End-of-Season Projections

Back on November 3rd, while the NBA season was still in its infancy, HangTime blogger, Sekou Smith posted this blog with some end-of-season projections based on average points scored and average points allowed by each team.  At that point in time, no team had played more than three or four games.  Several teams had not yet lost a game while others had not yet won a game.  As you might expect, the results were laughable, projecting three teams would top the 70 win mark with Miami winning a mind-boggling 78.  That means since they had already lost a game, they would have to go almost an entire season while only losing three more.

Whether Smith wanted to get fans riled up or was just entertained by the idea of such preposterous numbers, I knew their had to be a better way to generate projections, even early in the season.

So I got to work.

The formula Smith presents is called the Pythagorean Method, which gives you a ratio based on average points scored raised to x divided by the sum of average points scored raised to x and average points allowed raised to x.

I took this same formula as a starting point and started addressing the issues:

What if a team is having blowout wins and close losses (i.e. Miami Heat)?

I've included a team's actual win percentage in the formula and weighted it to have more and more significance as the season progresses (starts at 50% and increases to 100% by game 82).  If a team like Miami continued their trend until the end of the season, the projection may still say 78 wins, but the standings would tell a different story.  Including this variably weighted factor brings projections closer to reality as more reliable information is collected.

What if a team has a difficult or easy schedule?

I've also included a Strength of Schedule factor that will either bump a team's projection up or down based on the quality of opponents they have recently been playing against.  The Strength of Schedule is calculated by taking an exponential moving average of each opponent's projected win percentage.  I opted to use the projected win percentage over the actual win percentage for all these same reasons.

What if a team starts out undefeated or winless?

 I was a little stumped by what to do with this issue at first.  I thought of carrying over each teams' record from the previous season as a starter, but decided that would cause inaccuracies due to player movements during the off-season (i.e. Cleveland).  I thought about solving that by narrowing the spectrum and limiting it to .400-.600 or so.  That still wouldn't solve problems caused by teams like Cleveland (or more accurately, people like LeBron).  I finally settled on assuming each team was average and project them to finish .500.  This average number is what starts the Strength of Season calculation and allows it to reach realistic calculations early on.

There are other issues that could also be addressed, such as home-court advantage (home team wins approx. 57%) and inter-conference games (west beats east approx. 55%), but I think I've managed to address the most glaring issues.

Here are some end-of-season projections using my current system:

One game in:

Standing	Eastern Conference	Wins	Western Conference	Wins
1	Cleveland	50	Denver	49
2	Orlando	49	Dallas	48
3	Atlanta	48	San Antonio	48
4	Boston	47	Oklahoma City	48
5	New York	47	New Orleans	47
6	New Jersey	47	LA Lakers	46
7	Detroit	35	Sacramento	46
8	Milwaukee	35	Portland	45
9	Toronto	35	Golden State	44
10	Miami	35	Minnesota	36
11	Chicago	34	Houston	36
12	Indiana	34	LA Clippers	35
13	Philadelphia	34	Memphis	34
14	Charlotte	34	Phoenix	34
15	Washington	33	Utah	33

Five games in:

Standing	Eastern Conference	Wins	Western Conference	Wins
1	Miami	58	LA Lakers	64
2	Atlanta	57	New Orleans	62
3	Orlando	53	San Antonio	50
4	Boston	51	Golden State	49
5	New York	46	Portland	48
6	Cleveland	39	Sacramento	46
7	Chicago	38	Denver	46
8	New Jersey	36	Dallas	45
9	Indiana	33	Oklahoma City	41
10	Toronto	32	Memphis	38
11	Milwaukee	29	Utah	38
12	Philadelphia	28	Phoenix	37
13	Charlotte	27	LA Clippers	30
14	Washington	27	Minnesota	26
15	Detroit	22	Houston	23

Ten games in:

Standing	Eastern Conference	Wins	Western Conference	Wins
1	Boston	57	New Orleans	65
2	Miami	51	LA Lakers	57
3	Orlando	50	Dallas	57
4	Chicago	47	San Antonio	56
5	Atlanta	44	Utah	52
6	Milwaukee	44	Phoenix	49
7	Cleveland	38	Oklahoma City	45
8	Indiana	37	Portland	44
9	New Jersey	35	Denver	43
10	Detroit	34	Golden State	41
11	Charlotte	30	Memphis	38
12	New York	28	Houston	31
13	Washington	28	Minnesota	29
14	Toronto	28	Sacramento	28
15	Philadelphia	26	LA Clippers	21