Team ratings...onto OT football/soccer discussion

Darkson · Post by **Darkson** » Mon Sep 16, 2002 8:28 pm

Zombie wrote:The current team rating has one and only one use: to determine the handicap. You can say otherwise, but you're only fooling yourself.

interesting thread, this.

I have to agree the TR is only for determning handicap. Our groups Undead team had the 2nd lowest TR, but came 2nd overall. He did this by using money to freeboot Count Von Drakenborg when he could, and any player he "raised the dead" he turned into a asst. coach. This way he kept his TR low so got the handicaps, but had 13-14 skillful players.

I'd like to see a more acurate system, but onlt if it doesn't mean filling out pages of calculations after every game. For all it's faults, TR is at least simple to work out.

Ghost of Pariah · Post by **Ghost of Pariah** » Mon Sep 16, 2002 9:22 pm

Yes I do understand.
That's why I say it's impossible to have system that predicts a winner.

And it's also why I keep asking you what is a WIN.

You need to define what it is you are predicting.

Zombie · Post by **Zombie** » Mon Sep 16, 2002 10:29 pm

A win is when you score more TDs than the opponent. Plain and simple.

Ghost of Pariah · Post by **Ghost of Pariah** » Tue Sep 17, 2002 4:32 am

Is it? So if you out score your opponent but you can't even field 6 players for the next match you still won?

Mestari · Post by **Mestari** » Tue Sep 17, 2002 5:23 am

Do you record such a game as a loss to your teams game record sheet? No. You record it as a victory, Pariah, so your concentration on that thing seems rather pointless.

Pariah:
The point is not to predict the winner, but the odds of winning, as can be seen from Zombies example.

And the point in my last post still stands that there is a point for trying to predict the odds for both teams victory.

As the ideal odds do "exist" whether we calculate them or not. When we set up for the game, there is a x% chance that I win, and y% chance that you win (prob'ly x>>y

), whether we calculate it or not. If we manage to create a system that quite accurately predicts these odds, I say: great! Now we can tell, if we win, that it was really a predictable, easy victory, or that it was a struggle against nearly impossible odds. On the other hand, if we lose, we can say that well, it was 95% likely, or that damn, it was a one out of million chance that I lost, but it just happened.

I just can't see the point of not being interested in developing a system that produces numbers that can be used to predict the outcome of a game.

It can also be used in the handicap system, naturally, far more fairly.

neoliminal · Post by **neoliminal** » Tue Sep 17, 2002 1:32 pm

Pariah wrote:
Actually, you don't have to.. Given enough coaches and a way to rank them
Okay, what coaches were used, how were they ranked and who did it?
As far as I know there was less than 5 test leagues for the BBRC so I doubt Neo has alot of data to back anything up.

You having a bad week Pariah?

LOL

I used the data I had available. I'm sure we'll revisit the numbers in a year after NAF gathers tons of it. These were the best we could get and their a damn sight better than numbers you could pull from your arse.

Ghost of Pariah · Post by **Ghost of Pariah** » Tue Sep 17, 2002 3:45 pm

I'm not bashing your methods Neo. I think you did fine but what I'm saying is that the data I assume you have is not from a large number of coaches or leagues. So to say that your number is accurate is short sighted.
Hey, he needs a number. You gave him a number. What I'm saying is that he is going to canonize that number and go on about how inaccurate the TR system is.

Fine Mestari, you can focus on the impossible task of predicting winners in individual games. I would rather know my team's potential in the race for the cup.

Show me the data that proves Zombie's system predicts winners any better than the current system. It did not in our league and it was more work so we scrapped it.

Marcus · Post by **Marcus** » Tue Sep 17, 2002 4:06 pm

Interesting article in a recent new scientist mentioned discussed probabilities in football, mentioning there were fewer certainties because it was such a low scoring game. It also illustrated that even a slight difference in skill could generate a large winning percentage.

Similar observations could be made about BB, given the high randomness factor.

Couldn't find the article online but it's probably worth digging for it.

Post by **DoubleSkulls** » Tue Sep 17, 2002 4:20 pm

Marcus wrote:Interesting article in a recent new scientist mentioned discussed probabilities in football, mentioning there were fewer certainties because it was such a low scoring game. It also illustrated that even a slight difference in skill could generate a large winning percentage.

Are these two related? If so how?

I would have thought that because football was such a low scoring game (~3 goals per game) that it increased the uncertainty and made larger differences in ability necessary to get increased win %'s. Even Man U can't guarantee beating Bolton bloody Wanderers after all.

Ian

Marcus · Post by **Marcus** » Tue Sep 17, 2002 4:24 pm

I believe it made the point that when uncertainty is essentially a constant, differences in skill become more noticable over a league season, but for any one game, any team could win.

I'll have a reread and lend you the mag.

neoliminal · Post by **neoliminal** » Tue Sep 17, 2002 4:32 pm

Pariah wrote:Hey, he needs a number. You gave him a number. What I'm saying is that he is going to canonize that number and go on about how inaccurate the TR system is.

Actually, if you re-read his initial post:

"Obviously, i'm only looking for rough estimates "

You'll see that the number is not going to be canonized.

If anyone else wants to talk about how inaccurate TR is, I have more than enough ammo for them. It can't be accurate. It's a guide. It's likely that a team will win if it's 100TR above it's opponent, for example, but that's not a set result.

Norse · Post by **Norse** » Tue Sep 17, 2002 4:43 pm

My only problem with TR is that if a team has 4 or 5 players with NIs, of which perhaps 2 fail to turn up, their value is still calculated in the overall TR..

In one recent game v Lucy, 2 NIs did not turn up plus I managed to have his star player arrested on the handicap table. By the time the two teams faced each other, I had more SPPs than he did...

I still lost though...

Zombie · Post by **Zombie** » Tue Sep 17, 2002 4:50 pm

Marcus wrote:Interesting article in a recent new scientist mentioned discussed probabilities in football, mentioning there were fewer certainties because it was such a low scoring game.

I assume you're talking about soccer. Football is anything but a low scoring game!

Deathwing · Post by **Deathwing** » Tue Sep 17, 2002 11:13 pm

Zombie wrote:
I assume you're talking about soccer. Football is anything but a low scoring game!

Oh dear..here we go again....I'll post the link to the relevent thread if it's still there...otherwise; ready to do this again Lucien?

In the meanwhile I'll bite my tongue....

Zombie · Post by **Zombie** » Wed Sep 18, 2002 5:47 am

Here's what i think. You come to my place, you call it soccer. I go to yours, i call it football. As this is the internet, we're not at anyone's place. Also, the population of the forum is about 50-50 Europe and North America (plus the occasional Australian and such), so we can't use that as an argument either.

Because this is a forum about Blood Bowl, a game copied on American football (i say American in opposition to Canadian football, i couldn't care less about soccer), i think that sport is very much in the theme, and should win over soccer. So let's call it soccer while we're here, ok?