How to Draft a Fantasy Basketball Team Using Analytics

Describing an analytical approach to fantasy basketball ranking using Hashtag Basketball's projections and Modified Rarity Index Scoring.

Building a fantasy basketball team in a traditional category-style league is much different than building a fantasy football team. Fantasy football is much like fantasy basketball "points" leagues which removes a lot of the strategy involved in fantasy basketball. To win in category-based fantasy basketball, you need to make sure your team has a good mix of strengths in each category.
Based on the excellent projections done by Hashtag Basketball's Marc Roberts, I have ranked and analyzed the players based on position and category.

The Ranking System

The ranking system I have created is called Modified Rarity Index Scoring (MRIS). This system is similar to Standard Scoring, but with some additional features to compare value across categories. The result of a MRIS system is "Fantasy Point Equivalent" (FPE). The basic premise is that 10 FPE in steals is equally as valuable as 10 FPE in assists or field goal percentage impact.

The MRIS system calculates the minimum expected value from the worst player in each category and subtracts that contribution from the player's stats. Then, it calculates the total sum of each statistical category across the entire league. Each of the stats is set equal to each other, accounting for statistical rarity. Each Fantasy Point Equivalent is one point above the minimum expected (the "zero" point for the category).

For example, in a 150 person league, the expected points is 9.7 for the worst player in the league. Therefore, a player who averages 19.7 points generates 10 FPE in points (average - "zero"). Since the worst player in threes in a 150 player deep league will contribute minimal threes, the "zero" point is much lower - 0.4. To make threes equivalent to points, each three is worth 5 FPE (actually 5.11). This means averaging 2.33 threes per game is equally as valuable as averaging 19.7 points.

The Category Comparisons

Below are plots of production in the nine standard fantasy basketball categories. On the horizontal axis are the top 120 players by MRIS scoring. The vertical axis is their contribution to the total FPE (Fantasy Point Equivalents) in the given category. Please note that while games played does affect the overall rank, it is done later in the calculations, so each category is not weighted by projected games played. Since Embiid is projected at 55 games, this doesn't have much of an effect on the graphs!

To go through each category, I'll point out general trends and also point out the 20th ranked player based on the projections and MRIS scoring - Kyle Lowry.

The general strategy that your early picks need to concentrate on points is overstated, but you can see that the 9 players that are projected for over 15 FPE in points are all projected in the top 20 overall. There are still players you can target late in drafts (such as DeMar DeRozen, ranked 46th overall who is projected at 14.9 FPE and Andrew Wiggins ranked 88th but projected with 12 FPE in points), but not many - and they are likely going to be drafted too early anyways. It is definitely advised to target at least one of the top points producers in at least one of your first two picks for these reasons. If you don't, it is important to concentrate on above-average scorers later in the draft.

You can see Lowry's projected 21 points is 11 FPE, just above the red regression plot at the 20th ranked position.

Unlike points, rebounds are much easier to find later in the drafts. The top two projected rebounders this year are found at rank 29 (DeAndre Jordan) and rank 55 (Andre Drummond), although their value is a bit convoluted due to their free throw shooting. Even disregarding those two, there is lots of value to be found throughout the draft. 

You might imagine Lowry doesn't contribute a lot in rebounds, and you see the 20th ranked player by these projections just above 0 - his 4.8 projected rebounds are only worth 1.95 FPE.

While assists are also found throughout the draft, the value is much more concentrated in the top producers. This is obvious if you look at the value of the top players in assists - there are four players projected above 30 FPE and ten more above 20 FPE. If you want to compete in assists, you need to target the top producers. While players like Elfrid Payton are projected at 20 FPE in assists, the rest of his projections work out to zero (20 total FPE thanks to his lack of general stats and negatives in his shooting and turnovers). 

Lowry is a good contributor in assists with over 20 FPE (projected at 6.5 assists).

Even more than rebounds, threes are easy to find later in drafts. The general consensus that threes are the easiest category to find on the waiver wire is seen to be true with significant value in this category in the 120-200 range (not shown on the plot above). 

Lowry is projected at 2.8 threes per game, which works out to nearly 13 FPE. 

Interestingly (if you like math), you can see that steals are easy to come by in small amounts. It is very common to see players average 0.5 steals per game - much more likely than blocks! This makes steals even more "rare" in the MRIS system since the "true zero" is relatively high for such a rare statistic - the "zero" point is 0.7. Each steal is basically as equally valuable as a block, even though there are more steals overall than blocks. There is value throughout the draft in this category. Finding role players that average over one or even two steals per game is a great way to find a great value on the waiver wire.

Lowry's projected 1.7 steals per game make him a great value in this category.

You can see in the above plot that the value in blocks is concentrated in fewer players than in categories like steals. The top projected contributors in steals are under 20 FPE, while we have seven players projected over 24 FPE in blocks. Targetting multiple players with high value in blocks can win this category consistently throughout the season - but the variance in such small numbers will drive you crazy!

Kyle Lowry is projected at 0.4 blocks, which brings him very slightly above the "zero" mark at rank 20 in the projections.

Percentages and Turnovers

These categories are the only ones that can have a negative effect on your team of the standard categories. This makes these categories easier to "punt" to build value in other categories. While you will hear many experts disagree about what constitutes "punting", the most value can be gained by punting these categories since they can change the value "delta" between two teams if one is punting and the other is not.

The MRIS system can be designed to rank players for the variance of a weekly scoring league or the variance of a season-long league (rotisserie scoring). When doing preseason scoring, I prefer to use rotisserie scoring since that will be the overall contribution to your team's value. In simple terms, weekly scoring leagues discount the value of good and bad shooting because there is more variance in percentages (it is imaginable to shoot really well or really poorly for short periods of time, but it rarely happens over long periods).

The effect of poor shooting is based on both percentage and volume of course, and the graph of impact shows there is value in this category throughout the draft. 

Lowry is projected to shoot 44% and has an overall negative effect on your team's field goal percentage.

Very poor free throw shooting can have a catastrophic effect on your team's free throw shooting. DeAndre Jordan scores the equivalent of negative fifty points for your team with his projected 44% free throw percentage. 

Kyle Lowry is projected to shoot slightly above average (81%) but at a decent volume (5.6 attempts per game). Overall, it works out to 6.3 FPE. 

Many experts feel that turnovers should always be ignored when building a team. The graph above shows why - there is a very high correlation between turnovers and overall value! Of course, just like with percentages, if you build a team around players that tend to avoid turnovers you will be stronger in the category. I usually find myself punting turnovers but I don't have it as a hard rule and I recommend doing the same.

Lowry is projected to turn the ball over quite a bit (as you'd expect most point guards to do!) and is projected at -14 FPE in the category.

I hope this was informative - if you have any questions or feedback on the system or how to use projections to build a great fantasy team, leave a comment or find me on Twitter!

Like what you've read? Share it with your friends on      or