Stats Chat

Team Ratings for Week 1

A caveat regarding these predictions. This is a competition where four teams have been added, and there is little information about the performance of those new teams against the existing teams. There is also little information concerning the effect of travel when the new teams are in a different continent. I have used previous form from the Pro14, Rainbow Cup, Super Rugby, and Super Rugby Unlocked but have not been able to tune model parameters using years of past data in the way I usually do, nor obtain initial ratings to my satisfaction. Particularly early in the competition, results may be erratic.

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.

Predictions for Week 1

Here are the predictions for Week 1. 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 4

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 21 matches played, 14 of which were correctly predicted, a success rate of 66.7%.
Here are the predictions for last week’s games.

Predictions for Round 4

Here are the predictions for Round 4. 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 2

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 6 matches played, 3 of which were correctly predicted, a success rate of 50%.
Here are the predictions for last week’s games.

Predictions for Round 2

Here are the predictions for Round 2. 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 the Preliminary Finals

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 198 matches played, 144 of which were correctly predicted, a success rate of 72.7%.
Here are the predictions for last week’s games.

Predictions for the Preliminary Finals

Here are the predictions for the Preliminary Finals. 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 8

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 18 matches played, 12 of which were correctly predicted, a success rate of 66.7%.
Here are the predictions for last week’s games.

Predictions for Round 8

Here are the predictions for Round 8. 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 7

The NPC site calls this round Updated Week 3, but it is the original Round 7, less games involving the three Auckland teams. Game times are much the same, and sticking with original round names is easiest for my data.

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 14 matches played, 10 of which were correctly predicted, a success rate of 71.4%.
Here are the predictions for last week’s games.

Predictions for Round 7

Here are the predictions for Round 7. 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 3

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 14 matches played, 9 of which were correctly predicted, a success rate of 64.3%.
Here are the predictions for last week’s games.

Predictions for Round 3

Here are the predictions for Round 3. 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 Finals Week 2

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 196 matches played, 142 of which were correctly predicted, a success rate of 72.4%.
Here are the predictions for last week’s games.

Predictions for Finals Week 2

Here are the predictions for Finals Week 2. 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.

• Good Herald piece on the Covid network contagion model 
• Andrew Chen organised a bunch of people to write a letter about privacy for Covid location data. He’s also been saying the same things to journalists.  It’s not that we think the government has any intention of misusing the data or letting the private sector misuse it, but the protections aren’t all that strong, the data collection is not voluntary, and having high-quality data is very important.
• There’s an interesting poll on US vaccine attitudes from the Washington Post and ABC News. Highlights: 82% of unvaccinated people said the FDA full approval for the Pfizer vaccine won’t make any difference to their decision. Crosstabulations (don’t you love polls with actual detail) showed 18% of unvaccinated respondents were in favour of requiring vaccination for school teachers and staff, and 15% requiring vaccination for students when a vaccine is approved at their age.  72% of those employed by someone else and not vaccinated said they would resign if required by their employer to be vaccinated. Politics site The Hill headlined this as Over 70 percent of unvaccinated Americans in survey would quit their job if vaccines are mandated, which is unlikely to be true — it’s a lot easier to claim to a poller you’d quit than to actually do it.
• Derek Lowe writes about bad clinical trials in Covid “I’m all for trying out new ideas – that’s essential, in fact. But try them out for real. … If you’re going to do research on human beings, you owe it to the subjects of your trial and to the rest of the medical community – and to the rest of the world, in this case – to do it right. To ask solid questions and get solid data on them that will allow you to make a real decision at the end of it.”
• Animation of vaccination progress in NZ (from Jonathan Marshall)
  

There are different ways of testing for the SARS-Cov-2 virus that causes Covid-19.  Broadly speaking, there are three approaches to the actual measurement: amplifying and testing for the viral RNA, testing for viral proteins, and testing for antibodies against the virus.  On top of that, samples can be be taken in different places: way up in the back of your nose, less far up, saliva samples, blood.

These tests are useful for different purposes because they have different characteristics.  The viral RNA tests using a deep swab and PCR have essentially zero false positives if you can avoid contamination.  That’s important in New Zealand because we use positive PCR tests to lock down the whole country, at nine-digit costs, and because we use them to put people in non-voluntary medical isolation.

The swab/PCR tests also have reasonably low false negatives. Nowhere near as low as the false positives — the lab assay is incredibly sensitive, but sometimes the swab just doesn’t pick up virus.  Again, we know this from NZ data. We currently have tests as soon after exposure as possible, again at five days, and again at twelve days, and people do test positive at five days or twelve days for the first time.  The false negative rate is important in New Zealand because we don’t want to miss even one case and allow an outbreak to expand.

For the places where we use swab/PCR testing now, we don’t want to substitute anything else. It’s the best technology available. But there are limits. The swab is a bit uncomfortable and the PCR process is slow and requires lab equipment in limited supply. We couldn’t, for example, do daily testing of all customer-facing essential workers with swab/PCR: they’d hate it and the labs would struggle to keep up.

In other countries, it’s much more valuable to have easy, rapid, and inexpensive tests even at a slight cost in sensitivity. There’s some risk of infection all the time; the consequences of a false positive or false negative are lower; there isn’t the same need to make sure all positive tests get reported. There’s a lot more scope for other tests to be helpful.

Even in NZ, though, there are gaps where other tests could be useful.  The obvious one is frequent testing of high-risk people. In normal times that would be people at the border; during an outbreak it might be essential workers whose job involves being exposed to customers or crossing the alert-level boundaries.  If we compare to swab/PCR the rapid antigen tests are not as good; but that’s not the right comparison. The rapid tests would be useful in settings where there isn’t going to be a swab/PCR. In those settings, the chance of detecting a case with swab/PCR is obviously zero, and the chance of detection with another test is actually pretty fair.

There’s a theoretical downside that negative rapid tests might slow someone with symptoms from getting a swab/PCR test until they get a positive.  If we were getting nearly 100% testing among people with symptoms, this would be a big concern. We probably aren’t anywhere near that; but it would need monitoring. There’s also a theoretical downside that false positive rapid tests might make people take positive results less seriously. I don’t think that’s plausible during lockdown, but again it would need monitoring. Somewhat more likely, it might turn out that there’s nothing actually wrong with the tests but that they don’t detect enough additional cases to be worth the cost and hassle. But that’s worth investigating, and where the realistic options are additional rapid tests or just the status quo, the effectiveness comparison should be between additional rapid tests and just the status quo.