BAP league discussion

[SaintM]

I called the long con, early. It doesn't work anyway Mike.

/fluff

That final list includes season 11 data.

I'll be removing that from the calcs I do in future. I knew it was fucked up, but that graph sorta highlighted by just how much to me.

[SaintM]

Any team gonna be available for a warm-up game at 9.45 tonight?

[Wincho]

United will be about

[SaintM]

Coolio. 9.45 on the dot ok? (so we can fit a full game in)

[Wincho]

No problem

[Wincho]

Is any team not now a default 10.15 start?

Always need to remember to ask for a 10.15 start but it'd save the bother if that just became the default starting time

[SaintM]

We're 10.15 by standard now to give us time to get a warm up game in, so would be ok with it being assumed.

[SaintM]

As previously mentioned, I'm gonna add Team of the Week recording by the end of the season.

In addition to that I'm gonna change how the Team of Legends is calculated. It's currently based on avg rating in a position for all players who have played over 40 times.

The problem with that it twofold. First off, 40 games in a position isn't really that many given the number of seasons we've now reached, and the more you play the more the chance that your avg will be lower (eg two great seasons is given higher priority than two great seasons and one medium). Secondly, the way the ratings have worked have varied greatly over the years, which means players who were active in some seasons were penalised or given a boost compared to others.

The way around that will be to change it so it just looks at the number of appearances in the team of the week. It removes the rating issue and means that it's not got a fixed amount of games that starts to look a bit silly as the seasons go on.

That's me plan. Feel free to feedback or ignore.

Just to update, I'm gonna go ahead and implement the change to the team of legends calc at the end of this season.

Players eligible for a position will have to had played at least 40 times in that position. It's then sorted by percentage of those games they've appeared in the team of the week.

I'll also be excluding season 11 data from the calc for reasons I've already gone into.

Though it has no real-league-life impact on things, I feel this is a much fairer way of calculating it than the previous straight rating average (eg, any defender who played during the FIFA10 years will have a very hard time undoing those 2 ratings compared to defenders who didn't play then).

If anyone has concerns that this isn't as fair as the existing system, I'm happy to discuss as it could well be I'm not seeing the wood for the trees.

[dan_tm]

Just noticed the 'race' to reach 150 all-time goals is hotting up: http://www.rllmukfootballleague.co.uk/bap/bapstats.asp?tab=3

[Baring]

I scored last night as well but then the game needed to restart and was taken away lol Tyler will get there first as I probably won't be here next couple weeks, fair play to him, awesome player

[Wincho]

Average

[Manic Minor]

That's the apps by team and season.

Season 11 was obv the one with the FIFA11 ratings and all their glorious fuck'dupness

Been away from the forum for a couple of days, couldn't possibly read all of it (waaaay too dull) but this graph did particularly catch my eye, and make me laugh.

Anyone?

[Timmo]

What a run in. One point to seperate the top three, although Athletic's huge goal difference effectively gives them an extra point to put them two ahead.

However Town on paper have the easiest run in, facing the bottom two clubs before a showdown with Athletic who have to face Wanderers and United first. United are somewhere inbetween and must beat Athletic when they face them in two weeks time to stand a chance.

Completely wide open.

[SaintM]

Gonna be tight.

(this ones for you manic)

[tyler]

Just Town to complete the match reporting this week.

[SomethingWitty]

Gonna be tight.

Using my mathematical and statistical expertise, I have conducted lengthy analysis of the graph above, with the aim of predicting the final standings to within a 0.372% confidence level.

I started with a basic linear exptrapolation algorithm, as below:

This, however, proved inconclusive, as the nature of the population was such that simple linear expansion could not do justice to the complexity of the problem. I then turned to a skewed variant of French Curve extrapolation, to take into account the many and varied factors of decay and growth amongst the constituent curves. This was where I had my breakthrough – adding a further layer of sample points using a simple cubic spline interpolation as a pre-pass, before applying the French Curve to the expanded set. Any good statistician will tell you that extrapolation can never be truly 100% reliable, but with the coefficients applied here, I think I have come as close as science, as we currently know it, allows.

So, ladies and gentlemen, without further ado I present to you, the definitive destination of this season’s title:

[hannay]

[sladie]

:lol:

[SaintM]

[pompey88]

Just Town to complete the match reporting this week.

Done, it's a Wanderers and Athletic TOTW.

[SaintM]

Just to dip back into the discussion about how the team of the week works with the ratings etc (this is the cue for those prone to depression to leave), the main reason why the winning teams do seem to get placement in the totw over others is that it looks at the week as a whole, rather than individual games.

If you have a great match then a stinker, you're rated for both together.

If you only play one game in a position, you're heavily penalised for it.

Changing it to look at games in isolation isn't out of the question.

For example, this week's totw is:

If I just looked at individual reports instead, it would be:

LB - Mackenie - RPL Wanderers

CB - Rogue Soul - RPL Wanderers

CB - RickyDVT - RPL Town

RB - JaminTastic - RPL Athletic

LM - scooot77 - RPL Athletic

CM - Pilotwings - RPL Wanderers

CM - sixbuttons - RPL Wanderers

RM - Jazzy - RPL Athletic

AT - dan_tm RPL - Wanderers

AT - LittleJoe - RPL United

Not a massive difference, but significant. Two other teams get a look in for a start.

[tyler]

Now do that for S1-11 and get back to us.

[SaintM]

Hmm..

I may do that...

(assuming I can ensure I end up on top of course)

[dan_tm]

The problem with doing that is an increased chance of two people having perfect 10s competing for the same position in the TotW. Although I presume it wouldn't be too hard to revert back to an average over two games if two players are tied originally.

[SaintM]

It currently orders by rating, then goals, then assists. Maybe adding another sort of points won would help split things up in the odd times where the existing onces can't make the split.

[dan_tm]

What's the current penalty for only playing one game by the way, -2?

[SaintM]

Aye

[sith]

Mike don't suppose you could knock up a player rankings that would be up to date?

[SaintM]

I'm not at home so can't do anything, but I understand the desire from wincho was that the positional splits were dumped, so the website should tell you what you need to know (though keep in mind that the ratings for the season may well change when any final balancing is applied at the end of the season).

[SaintM]

On that subject actually, the ratings will probably need a secondary ordering. They are quite tight, so there are some players on the same amounts.