Thursday, November 22, 2012

Gameweek 13 Preview

Team Attacking Forecast
Team Defensive Forecast
Individual Player Forecast
In anticipation of the inevitable "why do you hate van Persie" questions, I thought I better add a quick note on his unexpectedly low ranking here.

The key difference the model sees between van Persie and Rooney is simply the percentage of United's chances he has accounted for when playing. Rooney sits at 29% in home games while van Persie comes in with a useful 20%, which based on United's forecast 14 SiB gives Rooney a forecast 1.3 shot advantage. Rooney also enjoys a distinct advantage in SoB, though given the low likelihood of those turning into goals, that doesn't have a huge impact. This all translates to a relatively slim, but not insignificant, 1.4 point lead and thus likely some scorn for these electronic pages.

If I'm honest, if I owned both players and hadn't looked at the above data, I would undoubtedly have captained van Persie but while many are quick to point to the drawbacks of stats, the above numbers are fairly self explanatory and thus I feel fairly confident standing behind them. Rooney currently has 2% of the captain vote at Fantasy Football Scout, to van Persie's 66% but I'm sure you can find your own 'following the crowd' quote here depending on how you feel about the above analysis. Any logical suggestions as to why van Persie might actually be better than Rooney would be more than welcomed below or at @plfantasy.


hthejambo said...

Could it b the fact that rooney is playing deeper than rvp therefore rvp has more chance of doing something than rooney

CDI said...

Hey Chris, I actually owned both as I believed it was pretty close between them but then Rooney missed the last game and I rage cut him(regret it now).

I think the main benifit RVP owners will cite over Rooney is the easy cap option he provides most weeks. Your data suggests the same with Rooney having worse stats away from home while RVP does well both Home and Away. Is this enough to justify the current 2 mil difference? prob not in terms of pure value but at some point you just have to pay for points or we would fill our teams with Maloney type players.

Team values are high enough now that owning RVP doesn't auto break your team anymore and I think alot got him when he cost 13 mil and not the 13.7 he costs today. I noticed you have quite a low team value but if you were given an extra 3 mil in the bank would you pick Rooney over RVP?

Turd said...

Surely a lot of the difference in the voting percentages in the FFS poll is due to the relative ownership of the players in the FPL game.

RvP is in 35.7% of teams, while Rooney lags way behind on 6.3%.

I bet the difference is even bigger amongst the self-selecting population of those who frequent the FFS website.

I'm not making any comment on their relative point scoring potential for this gameweek, mind!

Small Paul said...

Stoke at home have conceded one goal in 5 games, surely Berba shouldn't have such high attacking returns predicted?

Ryan said...

How do you calculate Berbatov's returns? Is it based on his past data at different clubs? Would you say that the pool is too small to have much of an idea about how he and Fulham will go away to Stoke?

Chris Glover said...

jambo - that's why I was so concerned about Rooney when van Persie joined United, but so far he has more shots per game, so do we care if he playing deeper? There could be argument that the quality of those shots is worse, and that is somewhat backed up by the % of his shots that were taken inside the box, but still, as a % of United's total SiB, Rooney's is higher. The standard small sample size issues are of course worth mentioning though.

Chris Glover said...

CDI - fair point but Rooney is projected to out score van Persie in total points, not on some kind of value basis. True, van Persie's standard deviation is lower, but I'm not sure that simply captaining a single player every week is particularly advisable.

I don't put much emphasis on team value, which in a vacuum is foolish, but I've just found that to make huge gains you generally need to be following the crowd (which I generally don't) and I've done just fine in the second half of seasons without a big budget.

As I say, the model likes Rooney to simply out score van Persie, regardless of cost. Now, I'm not saying the model is necessarily correct, but I don't think it's making any outlandish claims that more shots, particularly those inside the box, lead to more goals and thus Rooney looks a touch better.

I'm not confident (or arrogant) enough to say I'm right, in fact I assume I'm wrong, but I've yet to see a reason to support Van Persie definitely being better than Rooney, other than some vague "he's better", which just doesn't really show up in any of the stats.

Chris Glover said...

Turd - yeh fair point, though when I vote I don't take account as to who I have, I vote for the best player that week.

Chris Glover said...

Paul and Ryan - Stoke have enjoyed excellent results at home, but the process in getting there has been only solid. They're surrendering 9 SiB in home fixtures and if they continue to do that, it's very unlikely they'll average such a low GPG for the year.

Fulham have outshot the opponent adjusted average team this year by 25%, so we're giving them 11 SiB based on Stoke's form. As for Fulham themselves, they've averaged 9 SiB on the road, with Stoke performing slightly below average at 7%. Hence, both measures expect Fulham to get around 10 SiB which they've converted to goals at an 18% rate, a touch above the league avg of 15%.

Berbatov has accounted for at least 18% of his team's SiB in every game this year, averaging 32% away from home. So we expect him to get perhaps three chances in the game, which gives him a decent shot at a goal (behind Ba, Rooney, Giroud and Aguero). He then has a decent shot with assists too, so he gets a bump to push him up the rankings.

The danger with saying "Stoke have conceded X goals at home" is that if in those games they've given up massive shot totals, the sustainability of that success is questionable.

On the data point, I am only using current year data, other than for the % of times a player hits the target with his shots, as I've shown this rate to be sustainable over the years and based on something other than random variation.

There is an issue with small sample sizes for sure, and I hope I make that caveat enough during my ramblings, but I don't see an easy way round it, especially for players like Berbatov who have changed teams. Given that the season is only 38 games long, and the fantasy season is massively important for the first half (after which teams become homogenous), if we waited for a reliable sampel size, the season would be over.

This data is NOT a magic bullet and if you believe Berbatov will struggle without Kacaniklic, or that City will concede more goals with Maicon at right back, feel free to make that adjustment. I don't do so as those views as largely subjective and I don't want to force my own biases down the reader's throats. I worry that those narratives are applied after the fact to confirm our own biases, but that's just my view, so I let the data talk and then investigate where things go wrong.

Sorry for the long reply, and thanks for the questions

Pat Vivat said...

another vote for rooney :)

Tony S said...

Hi Chris,
Great article and thanks for replying to the comments above because the same questions came to my mind regarding Van Persie vs Rooney and Berbatov vs Stoke.
But as you point out this is what the Stats say, and you are only highlighting what the raw data implies without adding a personal or biased interpretation.

Personally, I like to analyse your data and then put on my own spin on the results based on my own gut feelings. For instance, I transferred out Ba this week for Lambert even though the numbers probably don't recommend it but I personally fear for the Newcastle squad over the coming busy weeks.

So please keep on providing the raw, untainted data and I'll interpret it as I see fit :)


Small Paul said...

Hi Chris, thanks for the reply about Berba's high predicted returns. I understand the model's logic behind it now, but this makes me think the model may need some tweaking. The prediction for Berba stood out as massively optimistic. I didn't see any prediction of Fulham scoring freely at Stoke apart from here. The danger of ignoring that Stoke have now conceded one in six at home is that the fact is, they've conceded one in six at home. Does the model allow for teams conceding SiBs but not high number of goals?