WE'RE ALL BAD AT PREDICTING

What if I came to you and said "I'm an expert predictor! Give me a 24 win range and I can tell you how well every team in the NBA will do?" Let me give you a practical example. I'm a Nuggets fan. Using the expert prediction skill I mentioned above, I will now make the following prediction:

Next season the Nuggets will win between 40 and 64 games!

Useful right? For the last several seasons, wilQ at Weakside Awareness has done a checkin on how analysts do in terms of predictions at the start of the season and how the season plays out. And the way this is typically interpretted is "Wow John Hollinger is great at predictions! He beat Vegas!" But there's a key flaw here. You see, the story isn't Hollinger is good at predictions, it's that everyone is bad!

The metric wilQ uses is standard error. Now, I trust that wilQ knows the math, he's a very sharp analyst. However, I don't buy that everyone else does. A better way to think of standard error is "What's the range of games I need to place all of my predictions?" The way it works is:

I need twice my standard error to properly predict 20 teams. I need four times my standard error to properly predict 29 teams. With that in mind. Let's redo wilQ's chart of how our analysts did.

Analyst Range of games to get 29 teams "right"
John Hollinger 24
Bovada updated 24
Zach Harper 25
ESPN forecast 26
O/U Line 26
Joe Schaller 26
Cantor Gamings 27
predictionmachine 27
Joel Brigham 27
Las Vegas Hilton 27
George J Monroy 28
Bovada early 28
Matt Moore 28
Sportsbook early 29
Robert Eckstut 29
Bobbo 29
David Williams 29
theNBAmodeller 29
Bradford Doolittle 30
DanielM ASPM 30
version “dumb” 30
Darryl Howerton 30
Ed Weiland 30
WoW1 30
AccuScore 31
WoW2 31
TeamRankings Projections (BETA) 32
Kevin Pelton 33
David Hess 33
SCHOENE 33
Royce Young 34
Derek Ayala 37
ESPN The Magazine and Basketball Prospectus 38
Wins in previous season 39
 
Hollinger needs 24 games to forcast teams (the difference between Boston and Miami) This was better than Vegas, which is apparently the gold standard, by three games. This was a full six games better than the best Wages of Wins model, which is what people are focused on.
 
However, as I mentioned, the key is not that any of these models is best. Rather, it's that they all fail! We're relative creatures, so we want to say - look Bovada did better than Schoene! But in this case, that doesn't matter. The range of games you need is so large, the information is meaningless!
 
A book that everyone purports to have read is "The Signal and the Noise" by Nate Silver. A topic Nate brings up is "Unknowable Unknowables" As the name pretty much says, these are things that you don't know and more importantly, can't know. For instance, Steve Nash got injured this season. No one could have known about this ahead of time and yet, it clearly impacted things. George Karl's decision to not play JaVale McGee, despite the fact he got a major contract, was not knowable ahead of time to anyone.
 
When making predictions, we are making assumptions about many things. One of the things we know pretty well is player evaluation. If a player was good last year, they will probably be good this year. However, even this is unexact. Players perform differently as they age. And of course, rookies are notoriously hard to predict. We have a limited number of things we have a firm grasp on. So it's not surprising that we are bad at predictions.

Updating vs. Predicting

What's key about this is to stop using predictions as the end all be all test. Hollinger is another great example. In the middle of the season Hollinger went to the Memphis front office. And the Grizzlies, who are now in the conference finals, made several moves to their teams. Of course, Hollinger could not have known about these to start the season when he made his predictions. And yet, these moves had some influence on the Grizzlies' success. Using the best data at the time and updating our "predictions" as things progress is the right method. And that's why using metrics that have the best combination of

  • Telling us what happened.
  • Being consistent over time.

Are useful. We can use them to guage what is happening and use them to update our analysis about teams. Trying to become "perfect" at predicting just won't happen. Hollinger has "beaten" Vegas three years now. He hasn't improved the range of games he needs to do it though. And, as I mentioned, that range is rather large. The number of moving parts in a season is vast. It's good to give judgements about the state we think teams are in. Being able to accurately guage how they'll do after 82 games when injuries, trades, breakouts, breakdowns, etc. occur though? It's simply not going to happen.

-Dre

 

Categories: Denver Nuggets, Predictions, John Hollinger

O.J. MAYO: REALLY MOST IMPROVED?

The other day, I called O.J. Mayo the "Most Improved of the 1/4-Way of the Season" Player (an award that I am sure he's going to put on his trophy shelf right next to his draft day picture). However, as a I pointed out, I highly doubt that Mayo will actually be the deserving winner of this award at year's end.

And that's because O.J. isn't really playing better basketball.

Dude...wait..wat?

Ok, sort of. Bear with me, I'll explain. Let's compare OJ's last season with this one:

SEA POS Min WP48 PoP48 WP PTS DRB ORB REB AST TO BLK STL PF
Mayo 12-13 (DAL) SG 784 .175 2.4 2.9 28.2 4.7 0.9 5.5 4.7 3.9 0.3 1.2 3.1
Mayo 11-12 (MEM) SG 1771 .072 -0.8 2.7 22.6 4.9 0.8 5.7 4.6 3.4 0.6 1.9 3.7
Average SG SG 1609 .099 0.0 3.3 21.6 4.4 1.2 5.7 4.6 2.7 0.5 1.6 3.5

Ok, so right now you're thinking my math skills suck, because .175 is a much bigger number than .072, right? Bear with me some more. Notice how most of these numbers are pretty much the same? Sure, he got better at fouls, but he's worse in steals and turnovers. So what explains his huge increase in productivity? Well, I should add O.J. Mayo to the FAQ, because his current season is really all the proof you need that Wins Produced does not "undervalue scoring" or "overvalue rebounds'. That's because basically all of Mayo's increased productivity comes from him transforming into a sharpshooting scoring machine:

  FG% 2FG% 3FG% FT% eFG% TS% FGA 3FGA PPS FTA
Mayo 12-13 (DAL) 48.9% 46.9% 52.5% 83.3% 58.6% 62.2% 20.4 7.5 1.38 5.1
Mayo 11-12 (MEM) 40.8% 43.4% 36.4% 77.3% 47.6% 51.3% 19.9 7.5 1.13 4.7
Average SG 43.4% 46.4% 36.6% 80.4% 49.0% 53.6% 18.0 5.4 1.20 5.0

52.5% from 3!? That's some straight shooting.

And here's where it gets interesting. This article started with an email inquiry from Professor David Berri:

Do you know what Mayo's WP48 would be if he shot from 3-point range this year as well as he did last year (and nothing else about this year changed)?

"Why no", I thought, "I don't know. But I bet my fancy database could tell me!"

Luckily, as the tables above illustrate, and as Dave's question alludes, there is already a ton of evidence that O.J.'s game is pretty much the same as it was last year, right down to the number of threes he is attempting. At the time of this writing, Mayo had attempted 115 threes, hitting 61 of them, and scored 460 points. Based on the per-48-minute averages, he would have shot 113.8 last year (the table above is rounding his 3FGA/48 numbers, in case you are wondering at the difference). If Mayo were shooting at career 36.6% (that's a more realiable sample, but it just so happens to be very close to last year's 36.4%), he'd have hit 42 of them (I'm rounding again). So, if that were the case, we would subract 57 points (19*3), and add 19 missed field goals to OJ's totals, and recalculate his production:

SEA POS Min WP48 PoP48 WP PTS DRB ORB REB AST TO BLK STL PF
Mayo 12-13 (FAKE) SG 784 .069 -0.8 2.6 24.6 4.7 0.9 5.5 4.7 3.9 0.3 1.2 3.1
Mayo 11-12 (MEM) SG 1771 .072 -0.8 2.7 22.6 4.9 0.8 5.7 4.6 3.4 0.6 1.9 3.7
Average SG SG 1609 .099 0.0 3.3 21.6 4.4 1.2 5.7 4.6 2.7 0.5 1.6 3.5

Kind of eerie how the result is so similar to his 11-12 season, isn't it? In other words, just about the only difference between last year and this year for O.J. Mayo is:

  • he's hitting more of his threes and hitting more of his free throws
  • his steals and blocks are down slightly
  • his fouls are down slightly, which mostly makes up for the steals and blocks being down

And truthfully, at the end of the day, if the only difference is that the shots are going in, especially when most of those shots are the same types of shots he has always been taking, is that really what we mean by "most improved"?

Now, I'm certain Mark Cuban will make an argument that this is largely coaching. That in Rick Carlisle's system, the looks that Mayo gets are better than he got in Memphis. But I have a very hard time believing this, for several reasons. The first is that Dallas doesn't have very much post scoring to draw defenses it; their best big is Wright, and he doesn't shoot a lot (one suspects that his unreasonably short leash might be the reason for that). The second is that defenses read scouting reports. They know that Dallas' big perimeter threat is Mayo, especially with Nowitski out. And the third is that...well, there's little evidence of it. If Carlisle's system is good at this kind of thing, why didn't have any super 3 point shooters last year? Jason Terry, Vince Carter, and Nowitzki were all good but hardly outerwordly. In other words, if it's the coaching, why is Mayo the only beneficiary? Does anyone honestly believe that suddenly O.J. Mayo is the best three point shooter of all time? It's much easier to belive that he is just running pretty hot. Occam's Razor is my motto.

If one is willing to concede that there is some element of coaching and general young-guys-getting-better involved here (and for the record, I am), it's more realistic to assume a percentage of around 40%, which means that Mayo has improved from "meh" to "average" play.  Which is OK, for $4 million annually. But one suspects that winning the "most improved player" award will probably lead people to believe that he is worth quite a bit more than that.

After all, someone paid DeMar DeRozan $10 million annually, right?

Categories: O.J. Mayo, Dallas Mavericks, Predictions, Most Improved 2012

SURPRISE AND EXPECTATION

I spent the weeks leading up to the season calling out a few things that I thought would surprise people. I promised myself to wait 10 games to do any player evaluation at all (even at 10 games, sample sizes are small, but at five they're pretty crazy), but of course I cannot quite resist.

Surprising: The New York Knicks

Well...sort of. We've been predicting the Knicks would be very good, and of course in the East, "very good" means a lot (as opposed to the West, where it might get you a 6 seed). But no one forsaw this start. Raise your hand if you thought J.R. Smith would be shooting 64% from three while rebounding at the best rate of his career? How about Jason Kidd having an 84% True Shooting? Ironically, the two players getting the most credit, Carmelo Anthony and Raymond Felton, are pretty much playing the same kind of basketball they always have, and at about the same level*. This is a team that I expect will come down to earth quite a bit (to the predicted "good/very good" level), because so much of their success is from insane shooting that isn't sustainable. What would worry me a lot if I were a Knicks fan is that Melo really isn't as great as we thought he would be (i.e. he hasn't benefitted from the switch to more PF as much as the end of last season made us believe he would), and this will get worse when Amare returns.

*As a side note, I hear a lot of people talking about how "it's the system" that makes the point guard so good, pointing at Raymond Felton (again), because he's always so good there. It amazes me how far off the path of logic people will go to justify their views. The last time Felton was in New York, he played in an entirely different system under a different coach. The irony is that "it's the system" might actually be getting some evidence, since Felton hasn't actually been that good (he's below average in rebounding, shooting, and turnovers), but no one seems to be noticing that. Anyway, If Felton were actually playing well, I'd be more inclined to believe that "it's the fitness level".

Expected (by me, anyway. You?): Andrei Kirilenko

Ok, I've been on this topic too much, I will shut up soon. His turnovers are a bit high but that's probably because the Timberwolves are the walking dead right now, and he's being asked to do a ton on the ball. I expect him to also come down to earth a bit (67% true shooting is likely not sustainable), but that would make him "only" an all-star. I honestly have no idea why he had so many doubters. Sadly, he probably won't make it the all-star game. Voters use strictly the "yay points!" player evaluation system.

The Timberwolves, though, are in real trouble until they get Pek back, and even then in mild trouble without Love. I said the Wolves were still a good team without Love, but they aren't a good team without Love, Pek, and Budinger. After their great offseason, the management has made the John Bryant error twice now (once with Louis Amundson, again with Josh Howard). Neither of those guys should be employed by NBA franchises. All of these injuries would be a perfect excuse to give some young guys a shot, and instead they're giving the time to washed-up veterans who won't win any games. And I thought they hired Howard to play a few minutes and give Kirilenko a spell? What business does the guy have putting up 10 shots in 14 minutes? How is that a thing that doesn't get you benched and yelled at, seriously!?

Surprising: The Nets

And by "the Nets" I actually mean Brook Lopez. He's still not that great at rebounding, but has improved to only "slightly below average", but he's been blocking an enormous amount of shots (he can do that!? I didn't know he could do that. Why didn't he always do that!?) and he's been shooting really well. That last, I expect, will regress to the mean, but even if the Nets get average-level production from Lopez, that is a lot more than I expected out of him.

Further surprising to me is Deron Williams. I really thought he would bounce back this year. But his rebounds, assists, steals and turnovers are all at career-worst levels (barring his rookie season assist rate). Here's hoping it's the sample size -- he's on my fantasy team.

Surprising: Kobe and the Lakers

The Lakers have been surprising in two ways: first, Dwight Howard is a shadow of himself. The legs, the back, they are not there yet. Even so, that shadow is pretty effective. so when he is healthy, that'll be scary. A lot scarier than he was in Orlando surrounded by second-string talent. Nash's injury and Dwight's health are the easy explanation for the Laker's surprising slow start.

The other surprise is that Kobe Bryant has actually been playing like the MVP he and his fans think he is. I find it the heart of irony that last year, when he was generally awful, scoring 30 points on 30 shots most nights (while Bynum and Pau were begging for shots in the post), everyone was chanting "MVP", but this year, when he is destroying defenses with 64% true shooting while maintaining his usural rate of 5-6 dimes and rebounds per night, all everyone wants to talk about is how the Lakers are underperforming (when the reasons for it are obvious and not really anyone's fault). Kobe is pulling a Barry Bonds right now and no one is noticing; he's the only reason this banged-up Laker squad with a horrible bench has won 4 games, and that is not something I would have predicted.

Expected: Kevin Durant

I hear lots of "Durant's scoring is down, they really miss Harden." This only tells a part of the story. Durant's scoring is down, but so are his attempts -- he's averaging about 20 shots per 48, down from 24 or so last year. It's possible that the presence of Harden got Durant 4 more looks per 48 but I doubt that the impact was quite that big (especially since they obviously weren't on the floor together 100% of the time). Russell Westbrook is, as always, taking way too many shots*, but he did that last year too. I'm not really sure what the explanation for this is.

BUT.

Lost in the shuffle seems to be the fact that Durant is having a stellar year anyway. His numbers are better in rebounding, assists, fouls, steals, blocks, and even shooting (a remarkable feat, given how good he already was). He's been simply amazing and all anyone can say is "his points are down":

SEA POS Min WP48 PoP48 WP PTS DRB ORB REB AST TO BLK STL PF
12-13 OKC SF 387 .302 6.3 2.4 30.4 11.5 1.1 12.7 5.1 5.0 1.9 2.0 1.6
11-12 OKC SF 2546 .226 3.9 12.0 34.9 9.2 0.8 9.9 4.4 4.7 1.5 1.7 2.5
Average SF SF 1622 .099 0.0 3.3 20.5 5.7 1.9 7.6 3.5 2.6 0.8 1.6

3.8

*Among players with > 50 minutes, Westbrook leads the team in fga/48 by a wide margin, and is last in true shooting. Averaging 11.4 assists/48 does not make up for this.

Surprising: The Rookie class

So far the rookies of this year have impressed me a lot. Small samples again, I know, but for a draft that was supposedly very weak after the #1 pick, there have been quite a few surprisingly solid rookies. This might turn out to be a memorable draft. I'll have more on this in a few games when the sample size is a little bigger.

Surprising: The 76ers

I expected this team to be bad, even with Bynum. Without him, they've gone 5-4. Jason Richardson has been playing like the other guy who wore #23, Jrue and Ivey have been a solid PG rotation, and Thadeus Young is making up for the loss of Brand. Of course, Nick Young is every bit as bad as I predicted he'd be (worse, really). Unfortunately, it's likely Richardson will come down to earth, and this team is going to be in trouble, especially if they keep giving so many minutes to Young. This team reminds me a lot of Toronto -- the players that management thinks are the stars are really its worst players.

Surprising: The Bobcats

For the record, I still don't think this is a good team. But "not good" is obviously a huge upgrade over a .129 winning percentage. And Kidd-Gilchrist is actually giving me a reason to watch this team; he's basically the exact opposite of every other player Michael Jordan has acquired. Love the way he rebounds and hustles on defense. And win did Kemba Walker turn into such a defensive menace? He's got 20 steals so far. He had 60 all of last year.

Expected: LeBron

Playing power forward, LeBron is putting up career-bests in rebounding, blocks and turnovers, while shooting his usual 60% true shooting and racking up a ridiculous 9 assists/48. This is so unsurprising to everbody that no one even mentions him in the MVP debate.

But, hey, pundits, playing the best player in the game, who's 6'9 or so and about 250, at power forward, is not "smallball". This is because LeBron James is not small. He is not a small guy who is strong enough to play power forward. He's a huge, strong guy who is quick enough to play small forward. The difference is significant.

 

Categories: Predictions, 2012 season

THE TIMBERWOLVES: I'M JUST SAYIN'.

The Timberwolves, who are "without their two best players," are now 4-1. Indiana is the only playoff team they have faced, so there's that. And tonight they play the Bulls. Still, a lof of folks are...surprised. Like this guy at Yahoo (h/t CanisHoopus):

Though many expected the Timberwolves to stumble as Love worked his way back from a broken right hand, coach Rick Adelman's team has proven to be one of this season's biggest surprises in getting off to a 4-1 start.

Here's hoping their play is as sharp as their dress when they return!

The bolding is mine. Here's me about 10 days ago:

I said this before, but then Love got hurt. My quick surprise prediction: the Wolves are not going to suck without Love. This whole roster is revamped, these are not the same losers that went 2-9 without Love. I guess I touched on that when I covered the Suns, who employ the actual players who are going to be terrible without Love (like they were last year). I think this is a solid .500 team without Love, maybe better given the schedule that seems to benefit them.

I mean, I'm not really saying anything here...but I'm just sayin'. Why doesn't anyone listen to me!?

And while I am on this topic, this nonsense about the Wolves being "without their two best players" just has to stop. I like Ricky Rubio (who doesn't like Unicorns? Only Black Mambas), but this team's two best players are, without a doubt, Kevin Love and Andrei Kirilenko. Rubio, who was a pleasant surprise last year, doesn't come close. Rubio will certainly be an improvement over Ridnour when he returns (although Ridnour has been shooting on fire so far), and if Rubio has actually improved his shooting during this rehab, then he'll certainly be a great player. But Kirilenko is tearing up the league right now. It's not the thing a lot of people notice because it isn't all done with "Yay! Points!", but he's been dominant. He's playing 2 positions, defending 3 positions, blocking shots, getting steals, passing, has nearly 70% true shooting, he's just all over the place. At this pace he deserves to be Defensive Player of the Year -- he's got 34 rebounds, 11 blocks, 8 steals and 2 (TWO!) fouls through 5 games. Those are crazy numbers.

On a side note, a casino here in Vegas gave me 50:1 for the Wolves to win the Western Conference. That's nuts. Before the season, the odds were 40:1. Then they opened 3-1 without Kevin Love and the odds got better for me, going to 50:1? What, now?

Do I think the Wolves will win the West? Hell no. They'll have too much trouble in coming games against playoff teams (like tonight's tilt against the Bulls) without Love to keep this up, so they'll likely be too far behind the Spurs and Thunder to catch up, once he's back (and that's assuming he jumps back in at full strength). But I'd certainly give the Wolves a 5-10% chance at it (let's say Chase Budinger has a breakout year, plays like a star, and gets the bulk of minutes at SG), which means if I am getting 50:1, that's between 250% and 500% on my money! I wouldn't mortgage the house on a bet like this (like I would have taking the over on 38 wins before the season started), but that's what we call a "high-variance +EV bet" in the poker world -- definitely worth a couple of hundred bucks.

Categories: Andrei Kirilenko, Kevin Love, Ricky Rubio, Minnesota Timberwolves, Predictions

TIMBERWOLVES MINI-PREVIEW

I got a call from a guy at Fox in Minnesota, asking me if he could feature my blog in a commercial because he likes my work. I suspect, though, what he really likes is the fact that I am pretty bullish on the Minnesota Timberwolves:

Note how my article on the Wolves being good enough to get 57 wins is featured, with the 57 wins prominently highlighted! I'm used to people coming along and saying "Dude, you high?", and often enough it's actually my sanity barometer -- when mainstream pundits start agreeing with me, I get worried.  If you are chiming in for the first time are going to be thinking I'm nuts. I'll point out, though, that Kevin Love still had healthy knuckles when I wrote that post.

Tonight (finally!) the Wolves play their first game. I've been looking forward to this for a while (how cruel of the NBA to go three whole gamedays without us). Tonight, I only have one prediction, and it is based on sound scientific principles and has been thoroughly tested by most robust statistical models:

  • "And the Peksecutioner will rise from atop a pile stacked high with the skulls of his victims. But the children of the kingdom shall be cast out into outer darkness: there shall be weeping and gnashing of teeth."

Get it? "Children of the kingdom"? Get it? It's funny because we're playing the kings. Get it?

Ok, I have one other prediction for this season. Like Ryan Anderson before him, Nikola Pekovic will win the NBA's "Most Improved" award, without actually improving. They should just rename it the "NBA Player Who's Always Been Pretty Great But Whom We Didn't Know About Last Year" award.

Categories: Minnesota Timberwolves, Predictions