Fantasy Basketball drafts can be tough to do without some sort of plan/ranking system ahead of time.For example, your turn comes up. Which player do you choose to build your team around?:
Player A: 19 pts, 0 threes, 10 reb, 3 ast, .5 stls, 1.5 blks, 529 FG%, 823% FT
Player B: 25 pts, 6.5 reb, 4.5 ast, 1.5 stl, 1 blk, 500 FG%, 758 FT%
Arguments could be made either way. So I tried preparing an objective ranking system for Rotisserie leagues. It can be applied/altered for Head-to-Head leagues as well.
1. Projections for each player
To make player projections, one could make a weighted average of each player’s past few seasons. Basketball-monster’s already done that: http://www.basketball-reference.com/friv/projections.cgi
(See https://amateurstats.wordpress.com/2011/10/17/fantasy-football-rankings/ for a simplified breakdown of doing this manually)
It’s a lot to go through their projections, but if you copy+paste, and delete the useless players’ projections, you have a leg up over basing a player on just their past season.
The only thing is that these are projections for how well they’ll do if they all played 36 minutes. That’s not true for most players. So I estimated each player’s minutes played, based on last year’s totals and changes to their team’s depth chart.
Finally, to get the players’ projections, multiply their per-36 projections by the number of minutes you expect them to play.
2. Weighing FG% and FT%
These stats should have each player’s FGA, FGM, FTA and FTM. You could calculate the FG% and FT% that way, but more importantly is to take volume into account.
For example, Durant making 90% FT at 9 FTA/game is much more valuable than Nash’s 90% FT at 3 FTA/game.
To adjust for volume, multiply FGA * (their FG% – league FG%). Let’s call this xFG.
Brandon Jennings’ xFG = 15.1FGA * (.397 – .473) = -1.15
That is probably the worst xFG, while the best is about +1.5 from Dwight.
3. Finding z-values (standard deviations above the mean)
The basketball equivalent of step 2 in my baseball post (https://amateurstats.wordpress.com/2011/03/22/building-fantasy-player-valuations/)
With your sample of players, find the average of each of the 9 categories.
Find the standard deviation of each category as well. Excel does so with =STDEV(H3:H150)
To find a player’s contribution in one category…
(Their Pts – league average Pts)/Standard deviation of Points
Durant is 3 standard deviations above average, meaning he’s the top 1% of scorers available.
If you do that for all the categories, you can add up their contributions to get their value.
4. Positional adjustments
The value above will give a general ranking that may undervalue starting PGs. The reason being that each team only has 1 player that can rack up ASTs, while 3s, stls, etc can be obtained by any number of players.
To adjust for PG or C scarcity, find the replacement level of each player. Basketball has a lot of players who qualify for more than one position, so instead of forcing a SG/SF/PF into one position, I made 3 simplified positions.
G – Players with more than 4 asts.
C – Players with more than 8 rebounds
F – all other players
This gave a somewhat balanced number of players for each category. I found that the 30th ranked player in each category were Rashard Lewis, Daren Collison, Marcus Camby. Droppable players, but also addable.
All players were then added points so that these 3 players had values of 0.
Gs had a adjustment of 1.77, Fs of .85 (intuitive sense, as mediocre wing players to shoot 3s can be found more easily than mediocre PGs), and Cs had a 1.29 adjustment.
These may be more accurate if I had used a sample of more than 133 of the top players in recent years. I don’t think the attached sample overrates bigs, but the more players used, the better.
Finally, to account for the 2 UTIL positions, I subtracted the 104th best player’s value from the 96th best player’s value. Demar Derozen and Mehmet Okur are replacement level players after this change.
5. Calculating $ Values
If you’re doing an auction draft, ranking players from top to bottom isn’t enough.
To convert z-score value to $ value, take all the positive z-scores and divide it by the number of $ in the draft.
I prefer using http://basketballmonster.com/PlayerRankings.aspx to see how many positive z-score values there were. As a collection of data from the past, it’s more accurate than my sample that may have left out players that have contributed value in the past.
280 z-scores were contributed, and a 12 league team with $200 budgets gives an economy of $2400.
Each z-score is worth $2400/280 z-scores = $8.59/z-score
Multiply that by each players value to get their final projected $ values.
Below is the file for reference. The first 3 sheets are the only used. Enjoy.