How To Draft In Fantasy Football
Last year, I outscored my opponents by more 300 points. It was a cake walk all the way to the championship game, where a middling 6–7 team absolutely decimated me 178–100 (with the help of a 50+ point performance from Alvin Kamara and 30+ from Mike Evans). I’ve never scored so many points before in a league. Yet, I’ve never considered the draft from an opportunity cost perspective before.
Consider the following in terms of performance for the top players in any position:
A couple things should stand out immediately:
- In terms of overall points, QB>RB>WR>TE
- There are noticeable drop-offs for every position.
Now we should note that fantasy teams are usually composed of 10 players, and sometimes they have 8, 12, or 14. The team structure usually comprised as follows: QB, RB, RB, WR, WR, TE, FLEX, K, DEF. Sometimes a defensive player is included — notice the performance drop-off for defensive players and team defenses is the modest and the overall scoring is near the bottom.
In terms of overall points, a person might be tempted to draft a QB first. But the error here is in terms of opportunity cost.
Most people recognize the importance of drafting a RB first. This is typically correct but misinterprets why.
Opportunity cost refers to lost value in making a specific selection. A typical draft will use a snake format. For pick order that will go 1,2,3,…,8,9,10,10,9,8,…,3,2,1,1,2,3,… and so forth. So every position has to approach the draft in a different manner given the available players.
Consider the first position. What is the best player available at every position? What would be the next best player available at a given position assuming every next pick by each team is at that position? In the first position the best position picks score as follows:
QB: 395
RB: 314
WR: 243
TE: 208
If every team drafted the best available player at each position, what would come back is:
QB: 333
RB: 142
WR: 137
TE: 93
The opportunity cost at each position is the difference between the two aforementioned values:
QB: 62
RB: 172
WR: 106
TE: 114
Therefore, the greatest lost in potential points would be RB in the first position. It is therefore appropriate to pick the best RB available.
This trend continues for the first few picks, but then it starts to change. Consider the graph from before. Where do we expect it to change? Opportunity cost is about ensuring maximum value, so we should expect it to change where there is a noticeable drop in performance. In this case, this would be position 4. At position 4 the best available, next best available, and opportunity cost are as follows:
QB: 395; 360; 35
RB: 217; 151; 66
WR: 243; 155; 88
TE: 208; 87; 120
At position 4, the value ranking is TE>WR>RB>QB. It is important to note that this is based on actual performance rather than predicted performance. Predicted performance is more important here since it would push this value ranking to position 6 with the inclusion of two RBs with high evaluations who became injured during the 2020 season (McCaffrey and Barkley).
In maximizing opportunity cost, we maximize our overall player scores, team scores, and increase the likelihood of winning.
To understand this simply, think of the performance graph where each position has a cliff, followed by a gently sloping valley. Our goal is to get on top of as many cliffs of possible and avoid the valleys.
Note that during any draft, the projected performance data indicates there are usually less than 20 players sitting on the cliffs — so the first 2 picks are critical.
The 1700+ point Draft
Following this approach, I thoroughly expected to take a TE and WR with the 8th spot in our 10-person draft. I expected to take Travis Kelce first and Davante Adams second — since most people draft RBs with hopeless abandon. However, the opportunity cost changed when the 5th, 6th, and 7th picks all passed on Dalvin Cook. Dalvin’s projections put him at the edge of the theoretical cliff, improving the opportunity cost for RBs, and netted my team 294 points (only 234 projected). Kelce was then drafted 2nd.
By the time my 3rd pick came around, every cliff player was taken except for one — Lamar Jackson QB. Jackson was projected to be one of the two top rated quarterbacks but ultimately finished 10th overall (just behind Justin Herbert, who I picked up off the waiver wire). My fourth pick went to Chris Carson. He was a the last remaining RB before a second drop-off in RB production.
Notice that for RBs and QBs there are multiple performance cliffs. While WRs and TEs essentially have only one.
Picks 5 and 6 went to Tyler Locket and Keenan Allen. 7 and 8 went to Robert Woods and Ronald Jones II. These picks were workable starters. The remaining picks after round 8 typically don’t matter — the opportunity costs are minimal that their value is found more in situational opportunities, bye week replacements, and waiver wire substitution.
It is worth noting that projections do not always match up with reality. Players get injured, don’t perform, or show unexpected improvements.
On injuries, of the top-10 predicted players at each positions, 70% of QBs were placed correctly (1 key injury — Prescott), 50% of RBs (3 key injuries — McCaffrey, Barkley, Mixon), 50% of WRs (3 key injuries — Moore, Jones, Golladay), and 50% of TEs (1 key injury — Kittle).
There is one improvement that can be made when drafting according to opportunity cost — factoring in likelihood of draft. Opportunity cost is maximized in my analysis (assuming every team picks the next 10-to-20 players of a given position). Of course, this is a extremely conservative assumption. It is much more realistic to make round specific assumptions such as:
Round 1: 9 Max RB picks; 3 Max WR picks; 2 Max QB picks; 2 Max TE picks
Round 2: 8 Max RB picks; 4 Max WR picks; 2 Max QB picks; 2 Max TE picks
Round 3: 6 Max RB picks; 5 Max WR picks; 2 Max QB picks; 3 Max TE picks
Round 4: 3 Max RB picks; 5 Max WR picks; 3 Max QB picks; 4 Max TE picks
Round 5: 3 Max RB picks; 6 Max WR picks; 4 Max QB picks; 3 Max TE picks
Round 6: 3 Max RB picks; 6 Max WR picks; 4 Max QB picks; 2 Max TE picks
Round 7: 2 Max RB picks; 5 Max WR picks; 3 Max QB picks; 3 Max TE picks
Round 8: 2 Max RB picks; 4 Max WR picks; 3 Max QB picks; 3 Max TE picks
These assumptions would provide a more realistic opportunity cost tabulation that includes a reasonable amount of risk.
Looking at opportunity cost from positions 1 and 4 from before, this would be the results (best available; next best; opportunity score):
Position 1:
QB: 395; 369; 26
RB: 314; 148; 166
WR: 243; 179; 64
TE: 208; 108; 100
Position 4:
QB: 395; 360; 35
RB: 217; 166; 51
WR: 243; 155; 88
TE: 208; 108; 100
Overall, I hope this helps provide a better understanding of draft strategy and insight into the mathematical considerations that provide value to this game.
