With the Lakers winning the first game of the Finals (and looking hella good doing it too) I need something to cheer me up during the insanely long wait for game 2. Drugs are certainly an inciting option, but I decided to go with the super geeky alternative: simulating the NBA finals as if it were a fantasy match up.
How? Way back in February, Erik and I took a look at the all-star lineups as if they were fantasy teams. Then I created a simulation program to simulate the all-star game as if it were a fantasy match up (Hint: The West destroyed the East).
Similarly, I modified that simulation program to take Lakers and Celtics lineups, sum up their stats across the nine statistical categories (points, rebounds, assists, threes, steals, blocks, FG%, FT%, and turnovers), and then compare each of those nine categories like you would in a typical H2H league to see which team wins the game. Since this is the Finals, it is a best of seven series; first team to 4 games wins the series.
Of course, things are not quite that simple. I had to make a lot of assumptions in the simulation about the rules of this league. I’ve included a longer write-up of those assumptions below if you are interested, but if you’re the type of person who doesn’t read any of the instruction manuals for the new crap you buy, then you might want to skip that part.
Because anything crazy can happen in just one simulation, I ended up simulating this Finals series 100 times. Here are the results of those 100 sims in terms of who won and how often.
Results of 100 Finals Simulations
Yes, you can take the ‘Celtics win 4-0′ out thanks to reality, but it was the least likely outcome according to the sim anyway. Overall, this is a close series fantasy-wise. The Lakers end up winning 53 out of 100 times while the Celtics win 47 times. Going into the playoffs, that’s probably what a lot of people were predicting for the actual Finals as well. In the sim, the Lakers were slightly more likely to win strong in 4 or 5 games (they did it 19 times in the simulation) while Boston was a little more likely to go 6 or 7 games in their wins.
Kobe Bryant: 39 times
Paul Pierce: 18
Rajon Rondo: 15
Pau Gasol: 14
Kevin Garnett: 12
Ray Allen: 2
Kobe was the runaway MVP of the series, winning it 39% of the time. Not surprisingly, he and Gasol were the only Lakers to take the MVP, while each of Boston’s big 4 took home multiple MVPs.
Comparing the Stat Categories
Not only can we look at the overall results of the series, but we can compare the individual stat categories across the entire simulation to see where each of the teams is strong or weak. Here is a review of the 9 cats with the percentage of times a team won that specific category.
FG% (Lakers 38%, Celtics 62%)
Any team with Ron Artest is going to have a leg down on FG%. Artest shot 41% from the field on the year, but Derek Fisher was even worse, shooting 38%. And Andrew Bynum’s great 57% shooting was offset by Perkins, who shot 60% from the field this season.
FT% (Lakers 58%, Celtics 42%)
A surprising loss for the Celtics given the Pierce (85%) and Ray Allen (91%) are two of the best free throw shooters in the league. Unfortunately for them, Rondo, Perkins and Tony Allen are horrible free throw shooters. The Lakers on the other hand, are a much more consistent FT% team, led by Kobe at 81% and Fisher at 86%.
Threes (Lakers 63%, Celtics 37%)
Overall the two teams are matched fairly evenly in threes with the exception of one position: PG. Rondo makes almost none, while Fisher average 1 a game this season. That’s all it takes to make a difference.
Steals (Lakers 44%, Celtics 56%)
On the other hand, Rondo led the league in steals. At 2.3 a game, he averaged twice as many steals as every player on the Lakers with the exception of Kobe and Artest.
Blocks (Lakers 50%, Celtics 50%)
I thought the Lakers would win this category consistently, but it appears that Perkins cancels out Bynum’s blocks and KG plus Rasheed equals Gasol. The rest of the two lineups don’t really have shot blockers.
Assists (Lakers 30%, Celtics 70%)
Rondo led the Celtics with 9.8 assists per game. Kobe led the Lakers with 5.
Rebounds (Lakers 68%, Celtics 32%)
Much like the real NBA finds, the Celtics don’t stand a chance on the boards. For the Lakers, Gasol averaged 11.2 a game, Odom 9.8 and Bynum 8.4. The Celtics on the other hand were paced by Perkins at 7.6 and KG at 7.3.
Points (Lakers 50%, Celtics 50%)
Another category I’m surprised the Lakers didn’t win. But while LA was mostly a one man show in the points category, Boston got contributions from their entire squad and even a little off their bench. The Lakers averaged about 2.5 more points a game as a team than the Celtics this season, so I’d expect the Lakers to come out slightly ahead if I ran this simulation again.
Turnovers (Lakers 58%, Celtics 42%)
Both Rondo and especially Perkins averaged more turnovers per game than their counterparts on the Lakers. Perkins was especially egregious at 2.1/game considering he only averaged 28 minutes a game.
Digging into the Simulation
A number of assumptions/rules had to be invoked to make this a worthwhile endeavor. Here they are.
The program I created simulates a single game at a time by randomly pulling a stat line from the game log of a player using their 2009-2010 box scores. This means that for any single game, Kobe (for example) might get his 44 point, 11 assist game from December 29th or his 10 point, 1 assist game against Dallas on January 13th. Or anything in between. The same thing happens for every player on the team.
Why am I using regular season stats and not playoff stats? One, I don’t have the playoff game logs yet, and two, the point of this simulation is to see who would win in a fantasy match up between the teams, not to predict the finals.
Yes, it is not a perfect system. The Celtics are (supposed to be) very good on defense and will likely hold the Lakers below their season averages on offense, Kobe and Gasol are playing out of their minds in the playoffs, rotations are different in the playoffs, etc, etc. The one thing that I did in the simulation is shorten the rotations so that each team only goes 8-9 deep. No need to get Adam Morrison or Brian Scalabrine involved in this; anyone who is not getting any playoff run had their stats zeroed out for the simulation as well.
After totaling up the stat lines for each team, there were compared like they would be in an H2H league. If the Lakers won 5 of the categories and the Celtics won 4, then the Lakers won that game. If the teams happened to tie at 4.5 categories apiece, that game was thrown out because there are no ties in the playoffs. Or crying.
After one of the teams won four games, the simulation calculated a Finals MVP by taking a rough weighted sum of each player’s series stats (for the winning team) and awarding it to the player with the highest score on his team.
The simulation was then reset and run again until 100 series were played out.