The Relationship Between Experience and Success

Oct 18, 2014; San Antonio, TX, USA; San Antonio Spurs players (from left) Tony Parker, and Manu Ginobili, and Tim Duncan during the national anthem before the game against the Miami Heat at AT&T Center. Mandatory Credit: Soobum Im-USA TODAY Sports

[ED Note: This post is a collaboration between Ben Dowsett (@Ben_Dowsett) and Andrew Koo (@akoo).]

“This is a veteran group they’ve got out on the floor right now, these guys really know how to play together.”

“They’re a young team with a lot of inexperience, how will they fare when the going gets tough?”

“Late in his career, what he lacks in explosiveness he makes up for with veteran savvy.”

Frequent NBA viewers are surely familiar with these sorts of expressions, typically repeated by our favorite television personalities.

Often intended innocently enough, a number of pseudo-clichés have emerged over time. Teams or lineups with predominantly younger players are “raw” or “unseasoned”; those with prominent veterans often possess “savvy” or “poise” in the face of on-court challenges.

Such descriptors are mostly harmless, but are often inaccurate representations of reality. Highlighting one or two roster members nearer to one end of the spectrum can ignore others elsewhere in the rotation who may not conform to the intended characterization. Roles are often overlooked or mostly pushed aside when they don’t fit the brief narrative being crafted, and “experience” is sometimes confused with “success,” particularly among the more well-known veteran names being highlighted.

This isn’t a vendetta against in-game commentators, who often lack the time or sophisticated audience for much more than such vague labels. But it asks the question: Is there any sort of true, demonstrable value to engineering rosters with age or experience in mind? And tangentially, do teams with more age and experience tend to have more success, or vice versa?

These are questions that have been asked and answered to some degree. Current ESPN analyst Kevin Pelton, while with Basketball Prospectus in 2009, explored the effects on team success of their “effective age,” also known as “minutes-weighted age” – a simple formula that factors in minutes played that season to avoid averages being skewed heavily by very old or very young players on the end of a given team’s bench. Pelton’s research found moderate correlations between age and both offensive and defensive efficiency (.382 for offensive, .388 for defensive), as well as a somewhat significant .515 correlation with team win percentage going back to the 1979-80 season. These basic observations would appear to jibe with popular perception. It’s rare to see a very young team at an elite level, and teams with at least some significant veteran presence are more prone to overall success.

We decided to delve a little further. Could a refinement of certain basics in the formula provide us a more intimate view into the effects of age and experience? We decided to test two new variables in an attempt to better define “age” as it tangibly affects players. The first: seasons experience, a simple measure of a player’s total NBA seasons rather than their calendar age. With each additional season, players learn more and train professionally through an offseason. Perhaps this better represents ability.

The second was career minutes played – that is, we weighted for each player’s minutes over their entire NBA career to try and more accurately assess the effects of long-term stress on the body. We called this minutes experience.

These three measures obviously correlate strongly with each other, but each definition holds possible details within – age may signify the presence or decline of athleticism, and seasons experience may indicate actual relevant experience. Minutes experience is an even finer representation of the latter.

Overlapping player careers (which would affect our career minutes-weighted numbers) kept the sample to a period from the 2002-03 season up until the 12-13 year, but this remains a robust sample of 328 teams. As a baseline, we ran the same numbers as Pelton for single-year effective age, and found similar correlations – a .530 correlation to team win percentage, roughly in line with .515 from Pelton’s study (keep in mind the different sample periods).

The data showed slight upticks in correlations when using the more detailed variables. When compared with years experience, the correlation to win percentage improved to .549 (lighter blue teams are more recent):

And when run next to career “minutes experience,” the number takes another jump to a .558 correlation:

These aren’t major revelations, but they’re instructive and on par with intuitive expectation. Players gain their true experience from playing actual minutes, not simply from being in the league for longer periods of time, and the data bears out this small differentiation. A larger gap between single-year and career effective age was always unlikely, if for no other reason than the fact that better players will typically tend to remain in the league longer and play more minutes than their less talented counterparts.

Our research yielded some more interesting tidbits. We pulled a number of other team measures for our data period beyond simple win percentage, and tested them against our three variables of effective age, years experience, and minutes experience. Using Pythagorean win probabilities (expected wins and losses based on points scored and allowed) rather than actual win percentage, correlations were actually slightly lower; still, minutes experience won out:

(Click to embiggen)

Diving into further granularity, possibly the most granular, minutes experience also produced the highest correlation to net rating – though it’s much closer:

(Click to biggify)

The more detailed age variables showed perhaps their largest effect compared to raw age when looking at shooting numbers. Career minutes experience and years experience showed correlations of .384 and .359, respectively, with True Shooting Percentage, while simple effective age clocked in at just .295. Again, this matches up with intuitive thought; guys who see more court time in their cumulative NBA careers, and therefore more shots and repetitions, see the results in their eventual shooting percentages.

(Enlarge by clicking)

Once again, nothing herein is groundbreaking or contrary to inherent logic. Particularly young teams lacking significant veteran presence have historically struggled to reach elite levels, but on the other side of the coin, no magic formula exists for the sort of experience/talent/youth combination that lends itself directly to winning teams. General managers can’t simply add a bunch of late-career veterans and expect instant success; as always, there remain enormous amounts of context to be considered within any and all team-building decisions. In the aggregate, the league average reflects a trend towards more experienced players:

… though some of this is because of recent, older teams skewing the average, namely the Mavericks and Celtics – and it’s worked for them. That’s not to say other teams are adopting the opposite strategy, or are even conscious of accrued experience: the 2013 Rockets were the least experienced team in seven years, and won 45 games with a weighted average of 1.4 seasons’ experience.

Overall, the number of career miles on players, and their team rosters, appears to hold slightly better correlation value at the team level for many relevant areas than their single-season effective age totals, as does “years experience” to a lesser degree. As the league changes – possible modification of the one-and-done rule, or players training overseas – and trailblazers like Gregg Popovich emphasize easing the long-term stress of vital players through minutes reductions, keep an eye on how these patterns advance.

Basketball-Reference.com provided data for this study.