Twitxy Saved

Derek Holliday

@d_e_holliday

AP @GWtweets | Former PostDoc @Stanford @PRL_Tweets | PhD @UCLA | Partisanship/State+Local Politics | #rstats | Trail Runner | PNW Expat | #YNWA #TridentsUp
Washington, DC derekholliday.com Joined June 2014
1.1K Following 1.1K Followers 286 Posts

Latest posts 5

Derek Holliday @d_e_holliday
I'm getting tired of this Kalshi victory lap. Below is a thread as to why. TLDR; forecasters and PMs are playing two very different games, PM performance does not suggest sophistication, and, when accounting for uncertainty, Kalshi is not reliably better. 🧵 1/9
Nicole Kagan Verified account @nkagan_official Nov 6
We’ve just made all of our Midterm model data comparisons public @KalshiResearch. You can see them here: kalshi.com/research/forec….

Today, the NYT claimed prediction markets were 'worse' than election models in the lead up to the midterms. But they ignore a few key pieces of the puzzle 🧵
View quoted post on X
2 replies 4 reposts 21 likes 1.3K views View on X
Derek Holliday @d_e_holliday
This is pretty clear when watching election night coverage. While it's mostly pundit slop, can't deny that when Kornacki goes to the board, there's genuine value-add there.

When Kalshi has their slot, the entire bit is "wow look at that line go up." No subtance. It's junk food.
Nate Silver Verified account @NateSilver538 Nov 5
Give me a break, the polls and models provided the direction in contast to vibes and dumb insiders. And yes it's good that traders were smart enough to trust them. But the better election models were more confident about the outcome than the markets and got there more quickly.
View quoted post on X
Derek Holliday @d_e_holliday
A recent blood-boiling assignment with a manuscript so clearly one-shotted through an Opus prompt brought me pretty close to this position.

After neutering the most acerbic parts of my review, I think I've gathered enough clear-headed thoughts to get in on the hot discourse (🧵)
Paul Novosad Verified account @paulnovosad Sep 23
Journals should have a zero tolerance policy for AI writing. I know we can think of justified cases, but these will be swamped by an avalanche of slop that will demoralize referees and kill journals entirely.

The bad money drives out the good.
View quoted post on X
1 replies 2 reposts 32 likes 13K views View on X
Derek Holliday @d_e_holliday
Perhaps an even hotter take: automated reviews of replication archives upon initial submission 🤷

All repos do now is confirm there is code that runs that produces the numbers in the manuscript. We (mostly) take it on faith the code does so correctly. Doesnt need to be that way!
Political science journals: You should do this too! Maybe even (gasp) before sending out for review or as part of the review process!
View quoted post on X
1 replies 2 reposts 10 likes 2.1K views View on X