2013 NCAA Tournament Picks

As some of you may be well aware, I have been working for over 2.5 years on creating a model that will accurately predict the tournament. We have built 8 models, 3 primary models, and 5 secondary models. We have the models broken down by 1st round, 2nd round, and 3rd round through the championship game. This allows us to break down different types of games that may have different influential factors acting on them. For example, the second round comes with only 1 day of rest, while the 3rd round comes with around 5 days of rest. In other words, everybody realizes that in the 1st round, there are different factors that could determine the outcome as opposed to the factors determining a national semi-final game.

We had over 1,000,000 different data points for the last 5 season of the tournament. We are predicting future success based off past performance.

Below are the picks, in order by Upper right of the bracket (Mid West), down to the West, and then back up to the South, and ending with the East. When looking at these tables, just look at the matchup, and where they would feasibly play in the tournament.

The right most column has what we (the professor working with me, as well as another student) decided upon after looking at our model, analyzing any current injuries/suspensions, and a strong schedule analysis (not strength of schedule metrics, but is a team trending up or down at the end of the season, big wins, big losses, etc). The human analysis consisted purely of variables that were too hard to place a numerical value on them, thus we didn't have a chance to try and factor it into our model. Along with the team here, Adam Stanco from ESPN contributed all year long with non-statistical analysis (ie: the eye test for every team and player imaginable....he's a college basketball genius).

If the number is green, the favorite will win. If the number is red, the underdog will win. If the number is yellow, it is in our region of error, and is considered a toss up game. If it is yellow in our primary model, we place that game into the secondary model(s). If they are all yellows in the secondary model(s), we likely put a "split" on the game, which means we will pick both teams in various brackets because our model says it is too close to call that game. There were a few games where we overruled the model and you will see when the team is going to lose in our model, is picked in the right most column. There are sometimes a team listed twice [or more] in the same round, and that is because they could be involved in a "split" in a previous round; we ran each possible simulation in the future.

If you don't see a game on here, but you want me to simulate it, tweet it over and I can simulate it! Or shoot me an email if Twitter isn't your style.

Follow @jacobsudek

1st Round (click to enlarge)

2nd Round (click to enlarge)

Remaining Rounds (click to enlarge)

Last night we worked in an awesome conference room at Chapman University where all four walls had "white board" paint. Here are some images from the 8 hour long day where we broke down each potential matchup for all 63 tournament games!

 

How Easy It Is To Stop LeBron James?

Was I the only one that thought the Oklahoma City Thunder looked like a JV team out there in the 2012 NBA Finals? 

LeBron James and the Miami Heat flourished in the 2012 NBA Finals. They absolutely put on a clinic. [Fine, you got it out of a Laker fan]. But how did they flourish? Was it really because they matured, learned from the past, and had a stronger drive to win it all? Or was it more that Oklahoma City did not play as good, as say Dallas did, a year before? Lets look at some numbers and make a case why it is the latter one.

As I watched each game of the finals this year, I kept telling the people I was watching it with, it's simple. Look how many shots LeBron is shooting near the basket. Now, I didn't have any numbers at the time, but my eyes were selling me on what terrible defense the Thunder were playing. Not just terrible defense, but they were giving LeBron everything he wanted; dunks and easy shots near the rim.

Percentages of where LeBron James attempted his shots.

Right away you can differentiate each year and see why LeBron James put on a clinic out there in 2012. He was attempting closer shots to the basket because he was put in better positions to do so. Want something more convincing? In the restricted area alone (that ring right under the hoop), LeBron shot 46% of his shots there versus the Thunder, where as in the whole playoffs, he only shot 37% in the restricted area. Talk about making it easy for LeBron. If you give the man wide open dunks, he is going to slam it with authority every time. It's not rocket science, Scott Brooks.

Mark Cuban and his team understood that and they capitalized. In his appearance on First Take, he debated with Skip and explained how LeBron played terribly in the Finals of 2011.

Watch for about 45 seconds, starting at 3:00.

As a matter of fact, to back Mark Cuban up, he forced LeBron to attempt nearly 3 less shots a game in the Finals than he was averaging in the 2011 playoffs. Regardless of who else shot the ball, when you can force LeBron [or any superstar] out on the perimeter more often and limit his shooting attempts, you have a way higher chance of winning the game.

For Oklahoma City, it really was more simple than than you would think. There are a plethora of ways to force LeBron outside of the key or for him to pass up the ball more often. "Defense wins championships." For a clinic on how to force LeBron outside of the key, watch a few games of the 2011 Finals.


Tweet

Evaluating Player Salaries - Part 2

Tweet
It's been a crazy off season so far and it is not over yet. With today's news on Jeremy Lin going to the Rockets, I needed to dissect it. 


I take a look at a small handful of the free agent signings that have happened already and see if they were worth it. Part 1 looked at win shares rather than PER. I have decided to use PER (which is Player Efficiency Rating defined by John Hollinger) because some of the players in my previous sample played less games, therefore altering the trendline for win shares slightly. Check out Part 1 for a background in case some of this flies over your head.

The Graph: Below are all of the players from the 2011-2012 season. I compared their PER (only players with a 15.0 or higher, just looking at that market) to their salary. I looked ahead to next season by taking a handful of new free agent signings and plotting them (all of them are labeled).

If the player is above the trend line, they are overpaid; if they are under the trend line, they are under paid. If they are within the band lines, their contract is spot on. If they are near, but outside the band lines, their contract isn't a huge problem (probably adjusting for other factors that go unmeasured by just PER).

What we learned:

1. Daryl Morey is an expert (bust most of us know that already). He pushed the Knicks to the edge with a pricey offer sheet for Jeremy Lin. However, Morey put him right on the higher band line and is not over paying for him. If anything, the extra marketing he brings in, he is probably underpaying him, if not paying him just on par. The Knicks failed on that one.
2. Deron Williams new contract is absurd. Would it have been better for the Nets to pass on him? Who knows. At this point, it may as well be championship or bust during the life of his contract.
3. The outcry over Brook Lopez's new contract needs to be stopped. He is barely outside the band line. Regardless if he only played 5 games or not, he is in the right place. Also, don't forget the market for centers is very slim and the Nets didn't want to miss out on a top center.
4. Steve Nash was an absolute steal for the LA Lakers. You mean to tell me they gave up no physical talent to get one of the top point guards in the league, AND not overpay for him? I'd say you're crazy. But hey, Mitch Kupchak has already locked up best GM in the history of the NBA.
5. Ersan Ilyasova got a solid deal by the Bucks. He is underrated by the public but the Bucks saw his value and locked him up with a solid contract for both sides.

Disclaimer: Keep in mind, I am no legal or CBA expert. Certain players may have a floor or ceiling on their contract and I may be unaware of that. Also, this is a small sample size, both in players and years. I also excluded every player that had a PER under 15 because I am looking at the 15+ market only. It won't tell us everything like, intangibles, risk, rookie contracts, city, injuries, etc. But it can tell us, on average, what a player is worth based on their win shares. Lastly, the salary I used for next year is a straight line projection (meaning if the deal was 5 years/$60 million, I used $12 million for the year). I understand each year is structured different, but for the life of the deal, it is better to compare the PER by a straight line projection.

Check out the Clutch Debate between LeBron and Kobe. This blog post has the right numbers from their entire career.

Tweet

Evaluating Player Salaries - Part 1

More and more these days, paying the right amount for a player is crucial with new salary cap penalties. Trying to forecast future production is hard enough, but finding the right value is even harder.

Taking a sample size of players who had a Player Efficiency Rating (PER) of over 15 (which is defined by John Hollinger to be "a pretty good player") and looking at their salary, we can see what the market is like. On the graph, we use Win Shares to gauge how valuable a player was to their team. In short, it is a metric that tells you how many games a player won for their team (defined here). If you add up all of the win shares per player on a team's roster, it will be very close to their actual win total.


Disclaimer is at the bottom.

[click to enlarge]

What we learned form the graph:

• Kobe Bryant is the most over paid player in the NBA; but we already knew that. It just is nice to see by how much he is over paid.
• James Harden, Blake Griffin, and Kevin Love are the top 3 most under paid players sorted by descending Win Shares.
• Blake Griffin will reportedly sign a 5-year, $95 million dollar contract extension this summer. This will drastically overpay him. Since year-by-year salary isn't available yet to the public, averaging each year will give us $19 million a year. He is worth, based on his win shares (using the band lines), around $10.5 to $13.5 million a year. However, the Clippers must feel Blake can offer more than just wins (ie. marketing, chemistry, etc).
• Kevin Love recently signed a contract extension worth about $60 million for 4 years. Averaging his deal equally gives us a $15 million average salary per year. Based on the graph, he is worth around $10.9 to $15.0 million a year. Right on Minnesota! Assuming he can produce near what he did last year, he is worth what he is getting paid for.
• James Harden is a free agent this summer and Kevin Durant doesn't even want him to hit the open market. He is worth around $10.4 to $14.0 million a year. However, OKC knows how valuable he is to their team and rotation, so I would not be surprised if he gets towards the upper end of that price, or even more.
• Note: Kevin Durant is paid $15.5 million a year (right on the trend line).
• Deron Williams just signed a $100 million contract for 5 years. For the Nets to get good value out of that deal, he will need to average 16.4 win shares per [82 game] year. His career best was in 07-08 when he had 11.3 win shares. You do the math here.
• Jeremy Lin played in only 35 games last season. Extrapolating that number out to a full season gives us 5.7 win shares for last year. That puts him in the salary range of around $7.3 to $9.1 million a year. He should be seeking this number right now. Currently the Rockets offered him a contract with year 1 valued at around $5 million. Whether the Knicks or Rockets sign him, they are getting a good deal, if he continues the way he was playing (or even close to that).
• Ryan Anderson is considered to be one of the top free agents this summer. He is vastly under paid per his current production. He should be seeking a yearly salary around $10.1 and $13.2 million a year.
• LeBron James, Chris Paul, and Kevin Durant are actually paid on par for their performance. However, they are superstars and status quo tells us that they should earn more.
• Looking at just Dwight Howard and Andrew Bynum, it is interesting to see how much of a better value Andrew Bynum is. More win shares, less cost. However, some would argue (including me), that looking for a future face of the team, Dwight clearly is more personable and less mopey.
• It is also cool to look at how age dictates this graph (shown by color darkness).

In summary, using win shares to predict a salary is good for everybody at the table. As long as teams know how much they are over paying their stars, and can justify that, then it is acceptable. However, teams can get lost in purely market value and can overpay for the wrong talent (ie. Paul Allen over the years).

Stay tuned for a similar post regarding Steve Nash and any future deals that stick out to me. If you are wondering where a player is located on the graph or want to know a player's value, tweet me @jacobsudek.

---

Disclaimer: Keep in mind, I am no legal or CBA expert. Certain players may have a floor or ceiling on their contract and I may be unaware of that. Also, this is a small sample size, both in players and years. I also excluded every player that had a PER under 15 because I assume those players don't have a chance at being a top 6 player in the rotation (could shift the trend line slightly). It won't tell us everything like, intangibles, risk, rookie contracts, city, injuries, etc. But it can tell us, on average, what a player is worth based on their win shares. If the player is above the trend line, they are overpaid; if they are under the trend line, they are under paid. If they are within the band lines, their contract is probably spot on (adjusting for other factors that go unmeasured by Win Shares). Also, these win shares are based on a 66-game season because of last year's lockout.

Does the 4th Quarter Matter?

Does the 4th Quarter Matter? What a stupid question, right? Of course the 4th has to matter, right? In the regular season, you may want to think twice about the 4th quarter versus the first 3 quarters.


After looking closely at many playoff box scores this year, I found something that kept reoccurring; the margin at the end of the 3rd quarter.

• Out of all of the playoff games so far (after G2 of the Finals), 77% of the time a team leads after the 3rd quarter, they win the game.

It seems reasonable if you think about it further. If the game is reasonably close at halftime, the 3rd quarter is huge because its right after crucial halftime adjustments. Dove tailing off of that, the better the coaching, usually the better response the players will have in the 3rd quarter. In the 4th quarter, certain players take over, and it becomes more of a make-or-miss game, rather than an execution game. The 4th quarter becomes a lot of damage control, most of the time.

Here is a look at the correlation coefficient (0 to 1) of how well the average margin for a team correlates to their regular season winning percentage. I break it down between the entire game margin and just the margin between the first 3 quarters.

As you can see, the margin after 3 quarters isn't that far off from the final score. Obviously, the final margin should be pretty close to 1 as that is a clear indicator to wins and losses.

• To put this in perspective the average 1st half margin has a correlation coefficient of roughly .74 between the two years.

Look for the team that is leading after 3 to take the win. What do you think?

Clutch Debate - Kobe vs. LeBron

Between Kobe Bryant and LeBron James, who is more clutch? That would depend upon how you define clutch and how far you look into the scope. But for now, I am focusing on "game winner" type shots between 2000-2012.


Over the past few weeks, there have been a number of statistics from ESPN about how bad Kobe was in the clutch [this season or just in the playoffs]. The problem? The scope wasn't well rounded enough. I solved that problem. I took play-by-play data from the beginning of 2000 to May 20th, 2012.

Now, lets look at how people define clutch. Here are three possible definitions of clutch that I am aware of.

1. Stepping up to take the last second shot every time and not backing down from the pressure. (Columns 1 , 3, and 5, below)
2. Taking the last shot if it's open, otherwise, "making the right basketball play" and passing it to an open teammate. (Columns 4 and 5, below)
3. Purely looking at the shooting percentage and judging success from that; whether that is effective FG% or just FG%. (Column 2, below)

For this timeframe: I looked at all field goal attempts that were 30 seconds left in the 4th quarter or overtime. I then narrowed that further to a shot to tie or take the lead.

After looking at these results, it's a lot closer than people would think. The two columns on the right are out of the total possessions. So Kobe takes a shot or gets an assist 61.5% of the time within the current timeframe. 

With that data, Kobe puts himself ahead of LeBron in clutch success in all three clutch definitions.

---

Now, lets take a look at where they shot the ball on the court.

 

Right away, you can tell that they each focus heavily on two areas. Is there an advantage to one of the players' choices? 

• LeBron seems to either shoot a 3 pointer or finish near the rim. 
• Kobe tends to focus from the free throw line or further. 
• This supports our basketball knowledge of Kobe having his go to move. Kobe loves a few spots on the court where he will back down his opponent and shoot a ridiculous turn around fade away; splash. 
• From a defensive standpoint, it is easier to defend LeBron than it is Kobe. The defender can guard LeBron tight at the 3-point line and have help near the basket; which forces a bad shot at the rim or a pass. Where as Kobe, he is harder to defend because he has a vaster area without a gap in the floor (along with exceptional footwork).

---

Lastly, lets narrow the gap even further. The chart below now describes a shot taken with 5 seconds or less left in the 4th quarter or overtime. Also, the shot must be to tie or take the lead.

Wow. Look how bad LeBron is when it comes down to the very clutch. 

• Once again, Kobe wins all three definitions of clutch. 
• What's remarkable is Kobe only drops 2% eFG% and LeBron drops roughly 15% when it goes from 30 seconds left to 5 seconds left! 
• LeBron's teammates are actually more clutch than him. You can't say that for Kobe.
• Under this timeframe, LeBron averages making one field goal a season where Kobe averages 2.33 successes a season.

What do you think? 

State of NBA Basketball

With the game evolving over the years, what types of things are impacting the game?

The game is slowing down, that is for sure. Since 1979, the graph below shows the average possessions per 48 minutes for the entire league.

Click graph to enlarge

Very interesting. The days of the fast break are fading with no sign of a comeback. More and more these days, the teams that win games in the playoffs don't have fast break centered offenses. Example? The Phoenix Suns when Mike D'Antoni coached them. More recent example? The Denver Nuggets in the 2012 playoffs. Half court execution becomes more and more important. Teams that win championships tend to execute better in the half court and limit the risk for mistakes.

Piggy backing off of the less possessions per game, points per game have also decreased. That comes with no surprise as teams that take longer on possessions to score, will often score less. Furthermore, with a slower pace, there are less fast breaks, with less easy baskets, thus leading to less points overall. Below is the graph showing the points per game over the years.

Click graph to enlarge

The last graph (below) is obvious in some ways, but astonishing in other ways. For example, everybody knows over the years, the common 3 point shooter has been abundant. There have been more offenses that create dribble penetration and force kick out three pointers. However, because there are more 3 pointers attempted, why are the points decreasing? I know it is a lower percentage shot, but the 3 point shooting percentage has been constant since 1979 (roughly between 34%-36%). 

It's no lie that three point shooters have become more abundant these days. Likewise, big guys have become harder to come around, and thus, more valuable.

Click graph to enlarge

Thinking out loud here....we all know 3 point shots are worth more, and because teams shoot more, they should be scoring more points a game, right? For example, if a team shot all 2 pointers, instead of 3's, they would actually score less per game. A team averages roughly 18 3PA per game, with a 35% success rate, that gives them around 18 points. If you replaced those 18 shots with all 2 pointers, with a 45% success rate, that would give them roughly 16 points.

With FG%, FT%, and 3P% all being relatively constant throughout the years, the league has clearly changed and pace is to blame.