Mike Waters and Ken Pomeroy

[SWC75]

 

[SWC75]

 

[IthacaMatt]

That left me a little less impressed with his early season forecasts. Not really any analytics of the impact of change in coaches, just statistical average that teams get a little worse, and the best teams get worse by a greater degree than average teams, and he uses 5 year program historical average.

Or really, not too much thought to transfers and how they fit a new program. He didn't mention that at all, just that transfers are much harder to predict their contributions than returning players already in a program, who are generally projected to improve from year to year at a certain historical rate for that program.

Eh, I thought that he metrics were more advanced than that. Stats maybe have gotten more sophisticated than his model.

He spends about half the time explaining that his model is not predictive, beyond how a team will do in its next game, well, and yeah, maybe predict the end of the year, well, of course it's based on the end of the previous year, but still ...

 

[Orangezoo]

That left me a little less impressed with his early season forecasts. Not really any analytics of the impact of change in coaches, just statistical average that teams get a little worse, and the best teams get worse by a greater degree than average teams, and he uses 5 year program historical average.

Or really, not too much thought to transfers and how they fit a new program. He didn't mention that at all, just that transfers are much harder to predict their contributions than returning players already in a program, who are generally projected to improve from year to year at a certain historical rate for that program.

Eh, I thought that he metrics were more advanced than that. Stats maybe have gotten more sophisticated than his model.

He spends about half the time explaining that his model is not predictive, beyond how a team will do in its next game, well, and yeah, maybe predict the end of the year, well, of course it's based on the end of the previous year, but still ...

Honestly I'm not sure why anyone would take much stock in advanced stats pre season rankings outside trying to make some early season bets.. and even then..

 

[IthacaMatt]

Honestly I'm not sure why anyone would take much stock in advanced stats pre season rankings outside trying to make some early season bets.. and even then..

I just thought, "if you've been doing this for 20+ years, it seems like your computer model is still pretty simple."

Of course, I'm a human computer who created a statistically accurate track & field game (based on Strat-o-Matic baseball, and the odds of using two dice, etc.) back in the 1970s, before we had computers, except for those little Tandy deals that you would put together, but they couldn't really do much of anything.

Just used a calculator, the past 3-5 years of stats from Track & Field News, their player ratings at the end of each year, top races, top times, what their average race looked like, what was the mean, etc.

For distance runners, were they guys who waited and had a great kick (sprint) at the end of the race, or were they a fast pace setter, but didn't have that extra gear at the end of a race? You try to quantify those things, develop different ratings for different races, figure out time adjustments for shorter races like sprints. So I had a "pace" rating for the beginning of a distance race, and a "kick" rating for the last lap, and the odds associated with the dice rolls would try to reproduce that.

But back before the Internet, when you were older than going out bike riding and you started getting into sports, there was nothing else to do. No Internet, barely any cable TV at all. No video games until a few years later.

The 440 yard track was 44 boxes, each representing 10 yards. Each lane away from the inner lane added one space, so you could run staggered races where each runner stays in their lane for the 220 or the 440, and if you were caught on the outside of the pack in a distance race, you would have to run farther. It took me years, and I kept refining it for years. Great hobby; all the guys on my team would come over to play.

 

[BuffOrange]

I don't see what's surprising/disappointing about it. People complain when these geeks are too smug. Well he's basically admitting there's a lot of unknown in the new world of CBB. If the model is too simplistic then there's a lot of $ to be made given spreads basically line up with him for the most part.

I was glad he mentioned tournament selections shouldn't use his rankings. Some of his cult followers disagree which is insane. Otherwise why ever empty your bench.

 

[richmondcuse03]

I just thought, "if you've been doing this for 20+ years, it seems like your computer model is still pretty simple."

Of course, I'm a human computer who created a statistically accurate track & field game (based on Strat-o-Matic baseball, and the odds of using two dice, etc.) back in the 1970s, before we had computers, except for those little Tandy deals that you would put together, but they couldn't really do much of anything.

Just used a calculator, the past 3-5 years of stats from Track & Field News, their player ratings at the end of each year, top races, top times, what their average race looked like, what was the mean, etc.

For distance runners, were they guys who waited and had a great kick (sprint) at the end of the race, or were they a fast pace setter, but didn't have that extra gear at the end of a race? You try to quantify those things, develop different ratings for different races, figure out time adjustments for shorter races like sprints. So I had a "pace" rating for the beginning of a distance race, and a "kick" rating for the last lap, and the odds associated with the dice rolls would try to reproduce that.

But back before the Internet, when you were older than going out bike riding and you started getting into sports, there was nothing else to do. No Internet, barely any cable TV at all. No video games until a few years later.

The 440 yard track was 44 boxes, each representing 10 yards. Each lane away from the inner lane added one space, so you could run staggered races where each runner stays in their lane for the 220 or the 440, and if you were caught on the outside of the pack in a distance race, you would have to run farther. It took me years, and I kept refining it for years. Great hobby; all the guys on my team would come over to play.

Note to self -- never doubt Ithaca's thoughts on anything statistics related.

 

[Brooky03]

That left me a little less impressed with his early season forecasts. Not really any analytics of the impact of change in coaches, just statistical average that teams get a little worse, and the best teams get worse by a greater degree than average teams, and he uses 5 year program historical average.

Or really, not too much thought to transfers and how they fit a new program. He didn't mention that at all, just that transfers are much harder to predict their contributions than returning players already in a program, who are generally projected to improve from year to year at a certain historical rate for that program.

Eh, I thought that he metrics were more advanced than that. Stats maybe have gotten more sophisticated than his model.

He spends about half the time explaining that his model is not predictive, beyond how a team will do in its next game, well, and yeah, maybe predict the end of the year, well, of course it's based on the end of the previous year, but still ...

Kenpom's numbers do a better job of telling you how good a team was at the end of the season or how teams from different seasons compare to each other. It falls on its face a little as a forecasting tool, as he alludes to. I didn't watch much of the video but, if he said it's good for predicting the result of the next game, that's murky at best. You wouldn't want to implement a betting strategy based on Kenpom matchup predictions.

In my experience, Kenpom isn't worth looking at until it has at least 10 games of data in the current season to look at.

 

[sufandu]

That left me a little less impressed with his early season forecasts. Not really any analytics of the impact of change in coaches, just statistical average that teams get a little worse, and the best teams get worse by a greater degree than average teams, and he uses 5 year program historical average.

Or really, not too much thought to transfers and how they fit a new program. He didn't mention that at all, just that transfers are much harder to predict their contributions than returning players already in a program, who are generally projected to improve from year to year at a certain historical rate for that program.

Eh, I thought that he metrics were more advanced than that. Stats maybe have gotten more sophisticated than his model.

He spends about half the time explaining that his model is not predictive, beyond how a team will do in its next game, well, and yeah, maybe predict the end of the year, well, of course it's based on the end of the previous year, but still ...

I really don't care to hear math guys analyze sports. I prefer sports guys analyzing sports. Let me hear opinions about guys' talents and skills.

 

[Orangezoo]

I really don't care to hear math guys analyze sports. I prefer sports guys analyzing sports. Let me here opinions about guys' talents and skills.

As a math guy and a sports guy I whole heartedly agree lol

 

[IthacaMatt]

As a math guy and a sports guy I whole heartedly agree lol

In a funny coincidence, I was listening to a soccer match the other day (or maybe a post-game analysis show), and the speaker, a previously "all league" level player, talked about how advanced metrics had taken over the game, particularly in the area of player recruitment. He said, "Yes, you can learn things from those stats, but I prefer to judge a player's performance with my own eyes."

There are so many movements and runs and positions a player takes up on the field (or on a basketball court) that may be important to a team's set-up and tactics that just don't fit into those neat little analytical boxes, like "take-ons" or "successful dribbles" in soccer. The +/- rating tries to quantify some of those intangibles in U.S. sports like basketball, but most would agree that it's an imperfect measure.

And I guess to solidify that point, my team, Chelsea FC, in the recent past signed a guy from Germany who led the league in tackles as a fullback (a defender along the sidelines). Chelsea signed that guy, and he couldn't get into the first team and seldom played before he was sent away. Two other attacking guys, both also from Germany, were considered great rising attacking talents in the world and we signed them both for a gazillion dollars.

Of course, after 3 years, neither one approached the levels they had played at in Germany. In part this is because, top-to-bottom, the Premier League is the most competitive of the big European leagues, because they have the most television money. That allows even the mid-level teams to have a lot of good players on them, especially compared to 15 years ago when I first started watching.

Now, the advanced metrics can tell you the relative strength of these leagues around the world, so you can project "OK, when someone moves from Italy to Spain, there is this difference in how many fouls are called (players tend to be protected better in Spain, and kicked a lot in Italy, where until the last few years they play with much more focus on defense).

Or you can say, "goal scorers from Germany moving to England tend to see their production drop by about 20%", and the opposite would be true, as well. Harry Kane's best years in England produced 25-30 goals, but this year playing for Bayern Munich in Germany, he's currently on pace to get over 40 goals.

But those are just numbers, trends, etc. There is more to sports than just numbers.

 

[BuffOrange]

I really don't care to hear math guys analyze sports. I prefer sports guys analyzing sports. Let me here opinions about guys' talents and skills.

Counterpoint: espn. Everyone in those studios for every sport for the most part, is a sports guy who sucks at analyzing. If anything Mina Kimes is way better than Keyshawn Johnson.

 

[baggerbob]

In a funny coincidence, I was listening to a soccer match the other day (or maybe a post-game analysis show), and the speaker, a previously "all league" level player, talked about how advanced metrics had taken over the game, particularly in the area of player recruitment. He said, "Yes, you can learn things from those stats, but I prefer to judge a player's performance with my own eyes."

There are so many movements and runs and positions a player takes up on the field (or on a basketball court) that may be important to a team's set-up and tactics that just don't fit into those neat little analytical boxes, like "take-ons" or "successful dribbles" in soccer. The +/- rating tries to quantify some of those intangibles in U.S. sports like basketball, but most would agree that it's an imperfect measure.

And I guess to solidify that point, my team, Chelsea FC, in the recent past signed a guy from Germany who led the league in tackles as a fullback (a defender along the sidelines). Chelsea signed that guy, and he couldn't get into the first team and seldom played before he was sent away. Two other attacking guys, both also from Germany, were considered great rising attacking talents in the world and we signed them both for a gazillion dollars.

Of course, after 3 years, neither one approached the levels they had played at in Germany. In part this is because, top-to-bottom, the Premier League is the most competitive of the big European leagues, because they have the most television money. That allows even the mid-level teams to have a lot of good players on them, especially compared to 15 years ago when I first started watching.

Now, the advanced metrics can tell you the relative strength of these leagues around the world, so you can project "OK, when someone moves from Italy to Spain, there is this difference in how many fouls are called (players tend to be protected better in Spain, and kicked a lot in Italy, where until the last few years they play with much more focus on defense).

Or you can say, "goal scorers from Germany moving to England tend to see their production drop by about 20%", and the opposite would be true, as well. Harry Kane's best years in England produced 25-30 goals, but this year playing for Bayern Munich in Germany, he's currently on pace to get over 40 goals.

But those are just numbers, trends, etc. There is more to sports than just numbers.

The last time the Yankees won the World Series was 2009, before Cashman went to the advanced metrics.
That many years he would have been gone years ago if George was alive.

 

[IthacaMatt]

The last time the Yankees won the World Series was 2009, before Cashman went to the advanced metrics.
That many years he would have been gone years ago if George was alive.

Yes, I think all the advanced stats in baseball (launch angle, launch velocity, bat barrel speed, etc.) have messed up hitting. The only hits that matter in baseball now are home runs, just like the only baskets that count in hoops are 3 pointers and layups. The midrange game is a statistical "no, you shouldn't do that", just like in baseball, it seems like advancing the runner, the hit & run, stealing bases, all of the real art and strategy of baseball has been lost between every pitcher throwing 100 mph fastballs until their elbow breaks, and hitters who either can't make contact, or hit it out of the park. Advanced metrics have ruined baseball, plain and simple.

 

[Brooky03]

Advanced stats just try to boil intuition down to numbers. Kinda like how BMI tries to boil fatness down to a number. It gets you in the ballpark but not all the way there. The Rock’s BMI has to be sky high. Every moderately athletic D3 football player is probably in or close to obese territory.

 

[sufandu]

Advanced stats just try to boil intuition down to numbers. Kinda like how BMI tries to boil fatness down to a number. It gets you in the ballpark but not all the way there. The Rock’s BMI has to be sky high. Every moderately athletic D3 football player is probably in or close to obese territory.

According to BMI nearly every football player is at least overweight. Going by SU's measureables, Damian Alford is barely normal at a BMI of 24.3. What's crazy is that at 160 lbs he wouldn't be underweight.