Thursday, January 19, 2012

Gameweek 22 Captain Data

Something which has bothered me this year are the weekly captain rankings, as I find they generally descend into the standard conjecture you find on other fantasy sites. It's not that there's anything wrong with that, it's just it is freely available elsewhere and I don't think you get much out of reading it all again here. With that in mind, I'm moving to a more stats based approach, though with the intention of providing a pool of data, rather than specific rankings (which will depend on how risky you want to be, last minute injury news etc which I struggle to keep up with 100%).

Update: There has been confusion as to what the original table was setting out. I do not really have 'rankings' as such, only a number of different metrics which I use to base my weekly captain decision on. I haven't found one to be the 'magic bullet' and thus I'm shying away from saying Player X is clearly the best pick of the week, unless of course he was the leader in every category.

The below data should all be taken into account and you can give as much weight to any individual metric you see fit. The tableau powered tables should allow you you re-order the data however you see fit. They are summarised as below:
  • P90 Points per 90 minutes at scored this season at home/away based on this week's fixture
  • QG Quality Games - Number of appearances with 6 or more points
  • BG Bad Games - Number of appearances with 3 points or less
  • % Percentage of goals accounted for (goals and assists) when the player has been on the field (save for a few rounding errors where a player is subbed, but you get the rough idea)
  • Rating Percentage of goals accounted for multiplied by forecast team goals to give the forecast number of goals a player will account for in a given week. Possibly the strongest of the metrics, but by no means a perfect solution.
As always, I welcome any suggestions or requests, specifically regarding any data you would find useful when making your weekly captain selection. Over the coming weeks I be working towards trying to build a 'catch all' formula which incorporates all these factors into a single, rankable number, but until then, enjoy the raw data:


17 comments:

Genevapics said...

As a Southie would say: "That graph is wicked good!"

satyam said...

its damn gud.. thats the best analysis one needs to pick a captain... i wd suggest chris to provide data for some more captain choices like silva rooney bale etc...

Voros said...
This comment has been removed by the author.
Voros said...

Given RvP is listed in the 17th placed in the attacking rankings (AR) for this week against a team listed 3rd in the clean sheet rankings, yet comes in a number 4 on the captain data, it seems a little misleading.

Is it his high % score pushing him up the list, even given the modest AR rank?

Chris Glover said...

Thanks Geneva

Satyam - do you mean add more data for the 'big names'? What extra data would you be interested in?

Voros - the rating is a very simple calc (probably too simple). RvP accounts for 63% of Arsenal's goals, who are forecast to score 1.1 goals this week and thus his rating is just 63% x 1.1. It's a crude calc, hence why these aren't really rankings as such (I would not captain Helguson for example).

I'm interested in the % of goals a player accounts for to try and show that Van Persie will be forecast to have a better week when Arsenal score 2 than Aguero will when City score 2.5 due to the heavy reliance Arsenal have on the Dutchman.

We basically say that Van Persie is getting a bigger piece of a smaller pie, which in this case isn't enough to get him ahead of Aguero but could be in some weeks.

I'm going to look in the off season as to how well these ratings correlate to actual production but I hope they're useful in the mean time, as a bit of an extra piece of information

Steven said...

Chris,
It would be useful to see % of assists as well. As we all know, goals and assists account for points and possible bonus points. Not sure how easy that would be. Anyhow, this is still very helpful. I'm looking at Silva this week myself.

unoriginallity said...

Hey Chris,

Digging the new graphs, the site looks the business.

I have a query, I'm kind of set on using the wildcard this week. I'm looking to bring in Dempsey and Aguero for Nani and Adebayor.

It would effectively end my United coverage apart from Evans. I am loath to do this but Nani's hit and miss form is frustrating and he ain't cheap. Should I try keep a United player?

Adebayor is more due to his next 6 fixtures. It's a move that will require adjustments throughout the team and I'm wondering if it's worth it.

Thanks man.

Kalix said...

Liking this data! :)


@satyam: Silva, Bale and Rooney are all on there...they're just pretty low down the list.

@Steven: The graph accounts for assists already. the value is the number of goals each player is involved in (either scoring or assisting). Obviously it requires some deduction as to which it'll be...e.g. If silva is involved its usually an assist, RVP: usually a goal.

stooshermadness said...

Chris - terrific piece. It gives some heft to one's captain selection as opposed to the usual weekly 'hunch'. The '%' column is based purely on the player's % of goals/assists in the games he has started or got more than a brief cameo viv-a-vis the goals the team has scored in the games in which that player started, correct ? Anyway, that's how it looked on the back of an envelope calculation for Zamora. Thanks, this is an incredibly useful chart.

dsb13 said...

I had already made up my mind to captain Helguson. He already had two double figures point. It's a punt I'm willing to take as my midfield of Silva, Sess, Walters and Dempsey would cover me if it fails. Up front, RVP and sturridge/ Torres.

Steven said...

@Kalix: Duh, I should read before I speak. Thanks!

Tony S said...

Hi Chris,
Brilliant, just brilliant!. This is without doubt the best and most useful piece of FF data I've ever seen. I've been crying out for something like this all season (mainly because I'm too lazy to do any research to compile a meaningful data set of my own).
The amount of data/stats compressed into such a small and simplified table is remarkable. I've already analysed the table several times and I’m sure I will do so many more times before the weekend. The inclusion of the QG and BG data was ingenious. It is so insightful into what type of player somebody is. I’m going to utilise this data when finalise my WC this week.
Thanks again, and I look forward to more of the same in the future (because there’s no point in me starting any research of my own at this stage).
Again, Brilliant!

manish said...

GRAPHS are good but I've been getting problems to load them up.. i don't know if it is case with everyone. i use mac and have latest firefox in it.

RaZor said...

I would suggest adding the "Form" factor in the equation.
I have long been thinking about this subject and how to consistently pick the right players to a given GW and the form of the player has to be taken into account.
Cheers,
Fred

Gummi at fantasyformation.com said...

Excellent graph. For many of us thinking about FPL all the week the "regular" articles do not give much extra information.

This graph does.

@unoriginallity: I have the same sitation. I'm wildcarding and have both Nani and Rooney. My hunch is that you would be wise to have some United coverage. Perhaps Valencia for Nani (instead of Dempsey)?

Tony S said...

Hi Chris,
I was Just wondering if it would be possible to add the number of home games in brackets in the QG and BG columns? It would nice to know instantly if the players good games were predominately at home.
Cheers

Yitzie said...

Dear Chris,
You are doing a fantastic job with your statistical posts, which i would love to see more of. Would be really interested to see a post about how teams bounce back from defeat, eg when spurs lost they won the next few by good margins, and some teams will not be as good as recovering. This would be fantastic to help predict games in the short term. Many thanks. Yitzie