Editor’s Note: This MFL10 Optimal Draft Strategies guest post was written by Sean Slavin. Follow him on Twitter @Slavin22.
If you’re reading this, you probably know what MFL10s are. You may be a casual player who drafts one league at a time, or you may be a full-on degenerate who is already worrying about diversifying their portfolio. If not, I recommend reading these…
MFL10s are wonderful because they allow us crazed fantasy footballers to fill the offseason with drafts, in a way that won’t pile up in-season work for us come September. It cured my habit of joining too many dynasty startups, simply because I wanted to draft.
Another great thing about MFL10s is that they are public, run all offseason long, and all have real money at stake. This creates a yearly dataset that is massive and untainted by home league biases or mock draft trolls (the worst kind of people in fantasy football, IMO).
These resources inspired me to do my own deep-diving research on MFL10s. The result is kind of a hybrid of their work, as I look at roster construction & ADP data in order to build an optimal draft strategy for MFL10s. My study is essentially VBD (value-based drafting) under a microscope. I would like a catchier name, but Micro-VBD will have to do for now.
Mike’s and Josh’s work on these subjects is stellar, but I wanted to look at it from a different angle. Knowing which roster constructions (3 QB, 5 RB, 7 WR, 3 TE, 2 DF for example) were successful in 2016 helps us a lot. However, you can end up with that 3-5-7-3-2 roster with utilizing completely different strategies. You can go Zero-RB in one draft, have a more balanced approach in another, and still finish with the same amount of players at each position. But those teams will look very different when you look at them through a positional-balance lens.
I began by collecting data on early MFL10 drafts. I ended up with 104 completed drafts, after filtering out tainted ones in which the wrong David Johnson was taken with the 1st pick or a similar blunder went down. These drafts ran anywhere from MFL10s Opening Day (2/27) to Wednesday, March 15. Considering the effect Free Agency and the Combine/Draft have on ADP, trends are already changing at the player level. See Rex Burkhead’s ADP Drift on Josh’s app:
My approach to evaluating ADP/roster construction is (mostly) immune to this. This is because I am looking at the strength of rosters based on draft strategy, rather than evaluating the players themselves.
Considering these drafts consisted of 1,248 rosters and just under 25,000 player selections, the dataset is significant despite only representing MFL10 drafts starting in the first week.
Measuring Strength of MFL10 Rosters
My initial goal was to find an optimal draft strategy that took advantage of how the field is behaving and found values accordingly. In order to do so, I needed a way to measure the strength of MFL10 rosters. I decided to use current ADP data to approximate this.
First, I calculated ADP for all players that were taken in any of these 104 drafts, accounting for when players went undrafted as well.
Normal ADP essentially gives a player one point if they’re taken firsts overall, 20 points if taken at 2.08, and so on. The average is taken to give ADP.
I tweaked this to get ADP*. This gives the player taken last one point and adds a point for every spot ahead of that in the draft. So a player taken in the middle of the draft gets ~120 points, and the first overall player gets 240.
Calculating ADP* this way gives two advantages over standard ADP:
- It gives a better player more points. This allows me to sum their scores, in a way that’s more intuitive
- Undrafted players get zero points, which drags their ADP* down. Standard ADP only accounts for drafts in which a player was taken, which is flawed IMO.
So while David Johnson and Devonta Freeman’s respective ADP’s are 1.6 and 13.1, their ADP* comes out to 239.4 and 228.9, respectively. This allowed me to sum up a team’s ADP*’s for all 20 players to measure team strength – I’ll refer to this as TS.
Since it’s based on ADP (not actual points data), we are essentially measuring how much value a roster has gotten out of a draft. I can go into a deep tangent on why I think ADP is the way to go here (rather than 2016 stats, 2017 projections, or expert rankings). But to save time, just google “Wisdom of Crowds” and you’ll understand my thought process. We run into a small roadblock of ‘groupthink’ within the fantasy community. Despite that, I believe this is the best (publicly available) approximator of value at this time.
Adjusting for Roster Balance
In a standard redraft context, TS would be a great metric to evaluate rosters immediately after the draft. However, in a draft-only league like MFL10s, we need to adjust for roster balance. Positional balance is not necessary in a standard league, where you can load up on wide receivers the whole draft and trade them later on to fix your holes. In MFL10s on the other hand, balance is crucial since trading/playing the waiver wire is not an option. Having the top wide receiver corps is still very valuable. But if you overload there, you will stretch yourself too thin in other areas without an ability to fix this.
So I adjusted TS in two ways to account for positional balance. I gave a penalty to each roster that only drafted the minimum starting requirement (or less) at any position. Also, I put a cap on how much value a team could accrue at each position. For instance, if a team already had Tom Brady & Cam Newton, they were right at the limit for QB value. So if that team drafted Derek Carr, they would not add any value to their team. Sure, they get the small benefit of weakening other teams’ quarterback corps, but that roster would benefit much more from a Kenny Britt or Jeremy Hill selection. After accounting for that, I ended up with ATS (Adjusted Team Strength). This metric allowed me to readily compare each of the 1,248 rosters in the sample.
Analyzing Draft Strategy
I then started digging into the data, focusing on how draft strategy/roster construction affected ATS. I looked at things like 2 QBs vs. 3 QBs, WR/WR/TE start vs. RB/RB/WR, as well as looking at how first-place rosters looked compared to 12th-place teams. It became pretty clear that draft slot affected strategy heavily. This is probably obvious, considering the uber-RBs are going heavily at the top of drafts, followed by a run of wide receivers.
So I split the teams into four groups based on draft slot, since teams drafting in the top-three should have a different strategy going in than those selecting in the bottom-three. It follows that a single optimal draft strategy is not sufficient, so I changed my goal to find one for teams drafting in each quarter of the 1st round. (I’ll refer to 1.01-1.03 as Early, 1.04-1.06 as Mid/Early, 1.07-1.09 as Mid/Late, and 1.10-1.12 as Late).
Overall Roster Analysis
Here are ATS averages based on number of players drafted at each position, for each of the four draft tiers:
- 2 QBs is the top option all around, but 3 QBs is not far off
- 2-3 TEs is optimal, with three having an edge across the board
- 2-3 DEFs is optimal, to varying degrees
- The optimal number of RBs is varied, settling in the 4-6 range
- The middle draft slots favor six RBs, as opposed to four for Early and five for Late
- This represents the fact that RBs are being taken in the first round more often in those spots, so less depth is needed at the position
- This should be obvious, but it’s encouraging that it’s backed up by data
- 7-9 WRs seems to be optimal
- The exact number will depend on how WR-heavy you draft early on
Focusing our Analysis
The tables above give a solid, yet flexible, plan of how to set up your roster construction based on your draft slot. However, we need to drill down our strategy further. 72 percent of rosters in my sample are in the optimal ranges at every position. So we will need to work harder to find an edge.
Besides the number of players at each position, the other aspect of roster construction is when to target those positions throughout the draft. So I investigated how teams fared in my ATS metric, based on when they drafted players at each position/depth.
The data below shows how strong rosters ended up, based on when they drafted their first RB:
The next set of tables is similar, but with a slightly different focus. They show the strength of teams, based on what position/depth they selected in the third round:
Based on both of the concepts above, I was able to develop an Optimal Draft Strategy for each of the drafts slots. I studied the tables for every round, as well as every combination of position/depth. This process was both an art and a science.
If I simply selected the top option for each particular round, it was possible that it would paint me into a corner later on. Also, the top option for RB5 could be round 12. But when I build the rest of my game plan, I might only have three running backs heading into that round.
So I used a blend of those concepts. I also looked ahead to avoid traps. The results are as follow, a specific game-plan to follow for each draft tier:
- If you are not familiar with Mike Beers’ Hyper-Fragile strategy, you should change that. The optimal strategy for the Early slot is very close to that
- His strategy calls for you to stop drafting running backs altogether after you collect your elite trio. But the data here supports taking a handcuff or third-down back later on. This provides a bit of safety for a high risk/reward plan
- Rounds 11-13 is the range to grab your quarterbacks
- The Early strategy suggests taking your QB1 in the seventh round. But if the elite options are gone, I would swap that with TE1 in the ninth
- Also, rounds 9-10 also provide decent value for quarterbacks, so it’s okay to take one there if the draft dictates it
- I wouldn’t wait any longer than the 14th to grab your QB2, the value drops off quickly from Alex Smith/Sam Bradford to Jared Goff/Mike Glennon, then a bunch of quarterbacks with no guarantee of starts
- TE1 shows up on round 8/Early 9 in every strategy, but rounds 5/6/11 were always close
- So if one of Jordan Reed/Travis Kelce/Greg Olsen falls, it’s fine to pounce and adjust your next few picks
- If Zach Ertz/Delanie Walker/Kyle Rudolph/Hunter Henry are all gone by your eighth pick, wait until the 11th and grab one of Eric Ebron/Cameron Brate/Martellus Bennett
- The optimal TE2s are all pretty late (15th round or later)
- If you feel like a high-end TE2 is too good to pass up earlier than that, go ahead
- If you do end up with a solid pair of tight ends early, you can nix that TE3 from the end of the draft, and reinforce another position
- If your strategy is to wait until the last two rounds to draft your defenses, that’s fine
- Bump those down, and grab your back-end skill players a bit earlier
- The position is largely a crap-shoot, and you can find value anywhere after round 16
- Drafts often can take a turn from the normal trends. That can force your hand if you’re too strict with any strategy
- So don’t hesitate to swap a couple picks around to adapt to your draft, especially in the later rounds
- If you are drafting near the turn at either end, this point is even more relevant – if you prefer T.Y. Hilton/Dez Bryant to the running backs available at 1.10, grab the wide receiver and take LeSean McCoy/Jordan Howard at 2.03
- If you have to stray from the strategy a little further, it’s not a huge deal
- For instance, you’re drafting from the late slot and go WR/WR at the turn. Just make sure you get to four wide receivers and two running backs after six rounds
- If your first few rounds look very different from the optimal strategy for your slot, look at the other strategies and see which one you match up with best
- For example, if one of the elite running backs falls to 1.05, I would suggest transitioning to the Late strategy unless you think you can pull off a Hyper-Fragile approach
Putting the Plan into Action
After developing the above game plans, I wanted to test them. So I started with the Early draft slot, and did a makeshift mock draft from 1.02 (right in the middle of the slot).
Then for each of my picks (1.02, 2.11, …), I filtered by the position my strategy prescribes for that round and picked the best player that had fallen to that pick (or later) in at least 50 percent of drafts. (This gives us a realistic pick at every round, so I am not just building a super-team, or even a best-case scenario). I repeated this process for each of the 20 picks. Then, I did the same thing from the 1.05, 1.08, and 1.11 spots, with the appropriate strategies.
Note: These teams were picked purely on the rules above, with no personal preference involved. Here are the results:
Evaluating the Results
Comparing these rosters to all 1,248 teams in the sample, the results are favorable. Every team ended up in the top 25 percent in ATS. For the most part, these rosters ranked very highly in wide receiver strength, just above average at running back, and below average at the other positions.
The Early team was great at quarterback, and the Late team was at running back. Otherwise, the trends followed my summary above. So while my strategies takes into account balance, they still build very good rosters while focusing on running back and especially wide receiver.
I believe the above teams are very realistic because of the way I selected them. In fact, I believe these are closer to a floor than a ceiling. If other teams in your draft reach even a bit, your results would be even better.
However, I would understand if you’re skeptical of trusting the results of a strategy on a hypothetical. So here are the early returns of my first draft utilizing my optimal game plan:
- The average roster has an ATS (Adjusted Team Strength) of 1,213 after six rounds
- The average roster drafting from the Late slot has an ATS of 1,208 after six rounds
- The mock rosters above average an ATS of 1,223 after six rounds
- My actual team drafting from 1.12 has an ATS of 1,225 through six rounds
- This is based off of live ADP data, so it’s not taking advantage of recent changes in a player’s ADP
The average MFL10 drafter is already doing well in roster construction, in terms of numbers of players to draft at each position. So the edge to be found is in targeting these positions in the right rounds. Follow the general guideline of the strategies above, make adjustments based on what the draft dictates, and you will improve your chances of cashing in on these MFL10s.
Draft trends change throughout the offseason. I plan to update these strategies at least once a month to keep up. But overarching strategies shouldn’t change nearly as quickly as specific player trends will.
Finally, I’d like to remind you not to be too stubborn in sticking to strategies, whether they’re mine or your own.
And if you start a draft in a way that doesn’t match up well with any of these strategies, don’t panic. Just hit me up on twitter or via email. I will make tweaks to find a new optimal strategy to finish your draft. The same goes for if you feel strongly about drafting a certain position earlier/later than I prescribe.
Hopefully, this helps you win some money. But more importantly, I hope you learned a thing or two. Good luck in all your MFL10 drafts!
Latest posts by TwoQBs Guest (see all)
- Blaine Gabbert vs. Drew Stanton: The QB Battle of our Dreams - August 8, 2017
- Cam Newton is a Fantasy QB1 with Upside You Shouldn’t Forget About - July 21, 2017
- #SFB7 Cheat Sheet Web Tool Primer - July 7, 2017