After Bristol City

[Pubey]

I say sub appearances only count if for more than 23 minutes.

sounds reasonable to me.

 

[pringlejon]

Great work!
The only thing I would like to add is that I think its a little harsh to point out a 'worst player' from each game. We all know players read these sorts of forums and if one persons name pops up often it could be very disheartening for them and given that the majority of the team is new I wouldnt want a player to see that and feel unwelcome so soon..
Is it possible to have a most improved rating from the last game, or point out another positive from the game?

I agree in hindsight that most improved player would be a more positive spin on it. I'll sort that out.

If the same people award the marks every week it doesn't become an issue, but not every one will attend every game.
You want the index to reflect the best performers, not who played in games that were attended by the more generous markers.

With regards truncating the marks, you've already said that you've left out one for Zaaboub, because someone gave him a 0. I agree this mark shouldn't count, but I don't think it should be an arbitrary decision as to what is an outrageous mark or not. It's far better to apply a filter (FWIW a good turnout of raters won't cancel it out, but rather lessen the impact).

The other area worth considering IMO, is how to treat subs. Not everyone ranks subs, and if someone only plays 5 minutes they may only get a 6 which may reflect more that they didn't have time to get into the game than their performance during the game. Making lots of sub appearances is therefore likely to drag someone's average rating down so that someone who'd started 20 games and come on as sub in 20 games is likely to have a worse rating than someone who just started 20 games. I'm not sure how best to deal with this, we could exclude sub appearances unless someone played for at least half a half (23 minutes) or just ignore them completely. Or have two overall ratings one including sub appearances and one excluding them.

I think this is a cracking initiative, so it would be a shame if a close-run battle was decided by a statistical anamoly which could easily be eradicated.

I say sub appearances only count if for more than 23 minutes.

The only issue with the above is if a super sub comes on for 15 minutes and wins the game for us, he deserves a great rating. Something needs to be sorted but i dont know the best way as i agree players only on for a short spell do often get rated less.

 

[EastStandBlue]

The only issue with the above is if a super sub comes on for 15 minutes and wins the game for us, he deserves a great rating. Something needs to be sorted but i dont know the best way as i agree players only on for a short spell do often get rated less.

This would never have been a problem under Tilson...

*runs for cover*

 

[OldBlueLady]

It's fantastic stuff just not too sure about showing the worst rated player - bit demoralising maybe when it gets back to them as it inevitably will...at some point, when all the newbies realise what a vehicle SZ is! ;)

 

[pringlejon]

Updated with latest suggestions:

• Appearances now in brackets beside player
• Worst player replaced by Most Improved
• Arrows indicate player performing better relative to other players

have yet to discount extreme values as am figuring out the simplest way to do this without overcomplicating it for myself.

 

[Yorkshire Blue]

have yet to discount extreme values as am figuring out the simplest way to do this without overcomplicating it for myself.

Are you using excel?

If so, it should be easy enough for someone to come up with a formula for you.

 

[BILLERICAY BLUE]

When you say most improved, is that his mark for that game compared to his season's average?

 

[pringlejon]

Are you using excel?

If so, it should be easy enough for someone to come up with a formula for you.

Yea i am, will look into it later unless someone here is a whizz at VB.

 

[Pubey]

Yea i am, will look into it later unless someone here is a whizz at VB.

there's a number of different ways you could do it.

You could estimate the mean, find the values that are a certain amount away from the mean (2xStdDev?), omit and then recalculate.

You could sort the data and then calculate the mean of the middle 90% in the array.

Personally though I don't think it's that big of a deal. Maybe something to look into for next season though.

 

[pringlejon]

When you say most improved, is that his mark for that game compared to his season's average?

Dont know whether to do compared to last game or season average. At the min, there is no difference since we only have 2 games, but need to decide before next game.

And cheers for suggestions pubey, will consider, but like you say (aaaaargh tilly has ruined my vocabulary) i wont implement any truncating unless it is really noticeable with external outliers.

 

[South Bank Hank]

Fantastic work on this PJ. I'm sure everyone is going to enjoy seeing how this pans out across the season. :clap:

 

[The Blue Stig]

Got to say very well done there fella like i said i would have loved to of done this but have limited knowledge of computers an accessing one most of the time, it will be good to watch an i think it is working really well, nice one

 

[BluesMike61]

Just loving the Pringello index. After Fabio has opted out of rating footballers (and
,allegedly,opting out of having any passion for managing England),this is just what
we need.

 

[manor15]

Maybe the most improved could be the result of the latest game compared against the average of the last 3 games before that?

 

[shrimperman]

It will be interesting to see how this reflects on our choice of Players of Season.

 

[matt1]

Yea i am, will look into it later unless someone here is a whizz at VB.

If you are just looking to remove the highest value and the lowest.
You could use the formula =(sum(range)-min(range)-max(range))/(count(range)-2))