18th position, from Game 19 to 38, is 0.07, predicting that the relegated team this year, will have 34.58 points (35). Although it has been as high as 0.21 (2008), and as low as -0.16 (2009,2010)
Using the data, we can predict that the magic number to survive this season will be 35 pts. Incidentally, 35 is also the exact average pts needed to survive over the last 10 years (perfect!)
So what does this leave Swansea City needing to do, in the second half of the season, to reach the total of 35 points?
In 2016/2017, Swansea City had 13 points from 19 games.
In 2017/2018, Swansea City have.. 13 points from 19 games.
Swansea City need a minimum of 22 points in the next 19 games, to meet our magic number and survive, 1 point more than we would have needed to survive, last season.
Last year, we amassed a total of 29 points over the second half of the season! An average of 1.53 pts per game. That will certainly do again this year, can we repeat it?
Swansea City will need to improve their average points total by 0.48 points per game to reach this total, last season they improved their second half total by 0.90 points per game. An improvement of 0.48 points looks relatively easy!
Statistically, the club is in an incredibly similar position to the one they faced last season, with 13 pts from 19 games. They have scored less goals, but also conceded less, with a goal difference today of -15, compared to -23, this time last year.
Not rocket science to any fan, we need points, and soon.
The key notable difference for me, from this time last year, to now, being..
Last year we were at 19 games played when Clement came in, and rocketed us up the table, thereafter.
This year, we are already at 19 games played, with seemingly no idea of who is going to be in charge, or in how many games time?
Will we have a team/manager in place, over the next few months, to repeat the scenario of last year?
We did it just one year ago, can we do it again? Does lightning strike twice? Let's all hope so.
Updated to Game 25
Observations since Game 20 -
At game 20, the 18th placed team had 17 pts, and an 0.85 points per game average, since then we have seen the same position in the table climb to 23 points, meaning a 0.92 points per game average. A 0.07 point increase in that time.
With this increase of ppg around the bottom of the table in the last 5 games, we now have a new projected safety total of 36 points.
With 13 games left, the average points needed (over 10 years) from this stage is 13, giving us our new total of 36.
The average rise in points per game, is 0.03, again which means that adding to Southampton's current total of 0.92, gives us a figure of 36.
In short, as long as the team in 18th averages 1 point a week increase, then 36 will be the minimum needed, at this moment in time.
Looking ahead to Game 30 -
For the projected total to go above or below 36, we would need to see the 18th placed team achieve the following totals by Game 30.
If we see the same increase in ppg as we did over the last 5 game weeks, then 18th place will have 30 points at Game 30, meaning a safety target of 38 points.
Could it happen? You never know in football, but I think with the league being so congested at the bottom, and teams inevitably taking points away from one another, I think 36 is still more than sound total for us to be aiming for
4 wins from 13 will do it! |  |
| |  |
| Game 25 - Looking at the stats for survival on 14:53 - Feb 1 with 8052 views | E20Jack |
Trying to work out a total based on an average figure doesn't really make much sense.
The very definition of average means that there have also been totals higher and lower than the one you have set. So to aim for that as a safety mark is useless.
It would make far more sense to look at worst case scenario as a figure to aim for safety with, not one that by its very drfinition does not represent safety at all. [Post edited 1 Feb 2018 14:58]
|  |
| |
| Game 25 - Looking at the stats for survival on 15:06 - Feb 1 with 7994 views | Jack_Kass |
| Game 25 - Looking at the stats for survival on 14:53 - Feb 1 by E20Jack |
Trying to work out a total based on an average figure doesn't really make much sense.
The very definition of average means that there have also been totals higher and lower than the one you have set. So to aim for that as a safety mark is useless.
It would make far more sense to look at worst case scenario as a figure to aim for safety with, not one that by its very drfinition does not represent safety at all. [Post edited 1 Feb 2018 14:58]
|
A number expressing the central or typical value in a set of data.
At least you've learned something today. | |
| |
| Game 25 - Looking at the stats for survival on 15:06 - Feb 1 with 7993 views | Professor |
| Game 25 - Looking at the stats for survival on 14:53 - Feb 1 by E20Jack |
Trying to work out a total based on an average figure doesn't really make much sense.
The very definition of average means that there have also been totals higher and lower than the one you have set. So to aim for that as a safety mark is useless.
It would make far more sense to look at worst case scenario as a figure to aim for safety with, not one that by its very drfinition does not represent safety at all. [Post edited 1 Feb 2018 14:58]
|
You are right, of course, but five wins or the equivalent from the remaining games (38 points) will probably be comfortable. An advantage is that there are still arguably 10 teams fighting relegation. Maintaining momentum (which I think is not just the 'dead cat' bounce now) is a key. And yes, I think we could win 5 games. Shame we don't get Watford again too. | | | |
| Game 25 - Looking at the stats for survival on 15:07 - Feb 1 with 7990 views | Professor |
| Game 25 - Looking at the stats for survival on 15:06 - Feb 1 by Jack_Kass |
A number expressing the central or typical value in a set of data.
At least you've learned something today. |
That would be more median or mode than the mean which I take it you are using? | | | |
| Game 25 - Looking at the stats for survival on 15:13 - Feb 1 with 7971 views | Jack_Kass |
| Game 25 - Looking at the stats for survival on 15:07 - Feb 1 by Professor |
That would be more median or mode than the mean which I take it you are using? |
The mean yes.
I haven't looked at using median or mode, in comparison, but I certainly will. | |
| |
| Game 25 - Looking at the stats for survival on 15:14 - Feb 1 with 7965 views | Swanjaxs |
We have to remember that all the teams around us will think four wins will be enough, i think for the first time in a number of years 40 points will be needed, maybe more ..... |  |
| |
| Game 25 - Looking at the stats for survival on 15:19 - Feb 1 with 7943 views | Jack_Kass |
| Game 25 - Looking at the stats for survival on 15:14 - Feb 1 by Swanjaxs |
We have to remember that all the teams around us will think four wins will be enough, i think for the first time in a number of years 40 points will be needed, maybe more ..... |
It can look like that on the surface
but we're forgetting that for every team that need 4 wins, they also play each other, making it almost impossible for them all to achieve it.
The last time it happened, with West Ham, they were on 20 points after 27 games, and went on a massive run, getting 22 points from 11 games. | |
| |
| Game 25 - Looking at the stats for survival on 15:22 - Feb 1 with 7930 views | Al_Bundy |
I agree with 40 points. If the bottom teams keep taking points off the top 10 sides 36 will not be enough. | | | |
Login to get fewer ads
| Game 25 - Looking at the stats for survival on 15:32 - Feb 1 with 7891 views | Swanjaxs |
| Game 25 - Looking at the stats for survival on 15:22 - Feb 1 by Al_Bundy |
I agree with 40 points. If the bottom teams keep taking points off the top 10 sides 36 will not be enough. |
Most definitely, with the exception of the top four or mabybe five the teams around us will regard the rest as fair game.... hold on to your hats lads 👠[Post edited 1 Feb 2018 15:34]
| |
| |
| Game 25 - Looking at the stats for survival on 16:10 - Feb 1 with 7791 views | swanforthemoney |
Thanks for the detailed analysis Jack. Interesting to see how it pans out. Last year I look ed at points totals of teams finishing 17 and 18th and came to the conclusion that except in exceptional seasons, 37 is enough to survive. Exceptional season would be where the bottom 2 get a very low points total, or where the top 4 or 5 don’t get really good points totals . In other words this season is not exceptional an 37 will be enough. I agree with your assertion that 4 wins (and a couple of draws will be enough. ) |  |
| I stand in the North Stand
|
| |
| Game 25 - Looking at the stats for survival on 16:17 - Feb 1 with 7774 views | Jango |
40 points will probably be enough to get around nd 14th/15th place this year. 35/36 should get survival. It’s closer at the bottom this year than it was last year and 35 was enough. | | | |
| Game 25 - Looking at the stats for survival on 16:49 - Feb 1 with 7710 views | E20Jack |
| Game 25 - Looking at the stats for survival on 15:06 - Feb 1 by Jack_Kass |
A number expressing the central or typical value in a set of data.
At least you've learned something today. |
and hopefully you have learned that using average data to set a safety mark is ridiculous.  | |
| |
| Game 25 - Looking at the stats for survival on 17:02 - Feb 1 with 7683 views | Jack11 |
You need to remove any outliers to ensure that your projections are accurate when you are working on mean. You need to employ standard deviation as well. | | | |
| Game 25 - Looking at the stats for survival on 17:04 - Feb 1 with 7671 views | E20Jack |
| Game 25 - Looking at the stats for survival on 17:02 - Feb 1 by Jack11 |
You need to remove any outliers to ensure that your projections are accurate when you are working on mean. You need to employ standard deviation as well. |
Don't confuse him, you naughty boy.  | |
| |
| Game 25 - Looking at the stats for survival on 17:04 - Feb 1 with 7672 views | Jack_Kass |
| Game 25 - Looking at the stats for survival on 16:49 - Feb 1 by E20Jack |
and hopefully you have learned that using average data to set a safety mark is ridiculous. |
What would you use?
Enlighten us.
Taken what the Prof said into consideration, in regards to final points, there is no difference in the median or mode in regards to the final calculation, as there is no real outliers in regards to either.
Only when we use the Median for ppg rise, do we see any difference, presenting a rise of 0.06 ppg (37), over 0.03 (36), which perhaps is a better calculation, as we do have arguable outliers, with ranges from -0.13, to 0.11
Imo, the average in points per game rise from one point (25), to another (38), gives us the best idea of the trends from this position.
If we wanted to use the worst case scenario, we would just say '42' points will get us safety every year, boring and plain obvious , why don't we just say that 'if we win every game, we'll guarantee safety'. I think most know that. | |
| |
| Game 25 - Looking at the stats for survival on 17:06 - Feb 1 with 7667 views | Jack_Kass |
| Game 25 - Looking at the stats for survival on 17:02 - Feb 1 by Jack11 |
You need to remove any outliers to ensure that your projections are accurate when you are working on mean. You need to employ standard deviation as well. |
Haha granted, I am not presenting this to the ONS, just the good people of PS.
Do you think standard deviation, would change the results significantly? [Post edited 1 Feb 2018 17:06]
| |
| |
| Game 25 - Looking at the stats for survival on 17:10 - Feb 1 with 7651 views | Jack11 |
| Game 25 - Looking at the stats for survival on 17:06 - Feb 1 by Jack_Kass |
Haha granted, I am not presenting this to the ONS, just the good people of PS.
Do you think standard deviation, would change the results significantly? [Post edited 1 Feb 2018 17:06]
|
I know mate, was pulling your leg as much as anything. But yes I think the standard deviation would change the results. You can also employ the chi square to a data set such as this to see if your hypothesis stands up. | | | |
| Game 25 - Looking at the stats for survival on 17:11 - Feb 1 with 7644 views | perchrockjack |
Stats eh
No way will Stoke finish bottom
I predict we could stay up along with West
It's gonna go right down to the last day of the season | |
| |
| Game 25 - Looking at the stats for survival on 17:21 - Feb 1 with 7617 views | E20Jack |
| Game 25 - Looking at the stats for survival on 17:04 - Feb 1 by Jack_Kass |
What would you use?
Enlighten us.
Taken what the Prof said into consideration, in regards to final points, there is no difference in the median or mode in regards to the final calculation, as there is no real outliers in regards to either.
Only when we use the Median for ppg rise, do we see any difference, presenting a rise of 0.06 ppg (37), over 0.03 (36), which perhaps is a better calculation, as we do have arguable outliers, with ranges from -0.13, to 0.11
Imo, the average in points per game rise from one point (25), to another (38), gives us the best idea of the trends from this position.
If we wanted to use the worst case scenario, we would just say '42' points will get us safety every year, boring and plain obvious , why don't we just say that 'if we win every game, we'll guarantee safety'. I think most know that. |
Well, in short, I wouldn't.
I understand using 42pts is plain, boring, obvious - but it is also historically accurate which is surely the most important thing here?
If I was to make a points projection to which I then claim "x more wins WILL see us safe" then I would certainly never use a figure that does not represent that, be that historically or otherwise.
So it depends on what your graph is trying to say really. If you are looking for a graph to show the minimum needed from here on in historically then that has some value. If you just want a graph to show what "may" make us safe then yours is perfectly adequate, although I am not entirely sure of its value. | |
| |
| Game 25 - Looking at the stats for survival on 17:42 - Feb 1 with 7552 views | awayjack |
4 wins from 13 games sounds achievable if we keep momentum going. |  | | |
| Game 25 - Looking at the stats for survival on 17:51 - Feb 1 with 7539 views | Professor |
| Game 25 - Looking at the stats for survival on 17:04 - Feb 1 by Jack_Kass |
What would you use?
Enlighten us.
Taken what the Prof said into consideration, in regards to final points, there is no difference in the median or mode in regards to the final calculation, as there is no real outliers in regards to either.
Only when we use the Median for ppg rise, do we see any difference, presenting a rise of 0.06 ppg (37), over 0.03 (36), which perhaps is a better calculation, as we do have arguable outliers, with ranges from -0.13, to 0.11
Imo, the average in points per game rise from one point (25), to another (38), gives us the best idea of the trends from this position.
If we wanted to use the worst case scenario, we would just say '42' points will get us safety every year, boring and plain obvious , why don't we just say that 'if we win every game, we'll guarantee safety'. I think most know that. |
I don't think the mode would be any use really-other than an idea of 'frequency of points needed' but the median is an interesting one and can be of more value when the data distribution is not 'normal'. So in terms of any outliers e.g.West Ham going down on over 40 points or staying up on 33, they would be taken into the value. I guess the difference may be an arithmetic mean of survival points is decreased in recent years, but the median may remain higher given it is based more on rank than the actual number. We have a lot of data where there are not normal distribution and so use medians, ranges and non-parametric tests more frequently than mean and t tests or ANOVA.
That said it is an unpredictable season. Although lower placed teams like us are beating the big boys, we all have to play each other too (though not Palace, Watford and Newcastle in our case). What may be to our advantage is that those around use were probably looking at us as a home banker or likely away points. My feeling is 36 or 37 may be enough. At the moment Soton, Huddersfield and Watford can't buy a win. Just get the feeling that Saints have sold the family silver once too often and Watford's revolving door is coming unhinged. It may go all the way but feel that us, WBA and Stoke have a enough nous and experience to stay up. Along with Newcastle and Brighton, the three aforementioned teams do look to be running on fumes to an extent. Only time can tell. | | | |
| Game 25 - Looking at the stats for survival on 18:41 - Feb 1 with 7475 views | Glyn1 |
| Game 25 - Looking at the stats for survival on 17:42 - Feb 1 by awayjack |
4 wins from 13 games sounds achievable if we keep momentum going. |
At the moment, any team above 30 points is probably already safe, and no team has yet been cut adrift (we almost were). It's going to go into the run-in isn't it?
The final day's fixture list
Burnley v Bournemouth
Crystal Palace v West Brom
Huddersfield v Arsenal
Liverpool v Brighton
Man Utd v Watford
Newcastle v Chelsea
Southampton v Man City
Swansea v Stoke
Tottenham v Leicester
West Ham v Everton |  |
| |
| Game 25 - Looking at the stats for survival on 20:21 - Feb 1 with 7377 views | plasjack |
Lets get this straight if we don't win 4 games and a few draws we will be relegated, I could have predicted that, without going into mathematical theories. |  | | |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
