Sunday, July 31, 2011

Goalkeeper combo strategy

[edit: thanks to Ben for pointing out that I forgot to include Foster's loan move to West Brom below. Replace all Myhill references will Foster but be mindful of his higher price]

I've been talking about the GK combo strategy for a number of years now, though the basic premise is worth repeating. There are two main issues which give to my support for this strategy:
  1.  If you go for a 'big name' keeper you effectively burn 4.0m as you are highly unlikely to ever play the lesser player. Only in a perfect storm where Petr Cech travels to the Emirates while John Ruddy welcomes Swansea will you bench the 7.0m man.
  2. The correlation between cost and points for 'keepers (right) has never been particularly strong with low budget options showing up every year (Al Habsi, Foster) to rank close to, if not above, their expensive counterparts (van der Sar).
With a combo strategy you play the best fixture each week, on the basis that if you stick to the logical projections, over the course of a season you will be sufficiently correct to generate enough points with your budget pair to match or exceed the points of a premium player, saving a couple of million in the process.

Method behind the madness
A couple of points to note on the methodology here:
  • GPG data is based on prior year defensive records, which I have found to be a fairly reliable indicator of future results. Where significant changes were made halfway through last season or this off-season I have allowed a small adjustment.
  • For the promoted teams, I looked at how previous newbies' goals scored declined upon entering the Premier League to generate a factor to apply to Norwich, Swansea and QPR's record from last season. 
  • Only 'keepers valued at 5.0m are included in the review. In prior year I had set this mark at 4.5m but this would leave us with just 8 viable options this year so I've expanded to 5.0m.
  • Where there is some uncertainty as to who will start (Mignolet vs Gordon, Begovic vs Sorensen) I have opted for the pricier of the two, which also happens to be the keeper I predict to win the #1 shirt.
  • The average GPG is calculated by taking a combination of a team's average GPG conceded at home/away and the opponents average GPG scored at home/away. We then select whichever keeper has the best fixture and finally take an average of these games to see which fixtures 'gel' best together. For this year, I have also included the number of games where the stats suggest a team will concede less than a goal (ie those with the best chance of a clean sheet).
Matching Pairs
I have chosen the first 10 gameweeks as a useful frame of reference to help select your keepers which will hopefully set you up nicely for the season, while allowing you to make the necessary moves if a particular team sees a substantial decline or improvement. For reference, I have included the data for the entire first half of the season at the bottom of this post.



Top Pairs

Rank Players Cost Average GPG GPG<1.0
1 Given / Begovic 10.0m 1.05 5
2 Begovic / Moreira 9.0m 1.11 4
3 Begovic / Robinson 10.0m 1.12 4
4 Given / Schwarzer 10.0m 1.12 3
5 Hart / 4.0m GK 11.0m 1.12 5
6 Schwarzer / Moreira 9.0m 1.13 3
7 Given / Kenny 9.5m 1.13 3
8 Kenny / Moreira 8.5m 1.13 3
9 Schwarzer / Robinson 10.0m 1.16 2
10 Schwarzer / Begovic 10.0m 1.17 2
19 Cech /4.0m GK 11.0m 1.21 2
38 Reina / 4.0m GK 10.5m 1.27 4
48 De Gea / 4.0m GK 10.5m 1.29 2

Once again, the stats seem to suggest that the GK combo pairing could top the elite 'keepers, and at a reduced price.A final caveat to add is that the new bonus system seems to favor 'keepers more than the old system and this could potentially be disproportionately to the benefit of the big name keepers. A boost in bonus points could help close the gap on the GK-combo system and hence this issue will need to be re-visited after a few gameweeks. Until then though, I am once again convinced that this is the best way to go, particularly as it will free up some cash to use on other areas of the team in light of the lack of reliable budget options this year.

The remaining team previews will continue to come over the next few days and remember to check back after Saturday for the write up of the inaugural SimPrem draft.In the mean time, all the latest news and links will be posted at @plfantasy.

First half data

9 comments:

Mell 10-11 said...

Nice post as always Chris, using EA Player Performance Index is such a grey area for the seasoned managers who has accustomed to the old system. But still it gives us a fair chance in the standing, I think. Not in favor of expensive keeper is justified though. For an extra kick code: 165858-164918

Ben said...

Hi Chris, great post as ever, and the previous one on bonus points was really useful too.

On GKs, how do you see Ben Foster's move to WBA affecting this? As you're basing data on last year, I guess the numbers would be the same for Foster as they are for Myhill? But I would think that Foster should at least deliver more save points, and with Roy Hodgson's increased focus on defence, WBA could be a bit stronger at the back. What do you reckon? For me, if he comes in at a good price, pairing Foster with Given looks like a winner.

Steve said...

This is an interesting area... if you have the time to delve into it :) Ultimately, unless the new bonus points change things, GK is not the position to spend big bucks.

I'm currently developing an Android FPL app, and have included some rotation stats based on last years results (see http://differentialfpl.blogspot.com/ for info on the app, btw!) wil which also update throughout the season. My conclusion is that something like Schwarzer / Moreira is probably an optimal budget combo (Kenny / Moreira would also be good, but too reliant on both promoted teams...).

Of course McLeish moving to Villa, and other changes will change these stats sooner rather than later - definately a good idea to look into this again thoroughly when wildcarding (hopefully much) later in the season!

Chris Glover said...

Ben - I should have updated my spreadsheet for Foster's move, so yes he will replace Myhill. I don't think however that I would move any rankings based on his arrival. He amy deliver slightly more save points than Myhill but I'd be surprised if the clean sheet totals moved considerably

Thanks for the comments guys.

Nikos said...

Chris,

Where, if anywhere, do saves play a role in this evaluation?

I'm thinking (maybe afraid I might be trying too hard to convince myself it's true) that a Given/Foster combination might be more valuable because they would be save beasts?

Thoughts?

Luke said...

Hi Chris, interesting thought that I haven't really considered before - was wondering which table you use to determine which of the two selected keepers to play each week? Really enjoy the site - Cheers, Luke

Simon said...

Just found this site, fascinating analysis.

The one thing missing here, unless I'm dim, is the factor that shows that one keeper in the pair is at home while the other is away.

Less goals should be conceded at home, so ideally you want the pair to rotate their home games as frequently as possible.

So surely the fixture list needs to be factored in?

S M said...

A well known tactic, and good stats to back it up. A few questions though:

Do you have any previous statistics for last season on how many points one can be expected to gain compared to the cost spent? Would be good with solid numbers to back this up compared to Star +4m keeper.

Would also be interesting to account for a certain error margin based on how you account for choosing the right keeper in the marginal games.

Agree with you though that combining right keepers with the fixture list is absolutely crucial here, perhaps more crucial than the keepers themselves.

ALM said...

Hi, thanks for pointing this particular strategy out to me, as I am usually one of the many to choose a pricey keeper and then 4.0m keeper whom I never play.

I have looked a few stats myself. The fixtures show that Villa and Fulham both play away from home in the same gameweek 15 times this season, with Fulham and Stoke doing so 4 times, and Villa and Stoke doing so 0 times. This would appear to suggest that going with Villa's Given and Stoke's Begovic would be a great combination, as you could rotate between the 2 knowing that you'll always have at least 1 Home goalkeeper for that week. Therefore Begovic is a definite for me, I just now have to contemplate whether to pair him with Given (0 away fixtures share) or Schwarzer (4 away fixtures share). Any thoughts on whether Villa or Fulham will have a greater propensity to concede goals this season, bearing in mind both have new managers