Barttorvik.com: A better and freer version of Kenpom

[proelitedota]

http://www.barttorvik.com/

My favorite feature of the site that Kenpom doesn't is the team trends

http://www.barttorvik.com/trends.php?team=Duke&y=2018&type=adjo
http://www.barttorvik.com/trends.php?y=2018&team=Duke&type=adjd

You can see a correlation between our switch to the zone and the downward trend of our offense. Therefore zone defense is making us worse on offense! You can't explain that!

:cool:

[duke2x]

Thank you so much! I like that his analysis is free and written for people who stopped at STA 100 at Duke.

I never knew Rick Astley had so much insight into basketball analytics going all the way back to 2008.

[WakeDevil]

Cincinnati is ranked ahead of Michigan State and UNC. Sure.

[proelitedota]

Cincinnati is ranked ahead of Michigan State and UNC. Sure.

Kenpom has Cincinnati between MSU and UNC.

[uh_no]

http://www.barttorvik.com/

My favorite feature of the site that Kenpom doesn't is the team trends

http://www.barttorvik.com/trends.php?team=Duke&y=2018&type=adjo
http://www.barttorvik.com/trends.php?y=2018&team=Duke&type=adjd

You can see a correlation between our switch to the zone and the downward trend of our offense. Therefore zone defense is making us worse on offense! You can't explain that!

:cool:

I see he made some improvements since the last time this was posted and was widely criticized for the eye-bleed inducing style sheet! Impressed with it.

[robed deity]

http://www.barttorvik.com/

My favorite feature of the site that Kenpom doesn't is the team trends

http://www.barttorvik.com/trends.php?team=Duke&y=2018&type=adjo
http://www.barttorvik.com/trends.php?y=2018&team=Duke&type=adjd

You can see a correlation between our switch to the zone and the downward trend of our offense. Therefore zone defense is making us worse on offense! You can't explain that!

:cool:

Great site. It's my go-to because I'm too cheap to pay for kenpom. I especially like the conference only option.

[cptnflash]

Kenpom is the best $20 I spend every year. By a mile. Seriously. It’s $20. Per year.

[jv001]

http://www.barttorvik.com/

My favorite feature of the site that Kenpom doesn't is the team trends

http://www.barttorvik.com/trends.php?team=Duke&y=2018&type=adjo
http://www.barttorvik.com/trends.php?y=2018&team=Duke&type=adjd

You can see a correlation between our switch to the zone and the downward trend of our offense. Therefore zone defense is making us worse on offense! You can't explain that!

:cool:

Thanks, I like this site. yes, being free helps. Just noticed Alex is shooting very well: 54% on 3s and 74% FTs.
Wendell is shooting well: 49% 3s and 73% on FTs. Looks like both could be used to open up the clogged lane for us. GoDuke!

[Hingeknocker]

This is the first year that I have used Bart Torvik's site, and it's really grown on me. At first I thought it was too much of a KenPom clone, in interface and how the rankings are compiled. But I've come to love all of the additional rankings, simulations, tournament projections, and historical comparisons that he includes. The fact that it's free is the cherry on top. He's also a good Twitter follow, if you can put up with some Wisconsin/Big 10 heavy content at times.

Oh, and of course the fact that Torvik has Duke at #2 compared to KenPom's #3 shows how much better his rankings are this year :).

[proelitedota]

I have found that Bartovick has more of a recency bias due to accounting for team trends that Kenpom does not. It's a better predictive tool for predicting games a week out versus Kenpom, which is better for predicting the rest of the season.

[COYS]

Thanks so much for sharing this. The site does come across as a KenPom clone on a superficial level, but it really does offer a few cool wrinkles. One extra stat that is particularly interesting is the G-score (http://adamcwisports.blogspot.com/2015/11/introducing-g-score.html). It's basically Torvik's way of rating how a team played in an individual game. A team that posts a 100 g-score in a game plays like a team with a .999 pythagorean winning percentage. A team that posts a low score like a 20 would only have a .200 pythagorean winning percentage. While even he admits it's not absolutely perfect (the g-score does not account for blowouts or how well the benchwarmers do at the end of a big win), it is loosely a way to measure which games are a teams best (or worst) performances.

To me, this puts some of Duke's losses this season into much better perspective. For example, in the loss at VaTech on Monday, Duke posted an 88 g-score. If Duke played that way the whole season, we'd be ranked 27th in Torvik's rating system, one spot above Louisville. Now, obviously Duke has higher aspirations than being ranked 27th (we're currently second in Tovik's rankings), but it really wasn't THAT bad of a game. It certainly was not as bad as some of our losses earlier in the season. The BC and NC State losses were both 69's (which would put us right around 100th in his rankings if we were to play that way all season) and the St. John's loss had a G-score of 73. So while the loss was disappointing, the performance was not nearly as bad as some of our losses earlier in the season. Going back over the previous 10 seasons (the "Torvik era"), the loss that it most closely resembles is the 2015 loss @Maryland to Greivis Vasquez, in which Duke posted an 87 g-score. And while we have been frustrated with this team due to its inconsistency, it is worth noting that our teams that had the three best performances in the NCAAT (2010, 2013, and 2015) all suffered losses in which the team played significantly worse than this year's team has in any game this season.


I went back through the g-scores of the previous seasons quickly and discovered that the worst loss of the past ten years was not the Mercer game, which surprised me a little (although that g-score wasn't good at all). Instead it was the 2009 beating we endured at Clemson in which Duke earned a g-score of 11 (which meant we played like the 320th best team in NCAA basketball for that game . . . that's BAD!).

Meanwhile, we Duke fans are incredibly spoiled because Duke has put up quite a few "perfect" 100 g score games. This season's lone 100 g score game against a major opponent was the Louisville game. Other notable "perfect" games from past seasons include the Gonzaga and Mich St. Elite 8 and Final Four games en route to the 2015 title, 82-50, and the WVU Final Four game in 2010. The team with the most perfect games in a season is the 2015 team with 7. 2010 comes in second with 6. This year's team has 3 so far, but here's hoping we add to that number starting on Saturday against the Cheats.

[robed deity]

Thanks so much for sharing this. The site does come across as a KenPom clone on a superficial level, but it really does offer a few cool wrinkles. One extra stat that is particularly interesting is the G-score (http://adamcwisports.blogspot.com/2015/11/introducing-g-score.html). It's basically Torvik's way of rating how a team played in an individual game. A team that posts a 100 g-score in a game plays like a team with a .999 pythagorean winning percentage. A team that posts a low score like a 20 would only have a .200 pythagorean winning percentage. While even he admits it's not absolutely perfect (the g-score does not account for blowouts or how well the benchwarmers do at the end of a big win), it is loosely a way to measure which games are a teams best (or worst) performances.

To me, this puts some of Duke's losses this season into much better perspective. For example, in the loss at VaTech on Monday, Duke posted an 88 g-score. If Duke played that way the whole season, we'd be ranked 27th in Torvik's rating system, one spot above Louisville. Now, obviously Duke has higher aspirations than being ranked 27th (we're currently second in Tovik's rankings), but it really wasn't THAT bad of a game. It certainly was not as bad as some of our losses earlier in the season. The BC and NC State losses were both 69's (which would put us right around 100th in his rankings if we were to play that way all season) and the St. John's loss had a G-score of 73. So while the loss was disappointing, the performance was not nearly as bad as some of our losses earlier in the season. Going back over the previous 10 seasons (the "Torvik era"), the loss that it most closely resembles is the 2015 loss @Maryland to Greivis Vasquez, in which Duke posted an 87 g-score. And while we have been frustrated with this team due to its inconsistency, it is worth noting that our teams that had the three best performances in the NCAAT (2010, 2013, and 2015) all suffered losses in which the team played significantly worse than this year's team has in any game this season.


I went back through the g-scores of the previous seasons quickly and discovered that the worst loss of the past ten years was not the Mercer game, which surprised me a little (although that g-score wasn't good at all). Instead it was the 2009 beating we endured at Clemson in which Duke earned a g-score of 11 (which meant we played like the 320th best team in NCAA basketball for that game . . . that's BAD!).

Meanwhile, we Duke fans are incredibly spoiled because Duke has put up quite a few "perfect" 100 g score games. This season's lone 100 g score game against a major opponent was the Louisville game. Other notable "perfect" games from past seasons include the Gonzaga and Mich St. Elite 8 and Final Four games en route to the 2015 title, 82-50, and the WVU Final Four game in 2010. The team with the most perfect games in a season is the 2015 team with 7. 2010 comes in second with 6. This year's team has 3 so far, but here's hoping we add to that number starting on Saturday against the Cheats.

I remember that Clemson '09 game vividly, and thinking it was one of the only times a K team seemed to quit.

Like you, I recently perused the "g score" history and was surprised to see how low some in past years were, notably NC State 2010, and Miami and NC State 2015.

[wobatus]

I see he made some improvements since the last time this was posted and was widely criticized for the eye-bleed inducing style sheet! Impressed with it.

Haha, yeah, I posted a link earlier this year and some of the reaction was ow, me eyes.

It really does have a lot of fun little features. I like points above replacement player adjusted for usage, and you can break it down to season long, conference-only, against top 100, top 50. And as mentioned above you can do conference breakdowns, trends for team, and for each player.

[-jk]

And he doesn't take himself too seriously (even if he misses the geek plural):
Compiled by Bart Torvik from sacred data and secret formulas.

-jk