Lau Sze Yui

@903124S

Data scientist

Hong Kong github.com/903124/ Joined January 2015

1.5K Following 5.9K Followers 8.2K Posts

Latest posts 5

Pinned post

Lau Sze Yui @903124S Sep 21, 2023

Replying to @903124S

Why would this happens? We could tell the story of not going for 2 down by 8. While it will as demonstrated in diagram will net you 62.5% of win rate, it would also trigger opponent to be more aggressive which would costly in long run. (7/n)

Lau Sze Yui @903124S 1d

A simple margin over expected model using House and statewise positions election as baseline
Though my main pet peeve is political "WAR" model has little to do with wins nor "replacement" can be well-defined given different primary formats

Lau Sze Yui @903124S 5d

Pareto frontier of House and Senate seat lost

Jason Corley Verified account @JasonCorley78 5d

It is not Watergate. 1974 was a House wave with a small Senate echo. 49 House seats, 4 Senate seats.

The Senate loss, if we call it 8 seats with Alaska, is 1994 and 1986. The House, if it settles near 250, is 2018.

Neither number alone is a record. Together they are. A Senate loss of 8 and a House loss of 35 or more on the same night has happened twice since the WWII, in 1946 and 1958. No one today voted in either of those elections.

The president’s party has not been beaten like this in both chambers at once in 68 years.
View quoted post on X

Lau Sze Yui @903124S Nov 2

I've a different take on this using rolling past 8 elections primary turnout data with no poll, my model show it'd be D 52.1% or D+4.2.
At minimum you probably want primary turnout vs house vote regression line pass through origin which yield D+8 similar to poll today

Ethan C7 Verified account @ECaliberSeven Nov 2

Also worth noting that while a D+11 tsunami is certainly a surprise, it was also forecastable - in that we LITERALLY forecasted it in August via primary turnout.

It’s not just polls. Primaries, special elections, WA primary: it’s all coming up Tsunami.

decalibrate.substack.com/p/the-data-una…

View quoted post on X

Lau Sze Yui @903124S Oct 31

Problem of @stat_ron thinking of addressing uncertainty is often in pro sports teams it'd just increase confusion without actually gaining knowledge (see except on Seth Partnow's book "The Midrange Theory")

Ron Yurko @Stat_Ron Oct 30

It's a great question Tej! Assumptions in this logic are pretty strong (see here argmin.net/p/should-you-g… and punkrockor.com/2024/02/06/wha…)

View quoted post on X