Hubert Davis Catch-all

[Calheel]

I would absolutely agree with you that if someone were arguing that metrics are "infallible" then what you are talking about would undercut their "infallibility." But luckily no one is arguing that! The metrics are not infallible. They are a fairly rough attempt to measure team quality - for numerous reasons, I don't think anything other than a very rough measurement is possible for college bball teams. I do think they accomplish their function of predicting outcomes better than any other system I've seen, even if individual game outcomes very frequently are far different from what the metrics would predict.

There is simply no such thing as "certitude" when it comes to comparing quality of teams and predicting outcomes in any sport, especially college basketball. So anyone who is looking for "certitude" from metrics is going to be disappointed, that's for sure. If anyone is using them to express "certainty" about anything, they should be mocked. But I would also mock anyone who thinks their own personal "eye test" is betetr than efficiency metrics at ranking every team in college bball.

The best validator of metrics is their widespread use in point spreads.

Oddsmakers would soon be out of business if they repeatedly posted soft lines. But they don’t. They set lines based on power metrics.

If metrics were unreliable, Vegas would be bankrupt.

 

[finesse]

The best validator of metrics is their widespread use in point spreads.

Oddsmakers would soon be out of business if they repeatedly posted soft lines. But they don’t. They set lines based on power metrics.

If metrics were unreliable, Vegas would be bankrupt.

Oddsmakers base their odds on what makes the money balance. If it's metrics today, that's fine with them. If the money balances and they get their piece they don't care. Yes, it's a selling point for it to be accurate but that's not nearly as essential what people believe.

 

[Enigma94]

The best validator of metrics is their widespread use in point spreads.

Oddsmakers would soon be out of business if they repeatedly posted soft lines. But they don’t. They set lines based on power metrics.

If metrics were unreliable, Vegas would be bankrupt.

And UNC is 17-13 against the spread this year, which is another data point indicating they are under ranked in the metrics.

 

[Calheel]

Oddsmakers base their odds on what makes the money balance. If it's metrics today, that's fine with them. If the money balances and they get their piece they don't care. Yes, it's a selling point for it to be accurate but that's not nearly as essential what people believe.

Incorrect. I know a lot about sports betting and if oddsmakers did what you said, they would be out of business.

You are partially correct in that lines do move based on public action but: (1) that is not relevant to opening lines, and (2) books rarely have 50/50 on bets nor do they necessarily want that.

Markets are based on metrics. Adjustments are made for injuries and the like. But oddsmakers are not using the “eye test” to set lines.

 

[heelinhell]

Efficiency metrics are best used as predictive tools and in tie-breaking/hair splitting scenarios.

I'm not sure they should play much of a role in actual seeding, those should be heavily determined by actual wins and losses weighted by the quality of those Ws and Ls (in my opinion).

At the end of the day winning the game is what matters.

Obviously, this guy fails to understand. Wins and losses don't matter. It's how you win the game or how you lose the game.

Give me a good loss against dook** over a bad win against VT any day and twice on Sunday...

 

[PTOHeel]

This is a great Hubert Davis discussion. LOL. Just kidding. Carry on.

 

[Bigs23]

The best validator of metrics is their widespread use in point spreads.

Oddsmakers would soon be out of business if they repeatedly posted soft lines. But they don’t. They set lines based on power metrics.

If metrics were unreliable, Vegas would be bankrupt.

No they wouldn't. They would adjust the lines according to betting trends.

Bookies were making bank long before analytics. Their eye tests were pretty profitable.

I do agree that these metrics are actually a great tool for book making. Not so much with NCAAT seeding.

 

[duluoz]

But oddsmakers are not using the “eye test” to set lines.

They absolutely used to use the eye test primarily, and they didn’t nearly, remotely go out of business.

They only started using statistical analysis once they got burned enough by sharps who were using those methods. Or they flat out hired them to set their lines.

Now that literally anyone who wants can have access to the same exact data as bookmakers, the eye test is just as critical a differentiator as it was in the old days.

 

[SnoopRob]

You might as well just say winning the game doesn't matter. And you think losing to ncst without our two best players and projected first round draft picks in a rivalry game is worse than losing to Pitt? I just can't get on board with that. It's ok, this is what I was talking about when I said some people think the metrics are infallible.

Winning the game does matter when you're evaluating how good of a season a team is having, but isn't as relevant when you're discussing how good a team is (in essence, the difference between predictive metrics and resume metrics).

Based in advanced metrics and Vegas odds, we can say that State is roughly 9-10 points (on a neutral court) better than Pitt. So losing to Pitt (at Pitt) by 1 isn't nearly as bad a loss as losing to State (at State) by 24. And it just makes sense at a basic level of logic, a blowout is going to be a much worse loss than a 1-point loss in pretty much every situation.

And none of that means that anyone thinks the metrics are infallible. That's simply a rhetorical device you're using to paint others in a negative light that has nothing to do with those folks and everything to do with you.

 

[Calheel]

They absolutely used to use the eye test primarily, and they didn’t nearly, remotely go out of business.

They only started using statistical analysis once they got burned enough by sharps who were using those methods. Or they flat out hired them to set their lines.

Now that literally anyone who wants can have access to the same exact data as bookmakers, the eye test is just as critical a differentiator as it was in the old days.

Please let me know what books use the eye test. I am ready to retire from the practice of law anyway.

 

[SnoopRob]

And UNC is 17-13 against the spread this year, which is another data point indicating they are under ranked in the metrics.

17-13 is two games off of .500, which means that the oddsmakers have done a pretty good job balancing the lines involving Carolina. The difference is likely attributable to single-game variance and that as a "brand" school Carolina picks up bets that the average team wouldn't get.

It does very, very little to suggest that we're under ranked in the metrics.

 

[Enigma94]

Winning the game does matter when you're evaluating how good of a season a team is having, but isn't as relevant when you're discussing how good a team is (in essence, the difference between predictive metrics and resume metrics).

Based in advanced metrics and Vegas odds, we can say that State is roughly 9-10 points (on a neutral court) better than Pitt. So losing to Pitt (at Pitt) by 1 isn't nearly as bad a loss as losing to State (at State) by 24. And it just makes sense at a basic level of logic, a blowout is going to be a much worse loss than a 1-point loss in pretty much every situation.

And none of that means that anyone thinks the metrics are infallible. That's simply a rhetorical device you're using to paint others in a negative light that has nothing to do with those folks and everything to do with you.

You keep bringing up the ncst game like it's the huge point, but you also keep avoiding the fact that our two best players didn't play in that game. The team we put on the floor that night was nothing like the team we really are. So if your best argument that Ohio State is as good as UNC is that the ncst game was a worse loss than Ohio St losing to Pitt then I'm not buying it.

 

[duluoz]

Please let me know what books use the eye test. I am ready to retire from the practice of law anyway.

You’re right, I’m sure no books ever incorporate the eye test to differentiate anywhere along the way. The machine spits it out and no human actually eyeballs it for sanity checks even, they just post it blindly. Good point.

 

[SnoopRob]

This is a great Hubert Davis discussion. LOL. Just kidding. Carry on.

This is essentially HD meta-discussion that lays bare a significant difference between HD supporters and HD detractors as there are clear overlaps between being an HD supporter and suspicious of the metrics and being an HD detractor and favoring of the metrics.

 

[Calheel]

You’re right, I’m sure no books ever incorporate the eye test to differentiate anywhere along the way. The machine spits it out and no human actually eyeballs it for sanity checks even, they just post it blindly. Good point.

They have bookmakers, it is not just AI. And they rely on multiple data sources. But if you think lines are being set on hunches, I just don't know what to tell you.

The best gamblers have the best metrics, better than KenPom. As do the books. And they are not saying, well the line should be -5, but I saw that look in Caleb Wilson's eyes and Hubert is clutch at the end of games, so let's set it at -3.

 

[SnoopRob]

You keep bringing up the ncst game like it's the huge point, but you also keep avoiding the fact that our two best players didn't play in that game. The team we put on the floor that night was nothing like the team we really are. So if your best argument that Ohio State is as good as UNC is that the ncst game was a worse loss than Ohio St losing to Pitt then I'm not buying it.

It's one data point that I brought out, you're the one who decided to focus on it beyond 1 data point within a much larger season.

My best argument is that when you look at each team's seasons in totality, it is understandable why they would be roughly equal in the predictive metrics.

But you focus on these 2 games and all of the outlier events surrounding one of them in order to keep from having to deal with the overarching quality nature of predictive metrics.

 

[Enigma94]

17-13 is two games off of .500, which means that the oddsmakers have done a pretty good job balancing the lines involving Carolina. The difference is likely attributable to single-game variance and that as a "brand" school Carolina picks up bets that the average team wouldn't get.

It does very, very little to suggest that we're under ranked in the metrics.

17-13 is a huge variance in betting. Professional gamblers would love to hit that kind of spread. And if "brand" was affecting the lines don't you think Vegas would account for that? Seems like if brand created some kind of mismatched line then gamblers would take advantage of that.

 

[duluoz]

This is essentially HD meta-discussion that lays bare a significant difference between HD supporters and HD detractors as there are clear overlaps between being an HD supporter and suspicious of the metrics and being an HD detractor and favoring of the metrics.

How does Wes fit into the meta-discussion? The HD-haters have clear overlaps with the pro-Wes guys at the time of Roy’s retirement.

Were the metrics behind the Wes support, gasp, inaccurate?? Impossible!

 

[Calheel]

This is essentially HD meta-discussion that lays bare a significant difference between HD supporters and HD detractors as there are clear overlaps between being an HD supporter and suspicious of the metrics and being an HD detractor and favoring of the metrics.

Interesting observation. And seems to be fairly accurate.

Ironic since metrics were the only thing that got Hubert and the boys into the tournament last year.

 

[SnoopRob]

17-13 is a huge variance in betting. Professional gamblers would love to hit that kind of spread. And if "brand" was affecting the lines don't you think Vegas would account for that? Seems like if brand created some kind of mismatched line then gamblers would take advantage of that.

Vegas sets lines that, after adjustment, are designed to bring in roughly equal amounts (dollars not number of bets) on each team. There are a lot of factors that influence the amount of bets (both dollars and number of bets) on particular games and on particular schools...and the "brand" of each team is one of those factors. (As a blue blood school, Carolina gets bets on every game that wouldn't otherwise be made by Carolina fans wanting to bet on their team.) But it is just one factor and doesn't dominate the other factors.

Vegas doesn't really care how Carolina does against the spread as long as the bets come out roughly even on each side and they don't get swamped with a tough loss.