25 Games, 25 Million: Using 25 Games to Predict a Full Season

One of the criticisms most often leveled at Jeremy Lin is that he didn't play for a full season, or even a majority of one. Lin said himself, "people are always saying, 'He's only started twenty-five games, there's so many uncertainties.'" He's right, 25 games isn't a big sample by any stretch of the imagination, and, at least for the time being, it's all we have to judge him by. The question many fans want to ask is, had he played a larger number of games, how would he have done? This article is a little calculation heavy, so if you just want the results, just skip to the last paragraph.

Looking back at prior seasons, there have been young players who in their first 25 games of the season put up extraordinary numbers, but then fell down to earth by the end of the season:


As you can see, in the 2009-10 season, after 25 games Brandon Jennings was averaging 20.6 points per 36 minutes, thanks in part to a 55-point performance in his 7th NBA game. However, by the end of the season, his points per 36 average fell to 17.1, a decrease of 16.8% from his average after 25 games.

On the flip side, there have also been young players that exhibited the opposite:


In the same season, Stephen Curry was only averaging 12.7 points per 36 after his 25th NBA game, but ended up averaging 17.4 by season's end, an increase of 37.1%.

Are these results typical? To find out, I compiled gamelogs for players' seasons since 1985-86 where:

  1. The player was younger than 25 at the end of the season.
  2. In the first 25 games played in the season, the player started in at least 20 games.
  3. The player ended the season with at least 60 games played and started in at least 50 games.
  4. The player averaged at least 30 minutes per game for the entire season.

Next, for each player's season gamelog, I looked at their basic box score stats 25 games into their season, and compared them to what they were at the end of the season. In essence, it's exactly like what I did above for Brandon Jennings and Stephen Curry's points per 36, except for other stats as well, including total rebounds, assists, turnovers, field goal percentage, free throw percentage, and true shooting percentage. With the exception of the shooting percentages, all statistics were calculated per 36 minutes. With the data, I was able to determine that the distribution of the percentage change from 25 games to the full season for each statistic adhered to a normal distribution with the following averages (mean) and standard deviations:

Expected percentage change, 25 games versus full season

Statistic Mean Standard Deviation Sample Size
Points 3.06 8.93 441
Rebounds -0.72 8.17 415
Assists 3.15 9.63 223
Turnovers -3.90 9.08 340
Field Goal % 1.17 4.60 440
Free Throw % 0.81 5.23 366
True Shooting % 1.13 4.11 441

Yeah, that's a lot of numbers. Essentially what they mean is that based on historical data, the expected percentage change in points from 25 games into the season versus the full season was an increase of about 3%, but may vary from a decrease of 15% up to as much as an increase of 21%. In fact, because the data is normally distributed, we can say with 95% certainty that the change in points will fall between those two values (within 2 standard deviations from the mean). This means that Brandon Jennings's and Stephen Curry's change in points from above were statistical outliers. The most common outcome was more along the lines of John Wall's rookie season:


After 25 games, Wall was averaging 15.3 points per 36, and finished the season with 15.6 points per 36 in 69 games, or an increase of about 2.2%.

The other interesting observation about the data is that offensive stats (points, assists, shooting %) tended to go up, whereas turnovers and rebounds tended to go down as the season progressed.

What does this mean for Jeremy Lin? In Lin's 25 starts last season, he averaged 19.3 points, 8.1 assists, 3.9 rebounds and 5.0 turnovers per 36 minutes, and shot 44.5% from the field, 79.6% from the free throw line and had a 55.1% true shooting %. Based on the data from above, with 95% confidence, he would have been expected to finish up the season with 19.9±3.4 points, 8.4±1.6 assists, 3.9±0.7 rebounds, 4.8±0.9 turnovers, and shot 45.0±4.1FG%, 80.2±8.3 FT%, and 55.7±4.5 TS%.

No cursing in title. No pirated material, such as links to online game streams. Do not cut/paste entire sections of content from other websites. Thanks.

Log In Sign Up

Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join The Dream Shake

You must be a member of The Dream Shake to participate.

We have our own Community Guidelines at The Dream Shake. You should read them.

Join The Dream Shake

You must be a member of The Dream Shake to participate.

We have our own Community Guidelines at The Dream Shake. You should read them.




Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.