There are a lot of factors to balance in trying to determine what fantasy players to target during drafts, and it does sometimes feel like juggling cats. One of the markers that is sometimes unremarked, but that is helpful to use when studying players for drafting in fantasy is to look at what players did in the second half of the year before. This is particularly useful in predicting performance in the upcoming season for three reasons:
- players are likely to be used in patterns and circumstances more similar to those in the second half of the previous season than the previous season overall;
- players prone to improvement or decline (through age, experience, longterm health issues or simply through the natural process of making adjustments to their game) are likely to show improvement or decline within the previous season, including from first half to second half; and
- particularly with regards to 2011-12, players had a long offseason break and abbreviated preseason, so second-half stats would be more reflective of performance under normal circumstances.
There are many reasons why you should be cautious of such numbers, of course. Injuries mount in the second half of many seasons, so rotations are disturbed. The sample size of the stats is smaller than a full season so statistics are less reliable as a predictor of actual quality of play or of player quality.
What we find in other sports I have studied, baseball in particular, is that examining both full-season and second-half stats are marginally better predictors overall of the following season than the overall stats alone. Both need to be looked at (along with much more, naturally) for a comprehensive picture, but second-half stats are at least useful, and where players have demonstrated marked improvement it’s well worth looking at the reasons why.
With all this in mind, I thought I would share some of the numbers I crunched from last year’s second-half statistics. If you want an easy place to acquire last year’s second-half only stats (in this case, from each team’s last 33 games played) you can get them in raw format from Doug’s stats page. (Always an excellent resource for raw stats; a very useful feature during the season is his “Last 10 Games” stats compilation).
What I will share here is a custom measure tailored for a standard format, nine-category league (FG%, FT%, 3PTM, PTS, REB, AST, BLK, STL, TO). I obtain this custom measure by looking only at the better slice of NBA players, so only the 160 players with the most second-half-of-the-season minutes, plus a few starters who didn’t make that category. In this case, we study 198 players in all. Normally, for a smaller league, I use a few less players than that, but essentially the 198 top players are pretty much the fantasy universe for most leagues.
The rating (which is far from the only one I use, but is one which I find easy to calculate and helpful) is the total number of standard deviations, in each category, that the player is above the average for this group of better players. What I’m trying to get at is what a player is doing to help you win or lose each of the nine categories while he is on the floor.
Crucially, these are per-minute stats. Since I am looking primarily at second-half performance as the relevant indicator here, how well a player played, I find per-minute stats the most useful. Since I develop my own measure of how much each player is likely to be used in the coming season, I concentrate on quality instead of quantity. The percentage measures are weighted by how many shots a player takes, so that high-volume shooters move up or down on that basis. So I measure total successful makes above or below average expectation: for this group of players, it was 46% for field goals and 77% for free throws.