Field Theory by Hadyn Green

I am current first equal in the work "pick the results" contest at work at the moment. World rugby rank explains 80% of the scoring results between teams. But you can then factor in how much particular teams are under or overpreforming relative to their rank in the tournament. I think a very narrow win to Aregentina against Ireland, as Argentina are scoring enough extra points for their rank they just overtake Ireland (on average)

Who won?

The All Blacks of course. I have to say that game was a stunner. The earlier one had moments of thrills as well. What TMO??? I don’t think i saw it once!! And that was with Mr Barnes too!!

This world cup is the first time (I think anyway) that we are seeing a strategy that says sucks to the media about every game needing a 50 point margin, lets work out the details, the plan and the execution. Lets try them out in real life for four games (plus the Championship) and then see how we go. Can anyone imagine any AB coach trying that? Even Henry? I hope people appreciate Wayne Smith. He was an AB coach once but I think he felt he was better in a backup role and Henry saw that and brought him back to the fold. It has paid off in spades twice (nearly hopefully) over when Hansen kept him on.

You should seriously consider getting a life.

For the semis I am saying NZ by 18 over SA, and Argentina by 10 over Australia,

and if I show my working https://thoughtfulbloke.wordpress.com/2015/10/22/rugby-world-cup-regression/

David,

Interesting analysis and code. I tried it but got:

> u <- “https://en.wikipedia.org/wiki/World_Rugby_Rankings”
> if(!(file.exists(“rugby.html”))){
+ download.file(u, destfile="rugby.html”, method = “libcurl”)
+ }
Error in match.arg(method, c(“auto”, “internal”, “wget”, “curl”, “lynx”)) :
‘arg’ should be one of “auto”, “internal”, “wget”, “curl”, “lynx”
>

What's "libcurl"?
Am I missing a library?

librcurl came in in R 3.2 as a cross platform way of getting https urls, so I would guess you are using a slightly older one. If that is the case, for the version you are using method = "internal" will work with Windows, method = "curl" should work with mac, and method = "wget" should work with most linux distros.

Thanks. I have 3.0. I downloaded the file manually and adjusted the line numbers in your code. The residuals looked a bit larger around sdiff=0, and I thought there might be some heteroscedasticity for a model of absolute values, but Breusch-Pagan says "no":

bptest(I(abs(sdiff)) ~ I(abs(rdiff)),data=scores)

One thing I didn't look at, so don't know if it actually matters, was starting from the logical principle that a team that played itself should tie (or a least win as often as it loses). So a zero difference in ranks/ zero difference in score should, in a "seems like a reasonable argument" kind of way, be an origin point for the expected results. I just generated a basic line from the data to hand.

Intercept=2.222e-15 and P=1 looks pretty convincing :).

How do you account for Pocock vs no Pocock?

So how are you going compared to David Scott's predictions on Stats Chat?

I don't- with only five matches to draw from I'm just taking a broad view- arguably though from the graphs you could claim that with Pocock on the field Australia have been underperforming slightly, and with him off they have been underperforming noticeably. Given the closeness of the Australia/ Argentina match that kind of thing could make a difference, but there are a whole bunch of other things that could make a difference on the day as well and I'm not really separating any of them out, figuring all in all they balance.

some stats- I am one up as of the quarter finals in terms of wins. But really, because of the variation on game day a better measure would closeness of points difference. If Argentina win by 1 then I will have picked the win but David Scott would be closer to the points difference between the teams.

Let's say I think we are in the same ball park :)

David Scott's model and my model seem to have been particularly disagreeing over Argentina in the past couple of weeks.

I'm happy I have a life to watch games like todays. :-)