Jeremy Lin helps teach us about Yay Points, Clutch and Risk

Posted on by
3

This season New York has been front and center when it comes to a lot of our basketball posts. With Wages of Wins Favorites Carmelo Anthony and Tyson Chandler (albeit popular by us for opposite reasons) joining forces and up-and-comer phenom Landry Fields teaming up with aging like milk Amare Stoudemire, there were simply too many interesting reasons to avoid the Knicks this season. That’s why it’s surprising that the biggest story this season from the Knicks is none of those players.

The story that has taken the NBA blogosphere by storm is Jeremy Lin. We’ve talked about the draft and how it is perfectly acceptable to say no one could have seen Lin coming. After all most young players — even future greats — don’t play well in their first starts. But what is not acceptable to say is that no team should have realized Lin was worth a look. In a league where mystery box players such as Bargnani get major minutes as young players, it is ridiculous to say the Warriors, Rockets or even Knicks were right in sitting Lin on the bench. However, with Lin now front and center, I am happy to say there are even more great lessons to be learned from this already amazing story.

Lin and Yay Points

Points drive perception in the NBA. Points get you drafted high, paid, voted to All-Star games and even sways voters for major awards. It is true that to win in the NBA you have to outscore the opponent, and this is a matter of points. Of course, missing shots gives the opponent an opportunity to score. As does turning the ball over. In Lin’s first three starts he scored 28, 23 and 38 points while shooting over 60% true shooting. That’s remarkable! In his last two games he’s kept up the 20+ points per game but he had a 33.3% true shooting percentage in the win over Minnesota and 54.3% true shooting percentage in the win over Toronto — also had Lin not scored his last second clutch shot he would have ended the night with a below average 50.3% true shooting percentage. Those marks are terrible and above average respectively. He’s also been racking up the turnovers. Of course, the focus is still on points. Lin shoots amazingly? He’s amazing! Lin shoots terribly? He’s still amazing! Lin shoots above average? Still amazing! People like winners. The Wages of Wins showed the biggest factor in ticket sales is a winning team. However the biggest thing people like to give credit to is scoring. Lin is scoring a lot of points on a currently winning team. Players like Joe Johnson show this is a good way to boost your popularity.

Lin and Clutch

Toronto has been a bad team this season. However, recently they’ve been a great opponent for some spectacular play (perhaps we should rename them the Washington Generals?) Both Kobe and Lin were able to sink last second clutch shots to will their team to a win. Of course, neither Kobe or Lin actually played that well. As I mentioned, Lin needed the clutch shot to get his scoring percentages above average and with 8 turnovers to his 11 assists, it’s not like he was being a great facilitator. Kobe’s game was even worse. He shot 9 for 23 and did little else. Yet at the end of the game, both players were lauded as heroes. The True Hoop blog actually had this to say about Kobe:

Sunday was a day in which Kobe Bryant made like Magic Johnson…

Wait…what? 9-23 shooting with two rebounds and four turnovers is like Magic? This wasn’t an article written by Henry Abbott and thank goodness Abbott did take another look at this game. That said, on the network that has been pounding the Kobe isn’t clutch drum, all it takes is one game against a bad team with a bad performance by a player that gets the game winner to draw comparisons to one of the greatest players ever? Yahoo Sports was similar in their talk of Lin and his game winner in Toronto:

Even after his amazing week, this one took Linsanity to a whole new level.

Again, a game with a below average performance against a bad team somehow takes Lin’s Michael Jordan like start (seriously compare Lin’s third start to Michael Jordan’s third start) to the next level? I enjoy clutch and excitement on Sports Center, but both of these claims are hyperbole to the Nth degree!

Lin and risk

The rise of Lin has brought out a lot of speculation and finger pointing. How could the Warriors cut Lin? (The answer by the way was so they could not sign DeAndre Jordan) How could the Rockets cut Lin? (The answer was to make room for Dalembert and to keep warm bench spots open for Jonny Flynn and Hasheem Thabeet) How could the rest of the league pass on him?

The issue in the NBA is this: playing an unknown player while your team is losing is a risky proposition for coaches. Much like going for it on fourth down in football will draw scrutiny in a loss, so will playing unknown players in front of known players. While everyone scurries to prove that Lin will be the greatest player since MJ or that they could have known he was coming (we’re somewhat guilty of that), there’s a bigger issue. Lin is proof of an inherent flaw in NBA management. Owners refuse to quit because they’ve sunk millions of dollars into players. Coaches refuse to play unknown players because of a risk to their job.

Lin is the very definition of a low-risk high-reward proposition. This season he will make less than $1 million. Compare that to even the most fringe free agent point guard signed in Earl Watson, who will make $2 million a year. Lin’s college numbers also suggested he was worth a look. On a losing team (such as Golden State), such a player should be great; you play him, and if he works out you’re better off. If he fails, then your cap is intact and you can roll the dice again in the draft. Yet the way the NBA management is structured, somehow playing a cheap, undrafted player is seemingly high risk. I say seemingly because, as Arturo has pointed out, if you are on a losing team your only hope of salvation is winning. That’s a problem that I suspect won’t be solved even after all the dust has settled and the finger pointing in regards to Lin has stopped.

Summing up

Let’s just get this out of the way: sample size. I didn’t even need to form a complete sentence around those two words. Lin has started less than ten games. All the narratives we are building stand a very good chance of looking foolish in the future. All I can say is that in all aspects (fun to watch, fun to talk about, fun to analyze the stats) Lin has been a joy for NBA fans. As the NBA is about entertainment that’s really all we can ask for. Regardless of how this plays out in regards to Lin’s career or NBA front office decisions, what we can say is that for two weeks in February Jeremy Lin was the most polarizing person in the NBA, and that in itself is pretty damned impressive.

-Dre

 

Jeremy Lin and the Ghost of NBA Draft’s Past

Posted on by
12

“Science is the belief in the ignorance of experts.”
― Richard P. Feynman

Yes, before you ask, as is contractually required by any and all bloggers I will be talking about the unlikely Jeremy Lin. Now, I know we touched on this yesterday but our goal today is different. My take will be different. You see rather than waxing poetic about the unbelievable and unpredictable nature of basketball or focusing on how no one could have seen this coming, I’m going to focus on how we kind of did.

Because when faced with a supposedly unsolvable problem, we brought the science and science once again beat the experts.

The problem I’m alluding to is evaluating talent in the NBA draft. Anyone who knows me knows I love to write about the draft. For those who don’t, hello you must be new here. Just in case, let me illustrate that by throwing some links up for your viewing pleasure.

This lead to a lengthy draft strategy segment in my guide to running an NBA franchise (Build me a winner rev.2).

The key takeaway was that talent was that I needed to build an effective draft model to predict player performance based on publicly available data. I built two (go here for the model build parts 1 & part 2 ). In very general terms the models use the available data to predict future performance for each player coming into the draft from college. Based on that prediction a ranking is done and a draft recommendation is generated.

Now this model is a work in process, I build it then publish it then go back at some future point to review to see if it worked. I will make corrections as needed over time.

One of the key ideas is having a public build to allow for peer review and answer the skeptics.

For the purposes of this discussion for example I will focus on the last 2010 build (see here) because at the request of some of our loyal readers I had included the best undrafted rookies. Can you guess who was number one?

Do you want to answer for the class?

Mr. Lin actually was the number tenth overall ranked prospect on our draft board and easily the best undrafted. The model had him slightly below the draft treshold. Given this and a few other similar data points, I moved the treshold slightly down  to .090 WP48 for Model #1 and .060 WP48 for model #2. You will see the results of that in the numbers that follow.

Why should you care exactly?

It’ll make more sense if I just give you the full story:

That’s every drafted player coming from the NCAA’s from 1995 thru 2010 who’s played at least 400 minutes in the NBA (2010 shows additional players who haven’t played those minutes yet). It shows the player’s draft year, where he was picked, the model predictions and the player actuals for his first 4 years and his career. For 2010 for example, we can see both the Knicks starting guards in the top 10 but this could simply be coincidence. Did the models actually do anything?

A simple test is to look at correlation between the place the player was picked, where the models suggested picking him and actual rank by draft in terms of production. Draft order vs production shows minimal correlation with an R-square of about 5%. It jumps to 25% for the predicted production rank.

A more complex and interesting test is to look at:

  • The probability of landing a better than average player (>.090 WP48)
  • The probability of landing a good player (>.150 WP48)

If I do this for all picks by the Models as well as  all draft picks and Model picks taken after the top 5  picks I get:

The models perform as well or better than the majority of lottery picks.  The only real difference is superstar talent at the number one pick (which isn’t really an every year affair).

So to review, using publicly available data we built a model that picks draft winners at a 75% rate which is better in general than having the #1 pick in the draft and big winners at a 40% rate which is better than everything but the #1 pick.

Science!

-Arturo

P.S. How about one more bonus table?

 

 

Ball Hogs, Long Meetings, and More on Jeremy Lin

Posted on by
2

My latest for Freakonomics – Ball Hogs and Long Meetings – allows me to express my general dislike of meetings (a sentiment — as my colleagues at Southern Utah University know — I express at every meeting I attend).

In addition to expressing my dislike of meetings, my post also illustrates the consequences of failing to measure performance accurately.  My intent was to illustrate that this issue goes beyond sports.

Beyond my post at Freakonomics, I also wanted to note the latest from Patrick Minton.  Patrick’s work can be thought of as a companion piece to what Greg Steele posted on Jeremy Lin (and both are well worth reading).

My next post at Freakonomics might be on Jeremy Lin.   At least, if I can find something to say beyond what Greg and Patrick said, my next post might be on Linsanity.

- DJ

Who Could Have Known About Lin?

Posted on by
12

The following comes from the talented Greg Steele (aka the Man of Steele). Greg normally writes about the Houston Rockets but as Rockets GM Daryl Morey decided to comment on Jeremy Lin, Greg felt compelled to reply.

It’s not often that a polished Harvard grad gets the short end of the stick. Yet that’s precisely what happened to one graduate of that distinguished institution. Jeremy Lin completed his four-year career at Harvard and prepared to enter a career with extremely high salaries. Despite his pedigree, Lin was not able to catch on in the business for an entire year after graduation.

Of course, Lin’s chosen career field was fairly exclusive: he chose to go into professional basketball. Although Lin accumulated a Position-Adjusted Win Score of 12.96 in his senior year at Harvard, he went undrafted in the 2010 NBA Draft. The average PAWS of college players selected in the NBA Draft is 10.18, so it seems relatively clear that Lin was a good prospect. However, there are two extenuating circumstances:

  • College productivity does not always lead to NBA productivity
  • Lin played in the Ivy League, midmajor or minor conference, and therefore faced weaker competition than many of the players who were drafted

Factor 1) above applies to all prospective rookies, not just Lin, so we’ll set it aside for now. Now, on to factor 2). Let’s say we discount Lin’s PAWS by 10% to account for the fact that his competition was fairly weak. While we’re at it, let’s also discount Lin’s productivity by another 10% to reflect the fact that he entered the draft pool as a senior, and thus had less potential to improve than younger players. So, if we go ahead and discount 20% of Jeremy Lin’s collegiate productivity, he is left with an Adjusted Position-Adjusted Win Score of 10.38 – still above the average production level of an NBA player in college. Actually, there’s no way we should even have an unofficial metric whose acronym is APAWS, since that sounds more like an endearing term for someone’s grandfather than a basketball metric. Instead, let’s call this second number Jeremy Lin’s Prospect Estimate.

Still, nobody saw Jeremy Lin until the past week. In each of his last four games, Lin has scored at least 23 points and handed out at least 7 assists, substantially bolstering the Knicks’ fortunes. In 209 minutes this year, Lin has recorded a 0.256 WP48, well above the level a star player achieves over the course of a season. Still, 209 minutes is a small sample size. How else could we have seen Jeremy Lin coming?

Well, last year Arturo came up with some absolutely smashing rookie performance prediction models (Editor’s note: More detail on that here) . These models, based in part on college performance and in part on preseason numbers, predicted that Lin was a good prospect, with a rookie year WP48 predicted to be somewhere between .055 and .090.

Though Lin only played 285 minutes last season, he managed to produce .157 wins per 48 minutes during that playing time, so I guess you could say that Arturo saw Jeremy Lin coming.

Still, surely Lin was only on the radar of the Wages of Wins network. Houston Rockets GM Daryl Morey, who signed and then waived Jeremy Lin without ever letting him play in a regular season game, said that nobody saw him coming. The New York Knicks don’t seem to have had very high expectations for Lin, since they acquired Baron Davis specifically in order to share the point guard position with Toney Douglas. Obviously the Golden State Warriors didn’t see him coming, since they waived him after last season. It would seem that we have to figure that the Knicks were lucky. After all, nobody really saw him coming.

So here’s someone who was interested in Jeremy Lin in August of 2010. Okay, I give up on the thesis that nobody saw Jeremy Lin coming. Even in the blogosphere, some people thought Jeremy Lin was a viable NBA player. Let’s shift to another question. How could anyone see Jeremy Lin coming? His sample size in the NBA is so small, what other numbers, besides his college performance, can we use to evaluate Jeremy Lin?

In 636 minutes in the D-League last year, Lin put up a 0.211 WP48, totaling 2.8 wins produced. So, even when we search long enough to find a statistically significant sample size (more than 400 minutes), we still find that Jeremy Lin was a good prospect. If only he had gone to Duke instead of Harvard, maybe he would’ve been a household name before last week.

On the other hand, maybe nobody could’ve seen him coming. At the very least, it’s unlikely anyone would predict a young player playing so well in their first four starts. Either way, New York scores … on accident.

-Greg

Should D’Antoni Get Credit for Lin’s Success?

Posted on by
19

I ran into a post on Marginal Revolution about Jeremy Lin, titled “Is Jeremy Lin a Fluke?”:

Nate Silver says no.  I say that in Mike D’Antoni’s offensive schemes a lot of point guards reap more than the statistics they would pick up on other teams and from other offenses, and since the D’Antoni scheme is not very generalizable, or capable of winning a championship, the “other team” metrics are more or less the correct ones.

Here at the Wages of Wins we have our own thoughts about Lin (more on that in tomorrow’s post), but for now I am much more interested in Tyler Cowen’s thoughts on Mike D’Antoni above.

First, I might ask, If it’s all about D’Antoni’s schemes, why is Steve Nash still one of the best point guards in the league? D’Antoni has been gone for several years, and to my knowledge his successors have changed the system a lot (and, in any case, isn’t Cowen making the argument that they couldn’t? I’m not sure what “hard to generalize” really means). Why didn’t Nash’s performance drop off with D’Antoni’s departure?

Careers per-48 Stats
SEA Min WP48 Wins PTS DRB ORB REB AST TO BLK STL PF
06-07 PHO 2682 .307 17.1 25.3 4.3 0.5 4.8 15.8 5.1 0.1 1.0 2.1
07-08 PHO 2780 .251 14.6 23.7 4.4 0.5 4.9 15.5 5.1 0.1 0.9 2.0
08-09 PHO 2484 .194 10.0 22.4 3.9 0.4 4.3 13.9 4.8 0.2 1.1 2.1
09-10 PHO 2660 .250 13.8 24.1 4.2 0.6 4.8 16.1 5.3 0.2 0.8 1.9
10-11 PHO 2497 .244 12.7 21.3 4.2 0.8 5.0 16.4 5.1 0.1 0.9 1.7
11-12 PHO 567 .231 2.7 21.8 3.0 0.5 3.6 15.2 5.3 0.0 0.8 0.8
Career Avg 2465 .241 12.4 23.4 3.9 0.8 4.7 13.9 4.6 0.1 1.1 2.3
Average PG 1624 .099 3.3 18.8 3.9 0.9 4.8 8.3 3.2 0.3 1.8 3.6

 

Career Shooting Efficiency
FG% 2FG% 3FG% FT% eFG% TS% FGA 3FGA PPS FTA
Nash 05-06 51.2% 54.8% 43.9% 92.1% 58.3% 63.2% 18.1 5.9 1.41 4.8
Nash 06-07 53.2% 57.5% 45.5% 89.9% 61.3% 65.4% 17.4 6.1 1.45 4.4
Nash 07-08 50.4% 52.7% 47.0% 90.6% 59.7% 64.1% 16.6 6.6 1.43 4.2
Nash 08-09 50.3% 52.9% 43.9% 93.3% 56.6% 61.5% 16.4 4.8 1.36 4.1
Nash 09-10 50.7% 54.0% 42.6% 93.8% 57.0% 61.5% 17.8 5.3 1.35 4.1
Nash 10-11 49.2% 52.5% 39.5% 91.2% 54.2% 60.1% 15.6 3.9 1.36 4.8
Nash 11-12 52.8% 57.7% 39.2% 85.0% 58.0% 61.3% 16.3 4.3 1.34 3.4
Nash (career) 49.7% 52.4% 43.3% 90.8% 56.2% 61.2% 17.1 5.1 1.37 4.7
Average PG 43.0% 45.7% 35.6% 80.5% 47.8% 52.5% 16.0 4.3 1.18 4.4

Although D’Antoni left the Suns after the 07-08 season, there hasn’t been an appreciable drop off in Nash’s performance. In fact, Nash is a bit of medical curiosity, since a player his age should be declining rabidly.

Furthermore, if D’Antoni’s system is so generous to point guards, why have Toney Douglas, Iman Shumpert, and Mike Bibby all been so terrible?

Raw Stats
Min WP48 Wins PTS DRB ORB REB AST TO BLK STL PF
Shumpert 751 .010 0.2 16.4 4.5 0.9 5.4 5.3 4.0 0.4 3.3 5.0
Bibby 272 -.003 -0.0 10.4 4.2 0.5 4.8 4.9 2.1 0.2 1.6 4.1
Douglas 533 -.141 -1.6 17.7 4.0 0.5 4.5 5.2 4.1 0.1 2.1 4.1
Average SG 570 .099 1.2 20.1 4.3 1.0 5.4 4.2 2.8 0.4 1.6 3.3
Average PG 617 .099 1.3 19.6 3.9 0.9 4.9 8.2 3.6 0.4 1.9 3.3

 

Shooting Efficiency
FG% 2FG% 3FG% FT% eFG% TS% FGA 3FGA PPS FTA
Shumpert 38.5% 41.7% 28.8% 84.6% 42.1% 45.4% 16.9 4.2 0.97 2.5
Bibby 28.6% 31.6% 27.5% 83.3% 38.6% 40.6% 12.4 9.0 0.84 1.1
Douglas 32.1% 38.1% 23.8% 85.7% 37.1% 39.3% 21.6 9.1 0.82 1.9
Average SG 42.5% 45.8% 35.7% 80.3% 48.3% 52.6% 17.2 5.6 1.17 4.3
Average PG 42.9% 46.1% 34.7% 80.9% 47.8% 52.4% 16.8 4.7 1.17 4.4

Regarding the second part of Cowen’s analysis, I am curious as to how exactly he determined that “the D’Antoni scheme is not very generalizable, or capable of winning a championship”? What huge sample of championship runs is Tyler using for his observations here? Even if we made the rather generous assumption that every Suns roster he coached was championship-caliber (as an aside, I hope he isn’t going to argue that the Knicks rosters of the past few years were), we’re talking about a sample of a few seasons, correct?

I’m fairly certain if a student presented a paper that came to economic conclusions based on evidence like this in one of Tyler Cowen’s classes, the grade would not be favorable.