by Janet Song
We’ve all heard about how Jeremy Lin was overlooked in high school, undrafted out of Harvard, and cut by both the Golden State Warriors and the Houston Rockets — all of which is seriously linsane. Many claim that someone with Lin’s skill set was passed over only because of external factors such as racial stereotyping and the soft expectations that come with playing in the Ivy League. After Lin’s Knicks defeated the Los Angeles Lakers 95-82 last Friday, Kobe Bryant concurred. “Players don’t usually come out of nowhere,” he said. “If you can go back and take a look, his skill level was probably there from the beginning. But no one ever noticed.”
To examine why Lin was undrafted, let’s compare Lin’s stats from his last year of college ball to those of the 2010 NBA draft class. With the advantage of hindsight, we know that Lin – at least for the last two weeks – has panned out better than many of his drafted peers. In the weeks leading up to the 2010 draft, however, was it reasonable to expect Lin to be drafted? How do his college stats line up?
I compared Player Efficiency Rating (PER) for all drafted players who played in the NCAA during the 2009-2010 season and adjusted conservatively for strength of schedule (SOS). My adjusted PER (aPER) took the z-score of the strength of schedule for each drafted player as compared to the total set of drafted players and multiplied the z-score to the standard deviation of the PER for all drafted players. This value was then added to the unadjusted PER to get a conservative aPER.
How did Jeremy Lin stack up?
He had an Ivy League-leading unadjusted PER of 26.8 in 2009-2010, which would very favorably match up with the PER of drafted players. Harvard’s extremely weak SOS of -3.9, however, dropped his aPER to 5.5. This is far outside the aPER for drafted players. Additionally, this would place all Ivy League players well below the league average PER of 15. This could indicate one of two things: (1) all Ivy League players really are that horrible or (2) it is near impossible to compare players who play under widely divergent strength of schedules. Due to the success of Cornell and Princeton in recent March Madness tournaments, it is much more likely that our current methods for comparing players based on statistics adjusted for SOS are simply inadequate to compare players in leagues like the Ivy League to powerhouse conferences such as the ACC. There is no true precedent on how to project player outcomes in widely disparate conferences. As a result, many draft analysts are likely to default to picking players that play against “adequate” competition and overlook players like Lin who hail from small conferences but still deserve a shot at the NBA.
It could be argued that looking at unadjusted PER could give a better picture of the true NCAA landscape. However, a quick look at all of the players with PERs of at least 25 who played in at least 20 games for 600 cumulative minutes numbers in the 70s in 2009-2010. (A PER of 25 is very high. As a comparison, John Wall, the number one pick in the 2010 draft had a PER of around 22.) Looking at the results, highly ranked players include Omar Samhan (PER of 33.8) who is now playing in the Phillipines and Jordan Eglseder (PER = 32.2) who averaged 2.0 points per game last year in the D-league. Perhaps, like Lin, these players are all waiting to be discovered, but the more likely explanation is that they simply are not as talented as the players who made it to the NBA and who generally, played against stronger competition in the NCAA.
Why Lin was not recruited out of high school to play in one of these power conferences is another topic altogether. Nevertheless, if NBA general managers do not want to miss out on the next Jeremy Lin, they would do well to consider smaller conferences and delve beyond strength of schedule to more seriously consider the players in the league.
 A conservative adjustment would least deflate Lin’s PER (given that he played on a team with an extremely low SOS). Since the analysis showed that Lin’s aPER could not match up with those of drafted players even with this conservative adjustment, less conservative adjustments were not needed for analysis.
 This is not strictly accurate as I adjusted only across drafted players and did not re-normalize to a PER of 15. Nevertheless, the true average would be much closer to 15 than it would be to 5.5 (Lin’s aPER).