This is easier than I expected it to be, but then again, isn’t everything? – Sterling Archer
I think I’m better than average at picking players, and I assume you do as well. I think we can fairly stipulate that we’re all better than average at picking players just like we’re all better than average at driving cars.
That said, if we define the quality of a pick purely through the lens of history, I make a lot of terrible picks.
In Part 1: Controversial Conjectures and the 2016 MFL10 of Death, I mentioned that I was lucky enough to win the 2015 contest. That’s not false humility:1 you need both general and specific luck to win an MFL10, and fate obliged.
Fantasy Douche recently referenced the infamous Sam Hinkie letter and the value of writing down predictions.
I’m often shocked when I go back and read old articles (that I remember as being correct) because I’ll discover that I made four predictions in the article and maybe half of them turned out to be right. Or maybe they turned out to be true, while the underlying reasoning was wrong. That’s the nature of making predictions, which is what the NFL draft is at its core. Hopefully by writing down those predictions we can either get better at making them, or become more humble.
I like to go back and look at my drafts in leagues I’ve won. This has the benefit of being less depressing because you can console yourself with the outcome, and yet simultaneously reinforce just how limited is our ability to predict the future.
2015 MFL10 of Death – Unfortunate Picks
|Player||Avg Finish||Win Rate||ADP||N Selected|
|Sankey, Bishop TEN||6.98||0.066||91.52||1021|
|Perriman, Breshad BAL||7.52||0.045||115.77||963|
|Bradford, Sam PHI||6.74||0.071||151.78||935|
Late Round Miscues
|RD||Player||Avg Finish||Win Rate||ADP||N Selected|
|Round 14||Marquise Lee||7.07||0.048||179||931|
|Round 15||Kenny Britt||6.36||0.096||191.54||858|
|Round 16||Jaelen Strong||7.35||0.05||165.96||959|
|Round 20||DeAnthony Thomas||No one silly enough|
The injury to Perriman was not strategy-related, but you can see that 7 of my 20 picks contributed very little, although Sankey’s fluke 22-point game in Week 1 provided the ultimate margin of victory.2 So even though I concluded in Part 1 that my win was structural and not player driven, there appears to be plenty of room for tinkering.
Just Pick Strategies
Part of any MFL10 success is going to be injury luck, but beyond that I want to make sure I’m using the positional allocations – and orders – that will provide the best chance at victory. My No. 1 operating thesis is this:
MFL10s are about buying the maximum number of player touches you possibly can.
I addressed the sexy positions of RB and WR from a rudimentary perspective in Part 1 and plan to expand in the future. In the present, let’s look at the unsexy keys to MFL10 victory.
Team Defense is Boring. It’s Also Where Championships Are Made.
In both the 2015 and 2016 MFL10s of Death, seven teams selected three defenses and five went with just two. Thirty-one total defenses were selected. Last season 2-DEF teams finished second, third, and fourth in our league, lending plenty of credence to the idea that you don’t have to select three.
And you don’t. Last season two of the top five positional allocations in terms of win rate were 2-DEF.
But I still think you should select three defenses, and here’s why: Selecting three defenses is an easy way to lock in points without having to actually make a good pick. Here was the defensive scoring of the aforementioned top four finishers.
I only defeated Denny Carter by nine points to win the league, but I owned a whopping 254-149 advantage on defense. If I hadn’t bested Denny on defense, I would have gotten crushed.
This fits with our pre-2015 intel that suggested three defenses was the optimal strategy.
A Shift in Thinking: Back to Three QBs
I entered 2015 with the idea that you couldn’t select more than two QBs if you wanted to maximize your chances. I now believe that to have been a mistake.
Why 2-QB Seems Pretty Straightforward
In updating optimal positional allocation for 2015, FD offered the following results.
|Top 10 Roster Combinations||2.2||5.6||7.2||2.6||2.4|
|Bottom 10 Roster Combinations||3||6.4||6.2||2.7||1.7|
Or, to look at it another way:
If it seems so obviously better to select two QBs, why would we select three?
Here’s where I think it matters where you select the QBs. My conjecture is that a lot of the 3-QB squads selected their initial QB in the first 7-9 rounds, which is an abject disaster. Early QB squads are contaminating our 3-QB sample and pointing us in the wrong direction. There’s actually quite a bit of evidence to suggest you should select three quarterbacks as long as you do it in the double digit rounds. The first comes again from Bessette and Meade. Here’s their optimal lineup, by round drafted:
RB TE RB RB RB WR WR WR QB WR TE WR QB QB WR WR K DEF K DEF DEF K
Keeping in mind that I’ve offered some evidence for a Zero RB approach instead and that we’ll be offering more shortly, we can still draw at least some tentative conclusions from their study. Their optimal lineup includes three quarterbacks, and once again that’s with only 19 rounds after you remove the kickers. As they mention in their piece, rostering 3 QBs provides some of the same advantage you get from three defenses.
Even at pick 175, you can expect 15 points per week out of a quarterback however, and the increase in standard deviation means that these low-end starters have more room for improvement against bad defenses or with a few big plays. No other position has such a combination of staying power and late-round upside.
Then we have this from the 2015 results.
Top Positional Distributions: Rounds 11-20
There are 80 positional distributions that were selected 100 or more times. Of those, the top two in terms of average finish were 3-QB and six of the top 15. Meanwhile, there was only one 3-QB distribution in the bottom 40, a group that is littered with 1-QB and 2-QB distributions. I asked RotoDoc his thoughts, and he replied that fitting a simple regression model to the Round 11-20 data table returned 3-QB as the optimal lineup.4
Finally, the best5 full-20 distribution selected more than 1,000 times was 3-QB, 3-DEF.
This fits with my overarching theory that you want to buy the most snaps and touches you possibly can. Defenses and starting QBs are still available late in drafts and should see the full complement of snaps. So while you’re buying snaps, you’re also protecting against disaster. If you only roster two QBs, an injury to one will probably eliminate you from contention.
Just looking at our league last season, Bloom was loaded with league-winners, but his team faded after the Andrew Luck injury. I put up a zero when Sam Bradford’s injured stretch overlapped with Eli Manning’s bye. Meanwhile, Carter hammered us with the trio of Cam Newton, Robert Griffin III, and Blake Bortles. The key may have been Newton, but the selection of Bortles in Round 17 allowed him to overcome Griffin’s benching and nearly win the league.
Of course, this is just one league that can set us up for all the pitfalls that accompany anecdotes, but investigating individual leagues can often lead to good questions if not concrete answers. Working through last year’s results drove my interest in re-examining the QB question.
In light of this information, these were my final 10 picks this year.
|11||(3 – RB44) McKinnon, J|
|12||(10 – RB48) Dixon, K|
|13||(3 – RB51) Booker, D|
|14||(10 – QB16) Dalton, A|
|15||(3 – QB18) Tannehill, R|
|16||(10 – Def5) Panthers, C|
|17||(3 – Def8) Patriots, N|
|18||(10 – RB63) Ervin, T|
|19||(3 – Def21) Ravens, B|
|20||(10 – QB27) Bridgewater, T|
Mike Clay and I were the only participants to draft three QBs and three defenses. This obviously leaves me a little thinner at RB and WR. Opportunity cost is very real in best ball. We have no waivers to bail us out here. And circling back to the beginning, the players you pick will matter.
For a more detailed discussion of RBs and WRs along with our first foray into TEs, stay tuned . . .
- But it is a casual way to appease the appeal to authority crowd. (back)
- He outscored all of C.D. Carter’s running backs by at least 10 points. (back)
- There were 22 roster spots and optimal strategy was to pick 3 kickers. (back)
- We’ll have more MFL10 analysis from RotoDoc in the near future. (back)
- in terms of win rate (back)