3rd September 2019 at 10:47 pm #46987
As we all know, when it comes to ranking our teams each year in the league we all submit to the league table and the points system. 3 for a win, 1 for a draw, the team with the most points is the winner and declared the undisputed best. We all then repeat “the table never lies” and accept the outcome like good sports. If my team gets 50 points and yours gets 40, my team is better we all agree. But how much better? 10 points you may say, but is that really the best method of answering the question or understanding the differences in the strengths of the team? I dont think it is. I think the league table tells us who is best, but that there is a better measure for “how much better?”.
I think the better measure is to consider the goals a team scores over a season relative to goals conceded. The reason I think this is better is that the league awards 3 points whether you win 8-0 or 1-0, and so over a league season just how much better team A is than team B is not fully reflected in the points totals. By dividing the number of goals scored by the number conceded you get number that is fairly representative of the overall strength of a team relative to all its opposition that year. This then becomes a very useful tool for comparison as every other teams number is a measure of their strength relative to the same opposition.
Now I say strength, what I mean is how many goals you score for every one you concede. This is the measure because its about winning and that means scoring more than you concede. Its not just who scores the most as obviously defence is a vital strength too, its about balance.
When you look at these numbers for last years teams in order, you can see how intuitive and right it looks, and also how close it is to the league table order. Remember GF/GA:
1) Man city-4.13
6) Man utd-1.20
10) Crystal Palace-0.96
11) West Ham-0.95
I think this further illustrates 1)Just how far ahead Pool and City were last year 2) just how close the margins were.
Now when it comes to measuring the differences in the strength of teams I think you could literally factor the numbers above.
Ie, If we round up for convenience and take pool as 4 and wolves as 1, I think it means that over the 38 games liverpool were, in general, 4 times stonger/better than wolves.
It means imo both city and Pool were over twice as strong as their nearest rivals, spurs and chelsea. Again when I say strong I am referring to their ability to score more goals than they concede relative to all opposition over 38 games
Looking at how impressive City and Liverpools numbers are, 4.13 and 4.05 respectively, I wanted to look at this ratio for all the premier league champions past. Please note before reading this list that it is not a list of champions in order of objective strength, it couldnt be as they never played the same opposition, but it is a very good way of measuring who was the strongest champions relative to the opposition the year they won
2) Man city(18/19)-4.13
3) Man city(17/18)-3.92
4) Man utd(07/08)-3.64
7) Man City(11/12)-3.21
8) Man utd(06/07)-3.07
9) Man utd(08/09)-2.83
11) Man city(13/14)-2.76
13) Man utd(00/01)-2.55
16) Man utd(02/03-2.18
17) Man utd(98/99)-2.162
18) Man utd(92/93)-2.161
19) Man utd(99/00)-2.15
20) Man utd(10/11)-2.11
21) Man utd(93/94)-2.10
22) Man utd(95/96)-2.09
25) Man utd(12/13)-2.00
27) Man utd(96/97)-1.73
I find this an interesting list and hope some of you guys too. What strikes me is that in terms of superior strength relative to all opponents, the only champions to ever out do Liverpool’s runners up effort last year are Jose’s first title winning chelsea side, and last years champions city.
I’ll leave it there, hope some of you find it interesting reading.4th September 2019 at 6:55 am #46988
Going to disagree with this, cyber chum, as the problem with statistical arguments is they can be proven to be quite wrong with other statistical logic.
Not going to deny the Liberoools and City are somewhat in front of the rest of the pack, we can all see that, but calculating a superiority on GD alone is going to be misleading and not entirely accurate.
4th September 2019 at 7:04 am #46991steveosnakeeyeParticipant219 pts4th September 2019 at 7:27 am #46993Hightown hopeParticipant177 pts
- This reply was modified 4 months, 3 weeks ago by Chucky McChuckface.
The older amongst us here may remember that goal average rather than goal difference was used to separate teams on the same amount of points. I reckon it changed to goal difference in about 1973. I think I was too young to ever really understand how goal average was calculated, but my feeling it was something similar to what you did here Lucky.4th September 2019 at 9:12 am #46998
banjo, the point is to get feel of relative strength. We all agree with points totals as a ranking method but it can skewed too if you consider that with a bit of luck 1 point can turn in to 3 for a team that is being outplaye. This average is just another imperfect way of comparing strength. It correlates very well with league order for a reason, the higher the average, the better the team.4th September 2019 at 9:18 am #46999
hightown, i would prefer this ratio used than goal difference as secondary ranking to points. It may have been previously I will have look into it.4th September 2019 at 10:01 am #47000
yeah stevo, was bored last night, spent half hour on bit about last season but couldnt help looking back at past champs, that bit took a lot longer 🙂4th September 2019 at 10:07 am #47001
LD, don’t know about everybody else on here, but I get a pretty decent idea of relative strength watching Match of the Day!! 🙂
I mean, using your “formula”, Man City are actaully 4.3020833 times better than Crystal Palace, and I’m pretty sure Palace turned them over at the Council House!
I’m very bored today too… don’t tell the boss! 🙂4th September 2019 at 10:08 am #47002newbalanceSpectator190 pts
Goal difference is already merited.
I wouldn’t want anything else as a change thinking David v Goliath analogy/theory.
But, in the same token, there has to be some kind of rationale hitting a cricket score which is already in place.
It’s how Man City won their first PL Title back in 2012.4th September 2019 at 10:10 am #47003
banjo, if team A wins 10 games 4-1 and team B beats the same teams 3-2, intuitively we would deduce team A were superior but this would not be reflected in points as they would have 30 each. The GF/GA ratio does reflect it, team A 4.00 team B 1.50.4th September 2019 at 10:13 am #47004
Or another way of looking at it, Spuds were an injury time OG from Toby from drawing both games with Liberools last year, which would suggest Liberool’s aren’t 2.3546511 better than the Spuds, as both games between the two were closer than your formula would suggest (I’m also sure we hit the woodwork in injury time at Wembly too so the 4 points Liberool got from those two games could have very easily been 1).
Still very, very bored! 🙂4th September 2019 at 10:18 am #47005
LD, but the thing is, your looking at it in very black and white modes.
If team A wins 10 games 4-1, then it’s only a true comparison if Team B plays those exact same 10 teams under the exact some conditions (and by this I mean same time of year, same opposing 11, same build up time etc…)
If Team A beat Team C on opening day weekend with a fully fit squad etc.. and then Team B plays Team C in January, just after a Cup game and before a Euro games, and now with team C now dealing with injuries, suspensions, relegation, and maybe a change of manager, can you really compare Team A & B’s results? Two entirely different scenarios in my book. (Double dare you to work that out with all 380 Premiership games!! 🙂 )4th September 2019 at 10:22 am #47006
banjo, strength is measured over 38 games, there will always be one off games that defy the odds like palace v city, but you are not trying to suggest that one result for palace means city are not vastly superior are you?4th September 2019 at 10:25 am #47007
banjo, why respect the points totals even if you look at it like that? @10:18am4th September 2019 at 10:33 am #47008
also comparing results against one specific opponent is flawed, but what are the odds that all 10 teams were weaker against team A than B in that scenario? let alone all 38 games.4th September 2019 at 10:35 am #47009
You’re missing my point (and made my point for me at the same time!), cyber chum, you’re trying to convince us by the very simple virtue of goal difference that City are 4.3020833 times better than Palace, despite Palace beating them on their own patch.
This is why we play the game, not just give it to the team everybody thinks is most likely to win it in the end. But by trying to put an actual number on it, based on what I would suggest is a flawed formula, is probably not a good idea (especially when I’m bored at work!).
Yes, City are the best team in the land, yes, Liberoools are the 2nd best team in the land, but hell no am I going to agree they’re 4x better than everybody else. As I just mentioned, two very minor differences in the two games against Spuds were the difference between 1 and 4 points and the goal difference.4th September 2019 at 10:43 am #47010
No Im with you banjo. That this game allows for such upsets is what makes it so great, but it was an upset none the less. city are still vastly superior, and regardless of that result city will still be at least 1 to 4 on with bookies to beat palace when they next meet.4th September 2019 at 10:46 am #47011
Don’t get me started on Bookie’s odds… 🙂
(Which are more to do with their desire to separate punters from their cash than an actual “reflection” on the actual odds of a sporting event! Yes, there are similarities, but not a true sporting reflection!)4th September 2019 at 11:02 am #47015PaganParticipant282 pts
So if we were to implement this it would encourage teams to go out to annihilate the weaker opposition, this will force all teams to play their strongest team at all times, so where is the opportunity for the younger players to be introduced when games are comfortable, answer it goes, we’ve had pages and pages on here about youth development this idea would be another nail in the coffin….Pagan4th September 2019 at 11:08 am #47016
banjo, my degree is mathematics. at uni we developed model for calculating football odds and when you account for a bookie margin of approx 13% their odds very closely correlated with ours. we only considered objective stats not levels of money wagered. The GF/GA ratio was more sophisticated in the modelling and tried to factor out luck by looking at abilities to create clear chances relative to ability to concede few clear chances, bit like expected goals eg
- You must be logged in to reply to this topic.