Quantitative analysis is great, isn’t it?
I make my living working for [insert name of very well known investment bank here], and I spend my days supplying financial data to institutional investors that guides their investment decisions. Essentially, titans of industry use my data to create projection models for how stocks will perform in the future. I am kind of like Nate Silver, but no one knows me.
I am also a Knicks fan, so naturally I was very interested in understanding why something called a SCHOENE was responsible for blowing up my Twitter feed the other day. Apparently, ESPN annually enlists stat-guru Kevin Pelton, the creator of the aforementioned win-projection model, to predict all 30 NBA teams’ performance in the upcoming season.
Never heard of SCHOENE, you say? No worries, all is explained by ESPN right here. Don’t have time to decipher all that nerd-speak? Well, if you’re a New York loyalist, all you need to know is that SCHOENE thinks that New York will win all of 37 games in 2013-14. And for the record, I am aware that SCHOENE is apparently pronounced shea-knee, but that doesn’t rhyme with baloney, so screw you.
Thirty. Seven. Wins. For the love of John Hollinger, surely, they can’t be serious?! But hang on, slow your roll. Before Knick Nation has itself a Carrie Mathieson style breakdown over a statistical model, allow me to make sense of what Pelton’s HAL-9000 is trying to tell us. There is little doubt that data can be invaluable in projecting performance in sports, but we must exercise caution and realize that data sometimes has limitations and flaws.
So what does it all mean, Basil?
First, we must understand SCHOENE’s data inputs — and put them into proper context — before simply accepting that its output, 37 wins in this case, dooms the Knicks to the 8th seed and ensures Carmelo Anthony’s departure from Gotham. SCHOENE starts by taking the last three years of each NBA player’s individual statistics to create a baseline projection that is weighted/adjusted for age and injury. Naturally, the Knicks are penalized because they are an older team. While this makes sense on the surface, the fact that this year’s roster is actually younger than last season’s 54-win squad is an inconvenient fact that cannot be ignored if we’re being fair.
In the case of Carmelo Anthony, 29, who is arguably close to exiting his prime, there is no reason to think that his style of play won’t age well, especially since SCHOENE is only projecting performance for this season. If Paul Pierce enjoyed one of his finest season at age 33 (.201 WS/48), age alone shouldn’t portend ‘Melo to experience a regression from last year’s league-leading scoring output. And we can assume that SCHOENE’s three-year lookback isn’t factoring in the drama that surrounded Anthony’s uneven 66-game campaign under the deposed Mike D’Antoni, either. Also, where does SCHOENE factor in a player’s increased experience and evolving guile?
SCHOENE also seems to place an inordinate value on a team’s assist-to-FGA ratio. While I agree that the Knicks’ dependence on isolation-heavy sets is disconcerting, elite offensive talents like Anthony and J.R. Smith have proven that they can score despite suboptimal ball movement. Again, no one is suggesting that this is the best or most efficient way to approach your offense, but the Knicks finishing dead last in the league in assists-per-game last year isn’t necessarily predictive of future performance. In fact, I would counter that SCHOENE inexplicably overlooks a closely related statistic: TURNOVERS. The Knicks had the lowest turnover rate in the NBA last season, and I am willing to bet that if SCHOENE emphasized assist-to-turnover ratio over assist-to-FGA ratio, New York’s win-projection would more closely align with last season’s 54 wins — not to mention fans’ expectations for the upcoming year.
Beyond the wins-projection, a holistic look at the results generated by SCHOENE — even if it’s accurate — suggests reason for optimism. Again, it’s all about context. SCHOENE predicts that 37 wins will be good for a seventh place finish in the Eastern Conference and a playoff berth. Sure, Knicks fans would prefer to compete with the likes of Miami, Chicago, Indiana and Brooklyn for a top-4 seed, but things could be worse if SCHOENE’s feel-bad prognostication becomes reality.
It should also be noted that the Knicks are not the only team that SCHOENE isn’t bullish on this season. The Knicks’ minus-17 win total is drastic, to be sure, but the Heat (minus-12 wins) and the Thunder (minus-9 wins) don’t fare so well, either. Heck, Oklahoma is projected to win as many games as the Timberwolves, for crying out loud! And yes, I have heard something about the return of someone named Kevin Love, but come on.
And what are we to make of SCHOENE’s treatment of teams that are expected to tank? Those teams, like the Celtics, Sixers, Bobcats and Magic, that intentionally roll with a substandard lineup are sure to increase the win-totals of their division and/or conference rivals, are they not? SCHOENE’s highest win projection is reserved for the Spurs (60), but the last time, excluding labor shortened seasons, that 60 or fewer wins led the league was during the 2002-03 season. I am pretty sure the Spurs are already 4-0, anyway. (They play the Sun four times.) All of this leads me to believe that SCHOENE, like PECOTA and KUBIAK before it, generally tends to predict a worst case scenario. (We will know more when the projections for every team is published.) In SCHOENE’s case, I think that the model’s reliance upon age, injury and assists are overly emphasized. How else to explain the performance of last season’s ancient, assist-allergic Knicks team. Simply put, relying upon the projections of a computer, especially without context, is pure folly.
So before we lambast the overloads at ESPN for trolling Knicks fans, let’s keep things in perspective and remember that we are talking about a statistical model here. And one with a weird name, at that. This isn’t some vast conspiracy by the “Mother Ship” to rile up the largest media market in North American sports — that’s Stephen A. Smith’s job, People — it’s simply a well-reasoned attempt to predict future performance based on past results. Pelton’s model, finely broken down by Tom Haberstroh, I might add, is down on the 2013-14 New York Knicks, that much is clear.
But the model doesn’t preclude the Knicks from exceeding expectations. It simply means that an almost-sentient robot isn’t blowing smoke up our collective New York asses. And I’m okay with that. You should be, too.