State of the League

[Stive Pesley]

On those numbers for promoted teams by position though:

1st: 17 - 40.2 - 59%

2nd: 17 - 40.9 - 65%

3rd: 7 - 39.0 - 71%

All 7 of those examples of 3rd being promoted made it through the playoffs, and with 71% survival they did remarkably well, even better than 2nd and 1st placed teams in that regard, and only just worse by points. Keep in mind that our 07-08 season is in those 3rd statistic there. Without that season it would read:

1st: 17 - 40.2 - 59%

2nd: 17 - 40.9 - 65%

3rd: 6 - 43.7 - 83%

 

Are the stats not skewed by the fact you have 17 instances of survival/non-survival for 1st & 2nd and only 7 instances for 3rd?

 

You'd also need to show the survival rates for those that get promoted from the "non-3rd" play-off spots - or only count the 1st/2nd survival rates from the seasons where the 3rd team make it through the Play-Offs

[Albert]

Are the stats not skewed by the fact you have 17 instances of survival/non-survival for 1st & 2nd and only 7 instances for 3rd?

 

You'd also need to show the survival rates for those that get promoted from the "non-3rd" play-off spots - or only count the 1st/2nd survival rates from the seasons where the 3rd team make it through the Play-Offs

The point about the sample size is actually a very important one. We can actually do some analysis with that though. What we're working with are averages here, and knowing the standard deviation of the values, the sample size and such we can actually find the standard error in the values, that is, the range over which it is reasonable to expect the true mean (for infinite data set size) to fall. For those sets, with error included:

1st: 17 - 40.2 ± 2.6 ­- 59% ± 12%

2nd: 17 - 40.9 ± 1.8 - 65% ± 12%

3rd: 7 - 39.0 ± 6.5 - 71% ± 18%

So there is less certainty, but to be honest, with only 17 values, I'd hardly call this a comprehensive survey. Add into that changes in how the leagues work since the mid 90s (although promotion and relegation worked the same way) and really, it's little more than a curiosity.

As another slight aside though, we can also consider the chance that the "true means" are actually higher than one or another, in this case the chances that 3rd's true means are actually above 1st and 2nd, and indeed, 2nd above first.

3rd being higher than 1st and second

1st - Av. Points: 43.3%

1st - Survival rate: 71.5%

2nd - Av. Points: 39.0%

2nd - Survival rate: 62.0%

2nd being higher than 1st

Av. Points: 58.7%

Survival rate: 63.4%

So it is more likely than not that what the data suggests is accurate, even if the data sizes are quite small. Even in the case that those rates for 3rd quite far out, there is still only around a 12.3% chance that the true survival rate is less than 50%, with a similar chance being for it to be actually be 93%. Long story short, even with a small data set, it's quite clear that 3rd's survival rate is higher than would be expected assuming it was just related to league performance in the previous season, and it seems that if anything those playoff games might actually help them (for third at least).

As for "non-3rd" promoted sides from the playoffs, those numbers were already listed on the last page, and it was clear that whilst they scored around the same points, their survival outcomes were less than idea, for the sake of completeness, here are those values again:

4th: 1 - 36 - 0% (not enough data for error analysis)

5th: 5 - 36.4 ±4.8 - 40% ±24%

6th: 4 - 40.0 ±5.2 - 25% ±25%

From all that it would be reasonable to conclude that the lower teams in the playoffs not only finish with less chance of winning the playoffs (41% ±12% of promoted sides by the playoffs are the 3rd place side, far higher than the 25% expected from it being merely random chance), but also have worse survival outcomes. The question here though was more to do with whether or not being forced to play those extra 3 matches, and use that time for the playoffs would have negatively affected the team heading into that 2007-08 season, and the result seem to suggest (at least the best that such a small data set could) that in fact teams that finished 3rd and won the playoffs actually performed better than teams finishing in the two automatic promotion spots. The survival rates for 4th-6th in such a discussion are irrelevant, particularly when the discussion started as being one of the merits of 3rd over them.

[Stive Pesley]

Excellent post - you are a bonafide stats nerd!

 

Interesting that only one team has ever won the play-offs from 4th!

[Albert]

Excellent post - you are a bonafide stats nerd!

 

Interesting that only one team has ever won the play-offs from 4th!

Keep in mind that's one back to the 95-96 season, the first in the current format. 94-95 was the last season of a 22 team Premier League, and only two teams were promoted from the old Division One; one automatic and one from a playoff system. The one from a playoff system that time was 3rd (this being a 2nd-5th playoff system, it would be like 4th in a 3rd-6th system), but it wasn't included as I thought that it was best to keep like for like stats, and that season being quite a change was a good cut off point. For those who are interested, the promoted teams from the playoffs by season:

~As Division One~

94-95: 3rd (1st automatic, 2nd-5th playoffs)

93-94: 4th (1st and 2nd Automatic, 3rd-6th playoffs)

92-93: 5th (1st and 2nd Automatic, 3rd-6th playoffs)

~As Division Two~

91-92: 6th (1st and 2nd Automatic, 3rd-6th playoffs)

90-91: 4th (1st, 2nd and 3rd Automatic, 4th-7th playoffs)

89-90: 4th, but awarded to 6th due to off field issues (1st and 2nd Automatic, 3rd-6th playoffs)

88-89: 3rd (1st and 2nd Automatic, 3rd-6th playoffs)

87-88: 3rd (1st and 2nd Automatic, 3rd-5th & 18th in Div 1 [22 teams] playoffs)

86-87: 18th in Div 1 (1st and 2nd Automatic, 3rd-5th & 18th in Div 1 [22 teams] playoffs)

85-86: No playoff system (1st, 2nd and 3rd Automatic)

So yeah, there have been more playoff seasons, but back to 95-96 is the total set for it being continuously the same. The total stats for promotion from the playoffs would read:

18th (Div 1): 0 / 1 [0%]

19th (Div 1): 1 / 1 [100%]

2nd: 0 / 1 [0%]

3rd: 10 / 26 [39%]

4th: 4 / 27 [15%]

5th: 7 / 27 [26]

6th: 5 / 24 [21%]

7th: 0 / 1 [0%]

Of course, since it has always been 4 teams you could also redefine it by the teams relative position in the playoffs, that is, for the current set up 3->1, 4->2, etc. This would allow for a general look at the performance for each team by position they were given for it. It seems that 18th in Div 1 was treated as the bottom team for these playoffs, and shall be treated as such. Keep in mind that the playoffs were at first used to determine if an additional 4th team would be relegated or stay in the First Division while it was being changed from 22 teams to 20 from 85-87. Anyhow, those numbers:

1st: 10 [37%]

2nd: 4 [15%]

3rd: 7 [26%]

4th: 6 [22%]

In any case, that 4th position in the Championship (2nd in the above) does seem cursed currently. The last team to win it was Charlton back in 97-98, some 16 seasons ago, and even taking into account playoffs during different circumstances (that is, with different numbers of teams promoted, different set ups for the playoffs etc.) they still only mustered a mere 4 promoted sides, one of which wasn't allowed to be promoted.

[PistoldPete2]

Excellent post - you are a bonafide stats nerd!

 

Interesting that only one team has ever won the play-offs from 4th!

Dont know about anyone else but it seems to make sense that Rams have about 30% chance of automatic and 90% chance of finishing in top 6. Not sure , if we lose to Burnley if that will change much, even tho we would be 5 points behind them. We will have played 5 of the top 8 teams twice so only have 2 more to play in the last 13. Our record against the lesser teams has been top notch so far.   

[PistoldPete2]

Thanks to Albert again.

My alternative measure also shwos Rams in good light. I look at where we wer at haf way, which was 44 points from 23 games, played everyone once.

Then compare the results we get in 2nd half to the correspondning fixtire in teh 1st half. Works quite well as our resulst are very similar so far. Beaten Brighton Yeovil, Wendies and Bournemouth home and away. Drew with Brum and Blackburn home and away. Losing to Wigan at home (won away) cancelled out by beating QPR at home (lost away) . So we are on target to match our 1st half of season performance and finish with 88 points from 46. That's enough for automatic in most years although maybe not this year.

Bad news is we now have 5 games coming up where we only got 4 points in the corresponding fixtures (lost to Burnley, Reading and Forest drew with Bolton beat Millwall) .

Good news is after that we have our last 9 games where we got 23 points in the corresponding earlier fixtures. Doubt we will match that this time, so we need to get more than 4 points from next 5 games.