Draft Strategy

MFL10 Roster Construction: What Worked in 2016?

In January, I took at look at the players with the best and worst win rates across 2016 MFL10 best-ball leagues. Here, the focus shifts to roster construction. Roster construction is defined by the number of players a team drafts at each position.

With 20 picks and five positions to choose from, there are many ways to build your roster – over 300 unique “constructions” were used at least once last year. By looking at how various approaches fared, we can arrive at some general guidelines for how we might build our rosters in the future. The following will break down 2016 results by position before pulling it all together to look at the best-performing rosters overall. A theme throughout this article is that there is not one “right” answer to the roster construction question, but there may be some wrong answers.

THE ONESIES

In MFL10s, the starting lineup requires only one quarterback, one tight end, and one defense.  The results of this analysis led to similar conclusions regarding the most successful construction at each of these “onesie” positions.

Quarterback

The drafting public appears to have a good handle on quarterback – 95 percent of teams rostered either two or three QBs last year. As this table shows, the win rates for teams with two QBs (9 percent) and three QBs (8 percent) are much greater than those of every other option. I do not read too much into the fact that teams with two QBs came out ahead of three-QB teams last year, particularly given that the win rates were virtually identical in 2015. Whether you go with two or three this year should depend on how comfortable you are with your strength at the position. If you take a QB in the early rounds, you are likely better off taking just two, while a three-QB approach is better suited for a late-round-QB strategy. The big takeaway here is staying within the two-to-three range.

QBs Drafted# of Drafts% of DraftsWin Rate
010.0%0.0%
11840.3%2.2%
22981150.4%9.0%
32689445.5%8.0%
419493.3%5.0%
52360.4%2.5%
6320.1%0.0%
740.0%0.0%

Tight End

As with QB, the range of winning strategies at TE was fairly narrow last year, and the majority of drafters took one of the dominant approaches. The win rates for two- and three-TE teams were virtually identical. For example, last year, RotoDoc’s Two TE1 strategy worked pretty well. But three-TE teams helped you capture more upside at a volatile position. Four-TE teams fared better than four-QB rosters, which makes some sense given that TEs can be used in the flex position, but there was still a drop-off in win rate from the two- and three-TE approaches.  Again, the conclusion here is two or three was the way to go.

TEs Drafted# of Drafts% of DraftsWin Rate
060.0%0.0%
11170.2%4.3%
22679745.3%8.3%
33115352.7%8.4%
410091.7%7.3%
5290.0%3.4%

Defense

Surprise! For defense, the answer again appears to be that we should stay within the two-to-three pick range. On a small sample, it looks like four-defense teams fared just as well as those with two or three in 2016, but I chalk that up to variance, knowing that win rates for four-defense teams in the preceding three years – again in small samples – were significantly lower (6 percent in 2015, 4.3 percent in 2014, and zero percent in 2013 with 2013 having a different format).

DEFs Drafted# of Drafts% of DraftsWin Rate
0210.0%0.0%
132695.5%4.4%
23695762.5%8.7%
31860231.5%8.3%
42390.4%8.4%
5220.0%4.5%
610.0%0.0%

THE FUN PARTS

While there isn’t much room for creativity at QB, TE, and defense, optimal running back and wide receiver construction is less strictly defined. I attribute the increase in viable strategies to a combination of a greater number of starters – two RBs and three WRs, plus the flex – and greater variance at the positions in terms of performance.

Running Backs

More than 75 percent of MFL10 drafters went with either five or six RBs in 2016, and both approaches generated above average win rates at 8.6 percent and 8.9 percent respectively.  Five or six was not the only way to go, though. While each combined for below average win rates, rosters with four, seven, and even eight RBs won close to seven percent of their leagues, showing that they were viable, if not optimal, approaches as the Monte Carlo simulations showed.

Then there are the three-RB teams, which, as a group, generated the highest win rate by a significant margin at 11.6 percent.  Last year I wrote an article promoting a “hyper-fragile” three-RB approach – a 3QB/3RB/9WR/2TE/3DEF construction which 90 of the 155 three-RB teams employed with relative success. Of course, 155 teams is a small sample, and I would not conclude from this data that three RBs is the best way to go in the future, especially before we have a good sense of average draft position.

In summary, five to six running backs was the safest way to go, but I would encourage you to experiment at RB based on your roster’s relative strength at the position.

RBs Drafted# of Drafts% of DraftsWin Rate
220.0%0.0%
31550.3%11.6%
435226.0%6.9%
52158636.5%8.6%
62440741.3%8.9%
7799713.5%6.8%
812732.2%6.7%
91560.3%3.2%
10110.0%0.0%
1120.0%0.0%

Wide Receivers

Seven WRs was the most popular construction in 2016, followed by six and then eight. Seven- and eight-WR teams led the pack in win rate with both at roughly nine percent. While seven and eight were the only approaches to finish above average in win rate, six and nine fared respectably, with both over seven percent. Staying between 6-9 wide receivers has been the way to go over the past several years, and I expect that to continue going forward.

WRs Drafted# of Drafts% of DraftsWin Rate
4700.1%4.3%
514692.5%4.1%
61531525.9%7.3%
72939149.7%8.9%
81090418.4%9.0%
916862.9%7.5%
102250.4%3.1%
11360.1%2.8%
12120.0%0.0%

PULLING IT TOGETHER

The data have pointed to multiple viable options at each position, but not every combination of those options is necessarily a good idea or even possible in some cases. Remember, we only have 20 spots to fill, so going with the highest 2016 win rate at each position (i.e. 2QB, 6RB, 8WR, 3TE, 2DEF) is not even an option.

To get a better feel for how to approach the choices between two and three at QB, TE, and defense, I took a look at the win rates for combined number of WRs and RBs over the past few years.  I found that allocating 12-13 picks between the two positions has worked best over time. That also means 7-8 roster spots across QB, TE and DEF has been the way to go.  So, while in a vacuum I would prefer to go with three late picks at each position over two earlier picks, this tells me I am probably better off going with only two at one of QB, TE and DEF, rather than limiting myself to only 11 WRs and RBs.

WRplusRBwinrates

The idea of limiting yourself to taking three at no more than two of the “onesie” positions is supported when we take a look at the 10 most successful roster constructions from 2016.1 None of the top 10 allocated nine spots to QB/TE/DEF, and nine of the top 10 went with either seven or eight.

Roster TagQBRBWRTEDefTotalDraftsWinsWinRateQB+TE+DEF
3/3/9/2/33392320901011.1%8
3/6/7/2/23672220445648911.0%7
2/4/8/3/324833204434710.6%8
2/5/8/3/22583220216422610.4%7
2/5/7/3/3257332046054559.9%8
2/6/8/2/2268222033753359.9%6
2/5/7/4/22574220163169.8%8
2/4/9/3/22493220114119.6%7
2/6/7/3/2267322054565139.4%7
2/6/6/3/3266332013641259.2%8

CONCLUSION

There is not one “optimal” roster construction for a 20-round best-ball draft. The number of players you take at each position should be a function of when you take the players (e.g. if you start the draft with three straight WR picks, then lean to the lower side of the 6-9 range). Hopefully these stats have provided you with some useful guidelines as you go into your drafts.

Subscribe for a constant stream of league-beating articles available only with a Premium Pass.

  1. Out of configurations used at least 50 times  (back)
By Mike Beers | @beerswater | Archive

Comments   Add comment

  1. Thanks for the article.
    That is great information.

Discuss this article on the RotoViz Forums