Whew! First of all, I have not posted here in a while - I've been devoting most of my writing energy to RantSports for the past couple of months. However, what follows is not for RantSports, but is a model I created based off of Arturo Galletti's Wins Produced metric, which can be found at boxscoregeeks.com. Wins Produced is awesome - it's incredibly accurate, and relies only on box score stats to create its values. By weighting different statistics differently, Galletti and co. were able to find the right way to value different stats.
As Galletti's pal, Andres Alvarez, points out on their website, while Wins Produced makes predicting players' year-to-year production relatively simple, predicting their minutes played can be rather difficult. Injuries, coaching decisions and various factors strongly affect players' minutes played, making win-loss predictions rather difficult to divine. Therefore, with the creation of my model, creating a formula to predict each player's minutes played took up most of my time and effort. Essentially, I used such factors as the average number of minutes played per season during each of the past three seasons, position on a team's depth chart and whether a player started or not to estimate projected minutes. While rough, the minutes formula gave me what looks like a pretty accurate depiction of projected minutes; however, injuries and coaching decisions still will render many of the predictions completely wrong, I'm sure. Also, the model doesn't account for teams that have different attitudes toward how their minutes are divvied up; a team like the Spurs, for example, play their starters for relatively few minutes per game (around 29 each), while a team like Tom Thibodeau's Bulls will pound its starters into the ground, playing them each about 37 or 38 minutes per game.
What follows is the prediction for each team that the model spit out; the Vegas over/under (via Bovada.lv), which provides a gauge of public perception of the team and how the model differs; and my own gut feeling about how the team will do this season. So, without further ado, let's get down to some divination!
As Galletti's pal, Andres Alvarez, points out on their website, while Wins Produced makes predicting players' year-to-year production relatively simple, predicting their minutes played can be rather difficult. Injuries, coaching decisions and various factors strongly affect players' minutes played, making win-loss predictions rather difficult to divine. Therefore, with the creation of my model, creating a formula to predict each player's minutes played took up most of my time and effort. Essentially, I used such factors as the average number of minutes played per season during each of the past three seasons, position on a team's depth chart and whether a player started or not to estimate projected minutes. While rough, the minutes formula gave me what looks like a pretty accurate depiction of projected minutes; however, injuries and coaching decisions still will render many of the predictions completely wrong, I'm sure. Also, the model doesn't account for teams that have different attitudes toward how their minutes are divvied up; a team like the Spurs, for example, play their starters for relatively few minutes per game (around 29 each), while a team like Tom Thibodeau's Bulls will pound its starters into the ground, playing them each about 37 or 38 minutes per game.
What follows is the prediction for each team that the model spit out; the Vegas over/under (via Bovada.lv), which provides a gauge of public perception of the team and how the model differs; and my own gut feeling about how the team will do this season. So, without further ado, let's get down to some divination!
No. 30: Philadelphia 76ers
Model Prediction: 20-62
The model, like most people, has the tankalicious Sixers as the worst team in the NBA for the second year running. While general manager Sam Hinkie's efforts weren't enough to get the league to change its lottery system, they will be enough to keep the Sixers at the bottom of the standings once again. The model somewhat likes Henry Sims, Hollis Thompson and Nerlens Noel, pegging them all as above five-win players; however, after those three, this is a bleak, bleak roster. Noel's hair should be a lock for an All-Rookie team, at least.
Vegas Over/Under: 16-66
Despite the model having the Sixers as the league's worst team, Vegas has outdone it. All I have to say is: it's really hard to lose 66 games. Last year's Sixers couldn't do it. And this year's Sixers team can't be as bad as last year's...right? Maybe?
Gut Reaction:
I think 20-21 will be right around where this year's Sixers will fall. I think Noel will be an NBA-quality big man right away, and MCW will be a quadruple-double threat every night (a triple-double, but with 10 turnovers to boot), but the loss of Thad Young (who deserves more than anybody else in the league to get traded to a contender) might hurt more than I think. It's hard to peg a win total on a team that can go on a 20-game losing streak at any given time.
No. 29: Boston Celtics
Model Prediction: 24-58
Boston is a difficult team to predict, given Rajon Rondo's tenuous position as trade bait and the role rookies Marcus Smart and James Young will fill. For the record, to estimate rookies' Wins Produced, I used the average Wins Produced for draft picks at that position for first-round picks, and an average of .009 Wins Produced per 48 minutes, the average for second-rounders and undrafted free agents, for the aforementioned two groups. Rondo's eight wins could leave with him if he's traded, so I wouldn't put money on this one.
Vegas Over/Under: 27-55
I think Vegas is more bullish on Brad Stevens' coaching, which the model doesn't account for, and the two rookies' production. Again, given the Rondo situation and unknown production of Smart in particular (who could be an immediate impact player, or shoot 32 percent from the field), this is not a lock to go under by any means.
Gut Reaction:
I think 24 wins is a bit low for this Celtics squad, but by just two or three wins. I would be incredibly surprised if it got to 30; at this stage, it's just not a well-rounded team with multiple weaknesses.
No. 28: Los Angeles Lakers
Model Prediction: 26-56
A lot of people like to see the Lakers succeed because it's good for the league. I don't. As a tortured Warriors fan, I love to see the Lakers suffer. Put it this way: Jordan Hill will be responsible for 10 of its 26 wins. I like Jordan Hill as a player, and the model does too (and D'Antoni would take him out at the half for good after he had 12 and 8 for some reason pretty much every game last season), but this team will be leaning on Jordan Hill. Yep.
Vegas Over/Under: 30.5-51.5
Linsanity Volume Three and/or a Kobe comeback not withstanding, this team is a lock to go under. Byron Scott does not mesh with this roster whatsoever (all D, no threes philosophy - the opposite of the entire team). Call me a cynic not betting on Lin or Kobe, and it may be my desire to watch the Lakers flounder, but I think this team will be sitting the basement of the West all season.
Gut Reaction:
26 seems like an accurate number to me. They could even go under, but I think anywhere between 24 and 28 is this team's destiny.
No. 27: Orlando Magic
Model Prediction: 29-53
Another team that's hard to predict given its two rookie starters, the Magic's three leading players by Wins Produced look to be Nikola Vucevic, Elfrid Payton and Aaron Gordon. Both rookies might be athletic sensations on the court, or could be zeros on offense that drag this team down. Only time will tell!
Vegas Over/Under: 27.5-54.5
Given this team's variance, 27.5 seems like a reasonable over/under. With this team, the difference between 29 and 27.5 wins is too close to make a wager on.
Gut Reaction:
29 seems a little high for this team given how Jacque Vaughn has yet to prove he's a decent coach. The model is high on Gordon simply by virtue of his being the third overall pick, but he may very well end up being a three-year project (even though I think he'll be awesome one day).
No. 26: Indiana Pacers
Model Prediction: 32-50
Poor Indy. It's still got a couple of eight-win studs in David West and George Hill, but the stopgaps at the wing positions following Born Ready's move to Charlotte and PG's gruesome injury aren't going to cut it, according to the model. I found it interesting that the model seemed to really like most rim protectors (Tyson Chandler, Andre Drummond and DeAndre Jordan all were rated really highly), but Roy Hibbert was rated as a three-win player.
Vegas Over/Under: 33-49
The model and Vegas reached similar outcomes for this team. While the team may seem to have a real dearth of talent, I believe Frank Vogel is a great coach and the team will grind their way to at least 30 wins through defensive effort.
Gut Reaction:
32 seems right to me due to the reasons I just mentioned regarding Vogel; however, the rumblings about Indy tanking to get a high pick (1997 San Antonio-style) have already began. If Indy decides to go for broke, trade West and pair a talented rookie next to PG next year, all bets are off.
No. 25: Milwaukee Bucks
Model Prediction: 32-50
Jason Kidd? Jabari? The return of Larry Sanders? Giannis, who grew 236 feet and six inches this offseason? Sign the model up! The model likes these Bucks evidently, in part because Jabari being the second overall pick places him as a seven-win player even as a rookie. The model also likes the addition of Jared Dudley, almost a four-win player - not bad for someone who's probably going to come off the bench.
Vegas Over/Under: 24.5-57.5
Vegas is much lower on the Bucks, obviously because of last season's stinker. This team got better in every aspect this offseason, but Vegas believes that those changers were worth far fewer than the 17 wins the model dictates.
Gut Reaction:
I like the Bucks as a 30-plus win team! I think Kidd is a huge upgrade despite his power trip, I think Jabari is NBA-ready (Paul Pierce 2.0, but with exactly the opposite temperament) and can put up 18 points a game, I think Sanders is at the very least a top-12 rim protector, and Ersan Ilyasova, Khris Middleton, John Henson, Giannis and Dudley are all competent role players. While they might not be ready for the playoffs this season, it's quite clear that a 7-foot-4 Giannis at the point is going to lead them their in 2015. Go Bucks!