Video: Official Cleveland Indians “Harlem Shake”
March 13, 2013NFL News: Free Agent CB Keenan Lewis to visit Saints; Cason to visit Cardinals
March 13, 2013Last week, I was all excited about the NCAA tournament and breaking down the bubble picture. I’ll have a little bit more March Madness today, but also wanted to share a few other stats-related sports thoughts about the Cavs, Indians and Browns. Hope you enjoy.
— The Cavs’ much improved turnover ratio
While I’ve written about this topic several times already in the last month (here on 2/20, here on 2/26 and here on 3/2), I don’t think it’s still that clear to the average fan: The biggest in-season improvement for the Cavaliers has been in lowering their turnover rate.
For the first half of the season, or so, the Cavaliers were one of the worst with respect to turnover rate. This statistic is a percentage of the ratio of possessions that end up in a turnover. Since that awful start, the Cavs have suddenly become one of the best.
Here’s another look at the improvements, via NBA.com/stats:
Split | GP | TOR | Opp TOR |
Games 1-10 | 10 | 16.80% | 18.10% |
Games 11-20 | 10 | 16.30% | 15.90% |
Games 21-30 | 10 | 15.10% | 16.00% |
Games 31-40 | 10 | 14.30% | 16.80% |
Games 41-50 | 10 | 11.80% | 15.70% |
Games 51-60 | 10 | 14.10% | 16.60% |
Games 61-70 | 4 | 13.60% | 13.80% |
I’ve had much difficulty re-creating TOR as a statistic in a spreadsheet because of the differing game logs for team turnovers and possession counts. Either way, we’ll just stick with NBA.com/stats’ splits for now.
So here’s the change: 16.7% in first 24 games through 12/14 (with a 96.6 offfensive efficiency rating), and then 13.4% in the most recent 40 games (with a 105.1 offensive efficiency rating).
Intuitively, it’s pretty clear that a marginal increase/decrease in turnover rate should obviously at least increase offensive efficiency by the same digits. The average offensive efficiency in the NBA has historically been about 1 point per possession, so an extra possession not lost should equal at least a normal (or better) chance at scoring.
When you factor in that the Cavaliers have been a slightly worse rebounding team during these splits, and about the same at shooting efficiency, the only marked improvement is in turnover rate. Thus, that 3.3% improvement has likely been a multiplier effect on the 8.5 point jump in overall efficiency.
Now, next time I hear people talking about why the Cavs have been playing better1, I better hear folks starting to bring up this point again.
——————————————————————————————————————————-
— Regression to the mean and the 2013 Indians
I’ve been thinking about the Indians and their potential 2013 record for a while. This thought process obviously likely began with Jon’s way back article on Jan. 14, then my follow-up about the difference Michael Bourn will make this season. Kirk then continued the theme with his jWAR analysis of getting to the 86-win mark he’s been talking about for a while.
But one other non-WFNY article has had me thinking about this topic quite a bit too: A pretty short Bloomberg Stats Insights blog post back on March 2nd. The article was only 150 or so words plus a table, but featured a fascinating topic: Regression to the mean in recent baseball history.
So, being typical Jacob, I decided to do another quick-ish study. I looked at all of the seasons with less than a 70-win pace2 between 1985-2010, and what said team accomplished in the next two years. Here are the results:
Split | Yrs | W0 | W1 | W2 |
<=58.9 | 14 | 54.4 | 67.6 | 72.1 |
59-62.9 | 17 | 61.1 | 72.8 | 74.4 |
63-64.9 | 17 | 63.5 | 74.2 | 76.8 |
65-66.9 | 26 | 65.4 | 77.3 | 79.8 |
67-67.9 | 21 | 67.1 | 71.3 | 76.8 |
68-68.9 | 15 | 68.1 | 73.3 | 74.3 |
69-69.9 | 14 | 69.1 | 74.2 | 73.5 |
Total | 124 | 64.3 | 73.3 | 75.9 |
Um. What? Really? Those stats are kind of strange at first glance. Again, I looked at the seasons of a less than 70-win pace dating back to 1985 (represented by the average wins in column W0). Then, the next two seasons are averaged up into the columns of W1 and W2.
So, on average, these 124 seasons had 64.3 win-paces. The next year? A jump all the way up to 73.3 wins, on average, per the same 162-game schedule. So yes, on average, a team with less than 70 wins per 162 games has then improved by 9 wins per 162 games the next season.
There could be a whole variety of reasons for this historical average. But, to a certain extent, it’s very valid to just say that under 70-win seasons are pretty darn bad, and it’s almost quite difficult to repeat such seasons by pure luck. Regression to the mean works in both ways, as Kirk argued already.
Yet, I’m not really sure what I was hoping to accomplish with this quick study. Maybe more logical parameters for improvements of bad teams based on Cleveland’s 68-94 season in 2012? Possibly. Let’s take one more look.
%tile | Change |
0.0 | -14.8 |
0.2 | +0.8 |
0.4 | +6.0 |
0.6 | +10.9 |
0.8 | +19.0 |
1.0 | +35.0 |
This then shows the percentiles for the range of improvements from W0 to W1, as shown above, on a 162-game pace again. The worst change? A 14.8-win drop. The best change? A 35-win increase. The middle 20% of outcomes? Between a 6-11 win increase.
So more likely, I’d put the Indians around that at least. Six more wins equals 74; 11 more wins equals 79. Likely, because of the team’s drastic offseason improvements — both on the 40-man roster and in the manager’s office — they could lean toward the higher-end of historical improvements for 70-under win teams.
——————————————————————————————————————————-
— NFL Draft aggregate big board update
With NFL free agency hot on our tails, I thought I’d give another update today on the aggregate big board I’ve been showcasing several times thus far this offseason. I won’t go too far in-depth with my analysis, but just to mostly let the numbers talk for themselves:
Rk | Player | Scouts | ESPN | CBS | WF | MAR. | FEB. | JAN. | NOV. |
1 | Joeckel | 1 | 4 | 1 | 1 | 1.75 | 1.50 | 1.75 | 3.75 |
2 | Floyd | 2 | 1 | 3 | 2 | 2.00 | 9.25 | ||
3 | Fisher | 3 | 3 | 4 | 3 | 3.25 | 6.25 | 13.50 | |
4 | Milliner | 4 | 2 | 6 | 6 | 4.50 | 5.00 | 8.75 | 8.25 |
5 | Warmack | 5 | 9 | 2 | 8 | 6.00 | 6.50 | 5.75 | 11.25 |
6 | Jordan | 8 | 6 | 9 | 9 | 8.00 | 18.25 | ||
7 | Loutulelei | 6 | 12 | 15 | 4 | 9.25 | 3.50 | 2.75 | 2.50 |
8 | L Johnson | 13 | 10 | 7 | 10 | 10.00 | 14.75 | ||
9 | Ansah | 7 | 11 | 12 | 11 | 10.25 | 16.75 | ||
10 | Jones | 21 | 5 | 11 | 5 | 10.50 | 7.00 | 3.50 | 1.75 |
11 | Werner | 14 | 13 | 5 | 12 | 11.00 | 4.75 | 6.50 | 10.50 |
12 | Mingo | 12 | 7 | 13 | 17 | 12.25 | 11.50 | 13.50 | 8.75 |
13 | Cooper | 9 | 8 | 16 | 16 | 12.25 | 16.25 | 15.50 | |
14 | Richardson | 16 | 14 | 17 | 7 | 13.50 | 10.75 | 10.50 |
Obviously, the potentially unprecedented movement in the top range of big board prospects has continued here into mid-March. I featured only last month how I thought this might be a sensationally weak draft because of the turnover.
Florida DT Shariff Floyd, who wasn’t anywhere on my list in either November or January, now sits at No. 2. Oregon OLB Dion Jordan, Oklahoma OT Lane Johnson and BYU DE Ziggy Ansah all also moved into the top 10 after similarly mediocre early ratings.
The players who have bowed out: Texas A&M DE Damontre Moore and Texas S Kenny Vaccaro. Moore’s poor performance at the NFL Combine has him looking like a mid-to-late first-rounder, as opposed to a top 10 pick. Vaccaaro now is just not as hot as some of these other prospects in the 8-14 range.
Personally, I’m still not certain where I think the Browns should go with the No. 6 pick. There have been dozens of players linked to them during this offseason — from QBs like Geno Smith or Matt Barkley, to CB Dee Milliner, to then a wide range of DE/OLB prospects like Jordan, Moore, Ansah and Jarvis Jones.
Although the draft is only next month, there still seems to be a lot of uncertainty and disagreement about what the top 5 might look like, thus affecting how fans — and the Brown front office — might be able to prepare for their pick.
——————————————————————————————————————————-
— March Madness odds updates
Last week’s edition of The Diff covered my thoughts on constructing the bubble picture for this year’s NCAA tournament. Before I get into an update of this bubble — like I did on Monday — I first wanted to tackle one subject I purposefully ignored previously: Ohio State’s seeding.
At the time, I said they would likely be closer to a 4-5 seed after their second-half drubbing of Indiana. Averaging 4 of the 5 outlets’ projections that I used previously3, here is an average of the current rankings of 7-18 best teams:
3/12 | 3/9 | Team | Lunardi | Stevens | Palm | B’ville | AVG | ||||
7 | 10 | Mich. St. | E | 3 | S | 2 | E | 2 | E | 2 | 2.25 |
8 | 7 | New Mexico | M | 3 | E | 2 | S | 3 | S | 2 | 2.50 |
9 | 9 | Michigan | W | 3 | E | 3 | W | 2 | W | 3 | 2.75 |
10 | 11 | Miami | W | 2 | M | 3 | M | 3 | M | 3 | 2.75 |
11 | 8 | Florida | E | 3 | S | 3 | E | 3 | E | 3 | 3.00 |
12 | 15 | Ohio State | E | 2 | S | 4 | S | 4 | S | 3 | 3.25 |
13 | 16 | Arizona | E | 5 | W | 3 | W | 3 | W | 4 | 3.75 |
14 | 13 | Kansas State | W | 4 | E | 4 | M | 4 | S | 4 | 4.00 |
15 | 12 | Marquette | E | 4 | W | 4 | E | 4 | E | 4 | 4.00 |
16 | 19 | Oklahoma St. | M | 4 | M | 4 | W | 6 | M | 4 | 4.50 |
17 | 18 | Saint Louis | S | 5 | S | 5 | W | 4 | M | 5 | 4.75 |
18 | 14 | Syracuse | M | 5 | M | 5 | W | 5 | W | 5 | 5.00 |
Of course, here’s the math breakdown you were looking for: the 1-4 best teams are No. 1 seeds, 5-8 are No. 2, 9-12 are No. 3, 13-16 are No. 4 and 17-20 are No. 5.
So, previously, OSU actually was #15 and thus, closer to a true 4-seed. Now? With their win, and the losses by a few other teams, OSU now looks to be a possible 3-seed. Joe Lunardi’s most recent update from yesterday even had them as a No. 2.
Now, let’s update the bubble as I promised on Monday:
# | Team | Conf. | Lunardi | Stevens | Palm | B’ville |
1 | Temple | A-10 | IN | IN | IN | IN |
2 | Saint Mary’s | WCC | IN | IN | IN | Last 5 In |
3 | Villanova | BE | IN | Last 4 In | IN | IN |
4 | Iowa State | Big 12 | IN | IN | Last 4 In | Last 5 In |
5 | Boise State | MWC | Last 4 In | Last 4 In | IN | Last 5 In |
6 | La Salle | A-10 | Last 4 In | IN | Last 4 In | Last 5 In |
7 | Tennessee | SEC | First 4 Out | Last 4 In | Last 4 In | IN |
8 | Kentucky | SEC | Last 4 In | Last 4 In | Last 4 In | Last 5 In |
9 | Virginia | ACC | Last 4 In | First 4 Out | OUT | First 5 Out |
10 | MTSU | S Belt | First 4 Out | First 4 Out | OUT | First 5 Out |
11 | Alabama | SEC | Next 4 Out | Next 4 Out | First 4 Out | First 5 Out |
12 | Baylor | Big 12 | First 4 Out | Next 4 Out | OUT | First 5 Out |
13 | UMASS | A-10 | OUT | First 4 Out | First 4 Out | Next 5 Out |
14 | Ole Miss | SEC | First 4 Out | First 4 Out | First 4 Out | OUT |
I mentioned Middle Tennessee‘s entrance to the bubble on Monday, after their semifinal loss in the Sun Belt tournament. As of now, MTSU is just on the wrong end of the bubble — too poor of a strength of schedule, and just too many other opportunities for the other teams.
Previously, I also mentioned how Tennessee, Kentucky and Virginia seemed to be battling for the two final spots. UVA now appears to be the odd man out of that conversation.
As of now, again, these are the 8 final spots in the tournament. This could — and likely will — change based on the outcome of a few other conference tournaments. Most notably, I’d lean your eye toward the Atlantic 10 Conference, Southeastern Conference and Big 12 Conference.
In any of these conferences, we could see a bubble team — or worse — sneak up to win the title. That’d change the outlook here for these teams, possibly meaning there’d only be 7 free spots for the taking for at-large teams.
I’d mark Temple, St. Mary’s, Villanova and Iowa State as near-locks. Their resumes have been known for a while, and their impressive streaks of late should bring them into the tournament no matter what. The rest of the bubble then is up in the air, as we’ll know for certain within these next few days what the tournament will actually look like.
- As a reminder, they started 5-23. Thus, they are 17-19 since that point. [↩]
- Obviously, because of the strike-shortened seasons in 1994 and 1995, along with random 160- or 161-game seasons here or there, everything had to be per winning percentage. Then, I just multiplied everyone’s winning percentage to a 162-game pace for the most familiar numbers possible. [↩]
- Warren Nolan’s site appears to be down. Poor timing, unfortunately. [↩]