This Featured Article was originally posted at RotoInfo.com. We have reposted it here for reference.
Ever been at the draft table and someone says: "I always go RB in the first two rounds." Well, as the inventor of The Machine, I say, doesn’t that depend on the scoring system. With certain roster restrictions and PPR leagues, maybe, but let’s dive into what I learned in the last three years playing in the FSTA Experts League.
First, The Machine, by Advanced Sports Logic, Inc. has performed better than the humans. The Machine achieved a 73.2 win percentage and made the playoffs each year. Seventy-one other teams have competed in one of two FSTA Experts’ Drafts over the three year span. Only two of the 71 other teams scored more fantasy points in a single season than The Machine’s average (2041 per year). The Machine also holds the record for the highest total fantasy points in a season (2191) and best win-loss record (13-1).
And guess what…The Machine used its nine picks from the first three rounds over the three years to select 8 WRs and only one RB.
The year The Machine had the best win record and the best all-time fantasy point total it did not draft a RB until the 5th round. So are we saying don’t draft RBs in the 1st round. Certainly NOT!
So, first the rosters. The FSTA starting lineup requires 2 RBs, 3WRs, and 1 TE. Also, WRs and TEs score 1 point per reception (PPR), but RBs only score 0.5 PPR. Thus, the greater than average depth at WR and the difference in PPR skews fantasy points toward WRs more than in most leagues.
The following tables show actual fantasy point results from weeks 1 thru 16 in the 2012 season.
The Machine makes selections on the ability of those players to improve a team’s probability to win. While it would take about 100 pages to describe exactly how The Machine works, the short of it is that The Machine creates a fantasy point probability distribution for each player in your league. Then starting with these probability distributions and using your league rosters, matchups, and playoff schedule, it simulates your entire fantasy season to determine exactly the impact how any decision you can make affects your probability to win, from your first pick of the draft to the end of the season.
Let’s consider five possible scenarios:
- Scenario 1: All teams pick an average mix of 3WRs, 2RBs and 1 TE in the first six rounds.
- Scenario 2: Six teams (Group A) draft in the order of WR, WR, WR, RB, RB, TE and the other six (Group B) draft in the order RB, RB, TE, WR, WR, WR.
- Scenario 3: The Machine drafts in the order of WR, WR, WR, RB, RB, TE and the other 11 teams draft in the order of RB, RB, TE, WR, WR, WR.
The table is color-coded back to the previous fantasy point table to help you understand the analysis. Scenario’s 2 and 3 show that drafting three wide receivers with the first three picks always yields the best result. Scenario 3 shows that when all teams, except The Machine, select RBs and TEs in the first 3 rounds, The Machine gains a 14.5% fantasy point advantage by selecting the top 3 WRs.
So why does picking 3 WRs generally gain an advantage over picking 2 RBs and a TE in the early rounds?
When selecting a WR, it helps three starting positions on your team and hurts three starting positions on everyone else’s team. The greater the extent of one team selecting WRs while others select RBs and TEs, the greater the extent of the advantage.
The Machine understands this dynamically at every point in the draft. DVBD algorithms do not factor in win probability.
This effect alone does not explain the degree to which The Machine has dominated the FSTA Experts’ League. Many team managers feel they can fill in with WRs later in the draft or throughout the year. The Advanced Sports Logic team has been doing a lot of analysis in evaluating upside potential of players based on tier and position. Our analysis has found that the reality is that it’s easier to find RBs with upside in the deeper rounds of the draft than WRs.
Normally 2nd and 3rd string WRs are on the field for most passing plays. Whereas 2nd and 3rd string RBs are on the bench for most running plays. This means 2nd and 3rd string WRs are already playing close to their potential and when a 1st string WR is unable to play, the 2nd and 3rd string WRs get more pressure and have a more difficult time getting open than when the 1st string WR was on the field. However, the opposite is true for RBs. When the 1st string RB is unable to play, the 2nd and 3rd string RBs get more carries. We also have evidence that running backs tend to decline and get replaced at a younger age than wide receivers. This means that a backup running back can often take over mid-season simply due to declining play, whereas this drop-off is not evident for WRs. This often leads to backs we have barely heard of being snatched late in fantasy drafts doing very well - think Alfred Morris, Joique Bell, and Vick Ballard from last year. While only one of them was a fantasy super star, all of them came from nowhere to produce decent flex starter numbers.
Let’s go to the numbers.
The first chart shows how AccuScore running backs projections changed from 12 weeks before the NFL season began to 15 weeks into the season. The data is broken into five tiers based on the projections 12 weeks before the season. Tier 1 includes the top ten RBs, Tier 2 includes the 11th-30th, Tier 3 31st-100th, Tier 4, all RBs after the 100th that were projected to score fantasy points in week 15, and Tier 5, RBs that either had no projections or were projected to score 0 fantasy points.
The blue and red lines show primarily the starting RBs in the NFL. The top 10 had more downside than upside and the 11th thru 30th stayed about even with their initial projection. The green line shows the majority of 2nd and 3rd string backup RBs in the NFL. You can see the green line is significantly skewed to the right and the right end tail jumps to over 10%. This means that more than 10% of the 2nd and 3rd string running backs did 2x or better than originally projected.
Now moving to WRs, the Tier 1 have less downside than RBs and about the same injury rate. The Tier 2 and Tier 3 WRs have far less upside and far lower injury rates than RBs. For example, looking at the red line, no WRs in the Tier 2 group scored more than about 50% better than their original projections and in the Tier 3 group, only 5% scored 2x or more than their original projection. The 4th tier are the 4th string WRs that started the season on the bench. Only at the 4th Tier do WRs start to have significant upside, compared with RB 2nd and 3rd tier.
So whether you take a RB or WR will depend very much on your league scoring rules and starting lineup requirements. Traditional strategies cannot be applied across all leagues.
However, knowing more about how player projections typically panout and why, and thinking about impact to the depth of your starting lineup and your opponents’ starting lineup as you take players off the board can help you make better draft decisions.
Or, just get The Machine and let it do all the deep thinking to figure out when you should draft which player positions, and based on its recommendation you can add your football knowledge to decide the best players available at each point in the draft.
Founder of Advanced Sports Logic, Inc. and inventor of The Machine