Stats Chat

Team Ratings for Week 18

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 136 matches played, 93 of which were correctly predicted, a success rate of 68.4%.
Here are the predictions for last week’s games.

Predictions for Week 18

Here are the predictions for Week 18. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Round 24

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 161 matches played, 125 of which were correctly predicted, a success rate of 77.6%.
Here are the predictions for last week’s games.

Predictions for Round 24

Here are the predictions for Round 24. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Week 14

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 63 matches played, 45 of which were correctly predicted, a success rate of 71.4%.
Here are the predictions for last week’s games.

Predictions for Week 14

Here are the predictions for Week 14. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Round 11

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 80 matches played, 47 of which were correctly predicted, a success rate of 58.8%.
Here are the predictions for last week’s games.

Predictions for Round 11

Here are the predictions for Round 11. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Week 11

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 81 matches played, 61 of which were correctly predicted, a success rate of 75.3%.
Here are the predictions for last week’s games.

Predictions for Week 11

Here are the predictions for Week 11. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Week 13

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 58 matches played, 40 of which were correctly predicted, a success rate of 69%.
Here are the predictions for last week’s games.

Predictions for Week 13

Here are the predictions for Week 13. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Round 10

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 72 matches played, 41 of which were correctly predicted, a success rate of 56.9%.
Here are the predictions for last week’s games.

Predictions for Round 10

Here are the predictions for Round 10. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

Team Ratings for Week 10

The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.

Performance So Far

So far there have been 72 matches played, 52 of which were correctly predicted, a success rate of 72.2%.
Here are the predictions for last week’s games.

Predictions for Week 10

Here are the predictions for Week 10. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.

RNZ has a report on the purported costs of weather warnings for (ex-)Cyclone Vaianu a few weeks ago.

Northland spending was down 48 percent. Auckland’s was down 46.5 percent, Waikato was down 52.58 percent, Bay of Plenty down 68.32 percent, Gisborne down 51.6 percent and Hawke’s Bay was down 56.34 percent.

That’s at least two more decimal places than I would want to use, but let’s assume the numbers are correct.  There are still a few issues with the attribution to weather warnings.

First, we can’t easily distinguish the effects of the warnings, the effects of the forecast, and the effects of the storm.  Presumably no-one is suggesting that weather forecasts should be edited to make approaching storms look less scary, but some of the reduction in spending would have been a response to the heavy rain and strong winds in the forecast, regardless of a specific warning.  Even if the forecasts were sanitised, people who remember, say, the Auckland Anniversary Weekend storm or Cyclone Gabrielle might want to be cautious about the risks of an approaching storm.  The report says the reductions in spending were highest in the areas where the forecast storm was most intense, but that’s what you’d expect if people were reacting to the forecast rather than to the North-Island-wide warning.

Second, there’s an economic issue that should be clear already to people in or near local government, because it’s part of how convention centres, stadiums, and major cultural events are oversold.  It’s much easier to get people to change where and when they spend money than to change how much money they spend.  Suppose you were planning a barbeque for your kid’s birthday on April 12th. You might well have cancelled it because of the storm warning, but that wouldn’t mean your kid didn’t get a birthday this year. You’d have the event some other time.  Similarly, if you wanted to see a movie (Project Hail Mary seems popular) or you needed to buy a new vacuum cleaner or you wanted to see the fireworks and racing at Waikaraka Park you might well not go out that day, but you might well go some other day when you could easily get to the cinema or Briscoes or when the racing and fireworks were rescheduled¹.  The reduction in spending on the day of the storm is a substantial overestimate of the impact of the storm on spending. (In the same way, stadiums collect money, but a lot of it is money that people would otherwise still have spent in the city).

And finally, there’s an important issue about uncertain predictions.  The storm didn’t in fact cause major damage across the North Island, but that doesn’t mean the warning was wrong.  If you play Russian roulette and survive, that doesn’t mean it was a good bet and you should do it again, it just means you got lucky.  The report gives no reason whatsoever to assume that less caution was warranted for this storm than for the two big storms in 2023 (one of which was seriously underpredicted).  It’s not inconceivable that our weather warnings are poorly calibrated, but any decision of this sort should be based on some actual evidence that they are.

1. Last night. Ask me how I know.

• Canada is starting its census on Monday
• Another ultraprocessed food headline: CNN says Even a single daily serving of ultraprocessed food may raise dementia risk.  As before, the research didn’t even try to compare people with no ultraprocessed food intake to those with only a single daily serving.  They used a model that approximates the relationship by a curve with no safe level, and so couldn’t address this question — nor did they claim to.  The “even a single daily serving” is CNN.  What the research claimed is that each extra daily serving translated to 10% more dementia risk.  I’m a bit dubious, for the simple reason that ultraprocessed food consumption has gone up but dementia risk (at a given age) hasn’t. But I don’t think anyone’s claiming ultraprocessed food is good.
• Strava makes an app that tracks cycling and running and confidential military locations. Over the last year, Seattle was the top US city for logging bicycle commutes, with Chicago and Minneapolis in the next positions.  The Seattle Times says Seattle front of the pack for bike commuting in U.S. cities, but given Seattle’s status as a West Coast tech and nerd haven, it seems quite possible that we’re seeing selection bias.  It could easily be that Chicago or the Twin Cities are ahead in commuting but people there are just less likely to report it to Strava.
• Nice blog post by Kieran Healy on maps of data and using hexagons to make all the regions visible
• A graph from The Economist. 
  
  I first noticed the unusual definition of Asia in the footnote, especially the exclusion of countries no-one would think to include. The time axis could also be clearer. I presume the slightly longer tick mark is either Dec 2025 or Jan 2026, but a stronger visual cue to past vs future would perhaps be more helpful.  The other aspect, which is very common in graphs of the fuel crisis, is the smoothness of the projected decline in the face of uncertainty as to both supply and attempts to manage demand.