Surprised that NS’s comments don’t reference the caliber of the teams the Knicks beat. Am i off in thinking that the West, not just the Thunder, were better than the East. Doesn’t that need to be discussed and factored in? Spurs in 5 or 6.
Here I go again. This piece is uninformative in the end due to failure to begin with a base-rate anchor. A base rate is a simple extrapolation of the past that you update with the kind of specifics contained in this piece. It's essential so you know which part of the distribution to analyze.
With a 15% base rate for a Knicks championship you look for low-probability upside--Wemby foul trouble, Castle's inexperience, San Antonio's half-court offense in a grind-it-out series--you don't waste time regressing your own team's shooting luck. With a 45% base rate you don't need reassurance that "the archetypes people have in their heads are too rigid." The piece is quoting arguments all over the map from those useful for updating 5% base rates and those useful for updating 95% base rates.
One easy base rate is prediction markets (36% Polymarket, 37% Kalshi) and betting (36% FanDuel adjusted for vig). Another is to use the two major statistics cited--regular season wins (higher total wins 77% of the time since 1984) and pre-finals playoff scoring margin (higher total wins 64%). In the 18 finals where the measures conflicted the low-win/high-margin team won 6 times, a 33% base rate.
If you take a harder look, it's even worse for the Knicks. Low margins hurt high-win teams, but high margins don't help low-win teams. High-win/high-margin finalists win 87% of the time, high-win/low-margin only 45% (Fisher p = 0.005). Low-win/high-margin finalists win 40% of the time, versus 21% for low-win/low-margin but the p is only 0.30 so this could easily be noise.
The one caveat is the Knicks' margin is off-the-charts high, the bookies and prediction markets are rewarding handsomely from the 3-22 record suggested historically for a 53-win versus 62-win team, or the 1-9 (2016 Cavs the only success) adjusting for the higher pre-finals scoring margin. So the size of the Knicks' margin is giving them something like a 25% bonus, with no historical data either way to calibrate.
My impression from reading the piece is it starts out rhetorically with something like a 15% prior--about what the Knicks were before the playoffs began. It reads as bullish because it starts from there, but it actually finishes with what I would call a range of 30% to 45%--a much wider range than any reasonable statistical analysis should provide, and centered on the consensus, and all in qualitative terms.
I asked Claude to read the piece and try to put numbers on the qualitative arguments. I armed him with the actual Polymarket price history and the analysis I did on previous NBA championships. I did not supply him with previous Silver Bulletin writings to judge style. His verdict:
Piece opening prior - 15%
"+19.4 margin," "elite company" - 34%, basically Knicks playoff success to date
"weight recent heavily" - 38%
"Spurs also dominant" - 33%
"Knicks playoff experience" - 36%
"Defense leap, 14th to 7th" - 40%
"Regress shooting luck" - 35%
"Offense good to great" - 39%
"More than sum of parts" - 41%
"I wouldn't bet on them at even money, especially with Game 7 scheduled for San Antonio." - this is the closest to a quantitative prediction saying it's under 50%, but might be close to 50% with home court advantage - 43%
closing paragraph with the 0-4 sweep mentioned - anywhere from 30% to 45%
I am not a Knicks fan although a was a season-ticket holder for over a quarter century starting in the early 1980s, and you cannot avoid some affection from that. I have been betting on the Knicks all season, and would be indifferent on taking them at even money in the championship--but for reasons entirely unrelated to those in the piece. I could well be wrong, I lose a lot of bets, but I am a consistent Bayesian in my analysis, and put numbers--and money--on the conclusions.
You point out that "the size of the Knicks' margin is giving them something like a 25% bonus, with no historical data either way to calibrate," but then you say that the language of Nate's conclusion suggests a posterior interval of 30% to 45% which is "a much wider range than any reasonable statistical analysis should provide." If the Knicks' margin is "off-the-charts high" with no historical precedent, doesn't that imply a very wide posterior?
The probability that the Knicks will win the championship is a point estimate--the mean of the posterior distribution if you prefer. You can have the widest possible posterior, something you know nothing about, but there is still a bet you would be indifferent between taking either side.
Bruno de Finetti's example was the probability that there was life on Mars a billion years ago. This is from the 1930s before any kind of data was available. Still, if forced to make a choice at some odds with a gun to your head--say 50 to 1 against--you would answer. That answer might be different at 5 to 1 or 500 to 1.
If you put a range on your belief, I think of it as a bid/ask spread. Say, I'd bet on the Knicks at +200 (33% breakeven) but take the Spurs at -170 (37% Knick win probability breakeven). That's a 4% spread which is too large for betting, but reasonable for an essay meant to speak to people with different priors.
Writing an essay to say you'd take the Knicks at +233 (30% breakeven) or the Spurs at -122 (45% breakeven) is pointless with a market of +172/-205. I'm pretty confident Nate Silver's actual bid/ask spread it much tighter than that, but it did not come through in the essay.
Moreover, while it's true there's no historical data to evaluate the Knicks' playoff margin, that is hardly the only information we have to go on to estimate their chances. If all you knew were the season records and playoff margins of the two finals teams, people with different priors would likely come up with different point estimates, but each one individually should still have a point estimate.
I'm taking the over on cut-aways showing Spike Lee, Timothée Chalamet, Jerry Seinfeld, Ben Stiller, Jon Stewart, Alicia Keys, Chris Rock, Howard Stern, etc, sitting court side. Apologies to any celebrities not mentioned in this comment.
If I squint hard I can see a rationale for questioning the conventional wisdom that you need a top 5 player to win a title. But putting that one and the same category as podcast/bar room pablum about the Warriors and Jokic just seems silly. It’s like the difference between saying you can’t have a great restaurant without great ingredients and you can’t have a great restaurant without great pasta. In any case, the Knicks don’t have anyone in the same zip code as top 5. Nate’s reasoning for Brunson being better than top 12 - he’d take him over Jaylen Brown - doesn’t inspire confidence. You won’t find a bigger Celtics fan than me, and I wouldn’t place Brown anywhere near the top 12. Nor do I think you will find any actual data that would place Brown that high. Or Brunson. For the record, EPM ranks Brunson 12th this season offensively. And 472nd on defense. Granted, individual defensive metrics are squishy, and less important than offense. But 472nd? Anyway, if you check your Northeast Chelsea biases at the door and spend even five minutes running through a list of top NBA players, you’ll pretty quickly conclude that Brunson isn’t sniffing top 10 territory.
The Knicks have beaten this iteration of the Spurs, though. It will be interesting to see if Brown thinks Tyler Kolek has something that’s kryptonite to the Spurs, as he was in the NBA Cup final.
The Spurs are so young that 3 months since that game of playoffs makes a real difference. They shot the ball poorly that night and committed a season high in turnovers.
We will see. If we’re name dropping…Stephon Castle a playoff basketball player.
I’m interested to see how Little Luka does against the Spurs platoon of guys they send at him with Wemby back there to clean it all up.
Yes, but basing on winning 2 of 3 games is pretty silly. KAT is a problem for Spurs to match up against. Castle and Harper improved their 3 point shooting dramatically during the second half of the season. As Edward hinted, they are so young that it is likelier to last. Spurs offense in the first half of the year felt stodgy when both Castle and Harper played because of the lack of outside shooting, but that changed as January went to February, Barnes went to the bench, and Wemby was fully healthy after missing a dozen or so games in November-December.
I see matchup advantages both ways - as alluded, KAT is a nightmare as he either drags Wemby out of the paint, or he has a major size mismatch. But at the same time, Spurs were designed in a lab to cover Brunson, as the Spurs have long wings who can play tight defense with no fear of dribble penetration - they just funnel that shit down to the big fella and most teams just recycle the possession.
Too close to call for me, but if pressed, I would pick Spurs in a tight series.
Another view is that the Knicks benefitted immensely from the Pistons' Jalen Duren collapse that allowed a middling Cavs team to advance. The Pistons matched up well with the Knicks all year.
I also really like Hot Hand Theory’s take on how the Knicks can try to get Wemby out of the paint, and agree with them that Anunoby is probably their best defender against Wemby: https://youtu.be/INM1a-MDQHo?si=TFj7lahHXFQnKeOj
Surprised that NS’s comments don’t reference the caliber of the teams the Knicks beat. Am i off in thinking that the West, not just the Thunder, were better than the East. Doesn’t that need to be discussed and factored in? Spurs in 5 or 6.
It’s funny how you can go all over the country and find Giants and Yankees fans, but basically no one outside New York is a Knicks fan.
Here I go again. This piece is uninformative in the end due to failure to begin with a base-rate anchor. A base rate is a simple extrapolation of the past that you update with the kind of specifics contained in this piece. It's essential so you know which part of the distribution to analyze.
With a 15% base rate for a Knicks championship you look for low-probability upside--Wemby foul trouble, Castle's inexperience, San Antonio's half-court offense in a grind-it-out series--you don't waste time regressing your own team's shooting luck. With a 45% base rate you don't need reassurance that "the archetypes people have in their heads are too rigid." The piece is quoting arguments all over the map from those useful for updating 5% base rates and those useful for updating 95% base rates.
One easy base rate is prediction markets (36% Polymarket, 37% Kalshi) and betting (36% FanDuel adjusted for vig). Another is to use the two major statistics cited--regular season wins (higher total wins 77% of the time since 1984) and pre-finals playoff scoring margin (higher total wins 64%). In the 18 finals where the measures conflicted the low-win/high-margin team won 6 times, a 33% base rate.
If you take a harder look, it's even worse for the Knicks. Low margins hurt high-win teams, but high margins don't help low-win teams. High-win/high-margin finalists win 87% of the time, high-win/low-margin only 45% (Fisher p = 0.005). Low-win/high-margin finalists win 40% of the time, versus 21% for low-win/low-margin but the p is only 0.30 so this could easily be noise.
The one caveat is the Knicks' margin is off-the-charts high, the bookies and prediction markets are rewarding handsomely from the 3-22 record suggested historically for a 53-win versus 62-win team, or the 1-9 (2016 Cavs the only success) adjusting for the higher pre-finals scoring margin. So the size of the Knicks' margin is giving them something like a 25% bonus, with no historical data either way to calibrate.
My impression from reading the piece is it starts out rhetorically with something like a 15% prior--about what the Knicks were before the playoffs began. It reads as bullish because it starts from there, but it actually finishes with what I would call a range of 30% to 45%--a much wider range than any reasonable statistical analysis should provide, and centered on the consensus, and all in qualitative terms.
I asked Claude to read the piece and try to put numbers on the qualitative arguments. I armed him with the actual Polymarket price history and the analysis I did on previous NBA championships. I did not supply him with previous Silver Bulletin writings to judge style. His verdict:
Piece opening prior - 15%
"+19.4 margin," "elite company" - 34%, basically Knicks playoff success to date
"weight recent heavily" - 38%
"Spurs also dominant" - 33%
"Knicks playoff experience" - 36%
"Defense leap, 14th to 7th" - 40%
"Regress shooting luck" - 35%
"Offense good to great" - 39%
"More than sum of parts" - 41%
"I wouldn't bet on them at even money, especially with Game 7 scheduled for San Antonio." - this is the closest to a quantitative prediction saying it's under 50%, but might be close to 50% with home court advantage - 43%
closing paragraph with the 0-4 sweep mentioned - anywhere from 30% to 45%
I am not a Knicks fan although a was a season-ticket holder for over a quarter century starting in the early 1980s, and you cannot avoid some affection from that. I have been betting on the Knicks all season, and would be indifferent on taking them at even money in the championship--but for reasons entirely unrelated to those in the piece. I could well be wrong, I lose a lot of bets, but I am a consistent Bayesian in my analysis, and put numbers--and money--on the conclusions.
I'm curious why you referred to Claude as "him" instead of "it".
Too much time spent working with AI. I'm starting to refer to people as "it".
Is it actually negatively impacting how you perceive people?
I would be the last to know.
You point out that "the size of the Knicks' margin is giving them something like a 25% bonus, with no historical data either way to calibrate," but then you say that the language of Nate's conclusion suggests a posterior interval of 30% to 45% which is "a much wider range than any reasonable statistical analysis should provide." If the Knicks' margin is "off-the-charts high" with no historical precedent, doesn't that imply a very wide posterior?
The probability that the Knicks will win the championship is a point estimate--the mean of the posterior distribution if you prefer. You can have the widest possible posterior, something you know nothing about, but there is still a bet you would be indifferent between taking either side.
Bruno de Finetti's example was the probability that there was life on Mars a billion years ago. This is from the 1930s before any kind of data was available. Still, if forced to make a choice at some odds with a gun to your head--say 50 to 1 against--you would answer. That answer might be different at 5 to 1 or 500 to 1.
If you put a range on your belief, I think of it as a bid/ask spread. Say, I'd bet on the Knicks at +200 (33% breakeven) but take the Spurs at -170 (37% Knick win probability breakeven). That's a 4% spread which is too large for betting, but reasonable for an essay meant to speak to people with different priors.
Writing an essay to say you'd take the Knicks at +233 (30% breakeven) or the Spurs at -122 (45% breakeven) is pointless with a market of +172/-205. I'm pretty confident Nate Silver's actual bid/ask spread it much tighter than that, but it did not come through in the essay.
Moreover, while it's true there's no historical data to evaluate the Knicks' playoff margin, that is hardly the only information we have to go on to estimate their chances. If all you knew were the season records and playoff margins of the two finals teams, people with different priors would likely come up with different point estimates, but each one individually should still have a point estimate.
Most models have the Knicks as a 30-40% series underdog. 15% is what they would have been against (a healthy) OKC.
I'm taking the over on cut-aways showing Spike Lee, Timothée Chalamet, Jerry Seinfeld, Ben Stiller, Jon Stewart, Alicia Keys, Chris Rock, Howard Stern, etc, sitting court side. Apologies to any celebrities not mentioned in this comment.
If I squint hard I can see a rationale for questioning the conventional wisdom that you need a top 5 player to win a title. But putting that one and the same category as podcast/bar room pablum about the Warriors and Jokic just seems silly. It’s like the difference between saying you can’t have a great restaurant without great ingredients and you can’t have a great restaurant without great pasta. In any case, the Knicks don’t have anyone in the same zip code as top 5. Nate’s reasoning for Brunson being better than top 12 - he’d take him over Jaylen Brown - doesn’t inspire confidence. You won’t find a bigger Celtics fan than me, and I wouldn’t place Brown anywhere near the top 12. Nor do I think you will find any actual data that would place Brown that high. Or Brunson. For the record, EPM ranks Brunson 12th this season offensively. And 472nd on defense. Granted, individual defensive metrics are squishy, and less important than offense. But 472nd? Anyway, if you check your Northeast Chelsea biases at the door and spend even five minutes running through a list of top NBA players, you’ll pretty quickly conclude that Brunson isn’t sniffing top 10 territory.
I’m trying to make “Joker KAT” a thing :-)
The obvious criticism of this post is it made no contact with the Spurs. However I do give the Knicks more of a chance than I did before reading it.
The Knicks have beaten this iteration of the Spurs, though. It will be interesting to see if Brown thinks Tyler Kolek has something that’s kryptonite to the Spurs, as he was in the NBA Cup final.
The Spurs are so young that 3 months since that game of playoffs makes a real difference. They shot the ball poorly that night and committed a season high in turnovers.
We will see. If we’re name dropping…Stephon Castle a playoff basketball player.
I’m interested to see how Little Luka does against the Spurs platoon of guys they send at him with Wemby back there to clean it all up.
Yes, but basing on winning 2 of 3 games is pretty silly. KAT is a problem for Spurs to match up against. Castle and Harper improved their 3 point shooting dramatically during the second half of the season. As Edward hinted, they are so young that it is likelier to last. Spurs offense in the first half of the year felt stodgy when both Castle and Harper played because of the lack of outside shooting, but that changed as January went to February, Barnes went to the bench, and Wemby was fully healthy after missing a dozen or so games in November-December.
I see matchup advantages both ways - as alluded, KAT is a nightmare as he either drags Wemby out of the paint, or he has a major size mismatch. But at the same time, Spurs were designed in a lab to cover Brunson, as the Spurs have long wings who can play tight defense with no fear of dribble penetration - they just funnel that shit down to the big fella and most teams just recycle the possession.
Too close to call for me, but if pressed, I would pick Spurs in a tight series.
It's certainly going to be an interesting chess match.
Another view is that the Knicks benefitted immensely from the Pistons' Jalen Duren collapse that allowed a middling Cavs team to advance. The Pistons matched up well with the Knicks all year.
I also really like Hot Hand Theory’s take on how the Knicks can try to get Wemby out of the paint, and agree with them that Anunoby is probably their best defender against Wemby: https://youtu.be/INM1a-MDQHo?si=TFj7lahHXFQnKeOj