Optimal Best Ball Roster Construction on Underdog (Fantasy Football)
One of the biggest edges we have in best ball is simply constructing a roster that isn’t dead the moment the draft ends with egregious roster construction mistakes. I wrote all about the mistakes I was seeing in best ball drafts last season, and a lot of those drafts had to do with roster construction errors – drafting too many QBs or too few WRs, for example. There are of course many nuances to building the best possible roster in best ball formats, but simply avoiding roster construction mistakes will help us improve our win rates over our opponents.
So, exactly how many QBs, RBs, WRs, and TEs should we be drafting on Underdog? Let’s dive into each position in detail and take a look at some data* over the last two seasons.
Editor’s Note: For Best Ball Rankings and a team-by-team breakdown including pace metrics, stackability, and our highest exposure players, check out the Best Ball Primer included in the Ultimate Draft Kit+.
Recommendation: Draft 2 or 3
On Underdog (and on most platforms), we’re only required to use one QB score each week, so we certainly don’t want to be going overboard on this onesie position, especially when you consider that a lot of the QBs in the 20-32 range (when you’re presumably loading up on these QBs) don’t have true difference-making potential. Last season in Best Ball Mania II, teams that took 5 QBs advanced to the playoffs 8.4% of the time, by far the lowest of any roster build in the tournament. By drafting 5 QBs, you’re sacrificing valuable spots on RBs and WRs, which are crucial to winning in best ball.
Similarly, we can’t just draft 1 QB, even if that QB is a stud like Josh Allen or Patrick Mahomes. While a great QB will count for your score most weeks, we still want access to at least one other QB on our roster in order to access spike weeks and team stacks, which we know are crucial to best ball win rates and advance rates in tournaments. Like 5QB builds, 1QB builds did not perform well last year. These teams advanced at an 11.4% rate, 2nd worst among all roster builds.
Drafting 2 or 3 QBs resulted in the two highest advance rates not only in 2021, but also in 2020’s Best Ball Mania. Interestingly enough, not only did 2 or 3 QB builds result in the highest rates of top two finishes in 12 team leagues, but these builds also resulted in the highest rate of teams finishing in the top 10% in last year’s tourney. The field is definitely on this build, as these two roster constructions were by far the two most heavily utilized builds in last years tourney. However, given how conclusive the data is that this is indeed the optimal way to build, this is not a spot to look for leverage in the tournament.
Conclusion: If you’re not leaving your draft with 2 or 3 QBs, you’re likely lighting your money on fire. The data suggests drafting 2QBs is slightly more optimal than drafting 3 QBs.
|# Of QBs||Advance Rate||Top 10 Percent|
When To Draft a QB
So now that we know drafting 2 or 3 QBs is ideal, when should we take said QBs? If using last season’s data, there’s no question about it – we need to be taking at least one QB before Round 10. Teams that took their first QB in Round 11 or later, really did not do well in last year’s tournament. The advance rate for this group was just under 12% compared to an advance rate of 17.4% for teams that took their first QB at some point before Round 10. The take-home here is that punting the position, while sometimes viable in redraft leagues, is not advised in best ball.
|First QB Taken After Round 10?||Advance Rate||Top 10 Percent|
As seen in this graphic below, which is a HeatMap from Rotoviz’s Underdog Advance Rate Explorer, it appears to be most optimal to select your QB1 in the middle rounds, which is conclusive with the data from Underdog. In addition, you’ll notice that the advance rate for teams who took QBs very early in Rounds 3 and 4 really weren’t that advantageous. We want to be careful in analyzing data from only one or two seasons, as that’s a very small sample size. However, the opportunity cost in drafting Josh Allen, Patrick Mahomes or Justin Herbert at or ahead of their ADP is massive relative to the “tier 2 QBs” who are going in the Round 6-9 range. This does not suggest that we should never draft one of the top few QBs in ADP, but this data* at least suggests that we should probably consider the type of players we’re passing on to draft a QB in Round 3 or 4.
Recommendation: Draft 4-7
Before we dive into the specific details on how the numbers looked last season, I want to take a second to apply one basic rule of best ball – and that’s to assume you are correct when drafting. That mantra is important to understanding how to approach the RB and WR positions. If you took a couple of RBs early in the first few rounds, it makes sense to assume those RBs will not only stay healthy but perform well. In this example, we probably want to lean more into those RB scores and not waste valuable roster spots on several backup RBs. On the other hand, if you load up on WRs early and avoid the early RBs, it probably makes more sense to try to add more RBs to your roster in the middle and later rounds of the draft in an effort to build with “quantity over quality.”
In general, we either want to take one of 2 approaches:
- Take fewer RBs but take them early OR
- Take a lot of RBs but take them later in the draft (after Round 5)
Trying to do both – taking multiple RBs early AND drafting a lot of them intuitively doesn’t make sense. Alright, now that that’s out of the way, let’s dig into the data.
The data definitely supports this philosophy. Last season, the highest advance rate (19.4%) was a 7RB build where zero backs were taken in the first 5 rounds. For the Zero RB bros out there, this is music to their ears. On the other end of the spectrum (and this further supports the concepts above), the 2nd lowest advance rate (9.1%) were teams that took 4 total RBs but didn’t take any early. The take-home here is that it’s okay to pass on RB early, but if you do this, you probably need 5 RBs at a minimum but up to 7 is acceptable.
On the other side of the coin, if drafting RBs early, taking the High T approach and loading up on 3+ RBs early seems counterproductive, likely because you’re passing on the elite WRs available in Rounds 3-5, which we know historically have massive upside. Last season, no teams that took 3+ RBs in the first 5 rounds had an advance rate above 15%, and teams that took 4 RBs with their first 5 picks had the worst advance rates of any build. As you can see in the table below, drafting 4-7 appears optimal, but adding context around whether or not you’re taking your RBs early vs. late makes a difference in how many we should be taking.
Recommendation: Draft 6-10
Before we dive into the data for WRs, it’s important to understand starting lineup requirements for each site. On Underdog, we’re required to use 3 WR scores every week, and ideally, we want to be using a WR score in the flex given that WR variance is associated with higher win rates. So, in an ideal scenario, we want to be using 4 WR scores every week if possible. This starting lineup requirement alone tells us that we need several WRs on our best ball rosters. Not surprisingly, the data agrees.
Last year, not a single team that advanced to the finals in Best Ball Mania II drafted fewer than 6 WRs. More specifically, teams that drafted 3 WRs advanced just 4% of the time. Teams that drafted 4 WRs advanced to the playoffs 10.9% of the time. On the contrary, teams that drafted 6, 7, 8 or 9 WRs all had an advance rate north of 16.5%.
So now that we know we need quantity at the WR position, what about quality? Or in other words, how early should we be taking WRs, or is it okay to pass on the WR position and just load up on 10 WRs in the double-digit rounds?
I looked at teams who took at least two WRs in the first five rounds, then sorted the data to include teams that drafted at least 5 WRs. The results can be seen in this table below.
As you can see, the advance rates not only stayed strong, but they improved pretty substantially when teams drafted at least 2 WRs with their first five picks. Again, advance rates are best when teams drafted 6-10 WRs, and it appears that the data suggests if you’re investing in WRs early in the draft, taking 11 is probably overkill. Intuitively, this makes sense if we go back to our RB conversation above. If you’re drafting as though you are correct in the early rounds, you’re most likely using your stud WR scores the majority of the time, making your WR11 irrelevant most weeks.
To keep this trend going, I next sorted the data* by teams that took at least 3 WRs before Round 5, and again included only teams that took 5+ WRs.
My biggest takeaway from this data is that if you even further lean into the WR position early in your best ball draft by taking 3 in the first 5 rounds, taking fewer WRs appears to be optimal. The advance rate for 10WR builds in this scenario actually got worse when further investing early draft capital at the WR position, again further pointing to our common theme – “Draft as if you are correct.” Interestingly enough, teams that went WR early and stopped at either 6 or 7 did very well last season. If we combine this with the above data that showed zero RB teams did well last year, these findings are obviously correlated.
Similar to the RB position, it appears as though we should be correlating the number of WRs we’re drafting with how early we take them. As we found with the RB data, if you’re taking RBs early, it does not make sense to also take a lot of RBs. The WR data is also suggesting that this is bad for our win rate. The more you lean into WRs in the first five rounds, the fewer of them you need, but we know we need at least six wideouts regardless of our roster build.
Recommendation: Draft 2 or 3
Just like with the QB position, we don’t want to over-invest in this onesie position. We’re only required to start one TE per week. Of course, we can use a 2nd TE score in the flex, but let’s be honest – how often is a TE2 outscoring our WR3 or our RB3? Not very often, obviously, and the data supports that.
We can start by debunking the 1TE builds quickly. Those rosters had the second-lowest advance rate among the position, even if you drafted a stud like Travis Kelce last season. When looking at the ‘robust’ TE strategy of taking 5 of them, again the advance rates weren’t good. 5TE builds last year had the lowest advance rate at the position.
So, now the question becomes do we take 2 or 3? Or heck, even 4?
|# of TEs||Advance Rate|
Interestingly enough 3 TE builds had the best advance rate last season, but 4TE builds did appear to have some success. I’m hesitant to buy into that data, however, as sacrificing an RB5 or 6 or a WR7 or 8 for a TE4 seems sub-optimal given the data above. Remember, this is only one year’s data so we want to be hesitant not to take this information as fact. I’ll be interested to see how 4TE builds do in this year’s tournament when we analyze the data next season to see if this is sticky year over year.
One question a lot of people want to know is whether or not taking 2 stud TEs early is a viable strategy. I mean after all, if you’re using one of Travis Kelce or Mark Andrews‘ scores every week, shouldn’t you in theory have a massive leg up on the competition? I sorted the data to include teams that drafted exactly 2 TEs but took both before Round 5.
Those teams advanced to the playoffs just 10.2% of the time, well below our advance rate of 18% for teams that drafted 3TEs last year or teams that drafted 2TEs but didn’t take 2 TEs early (table above). In fact, not a single Best Ball Mania II team who made it to the finals last year employed this strategy. It seems as though it’s better off to let our opponents try to get sneaky with this build, knowing it’s not an optimal strategy.
In my articles that include a lot of data, I like to include a short section with the pertinent information that you need to know after reading this article. Here are my takeaways for each position:
- Draft 2 or 3
- Middle-round QBs seem to be the sweet spot for advance rates as the opportunity cost isn’t massive the way it is when you select a QB in the early rounds. At the same time, these middle-round QBs still have plenty of upside that we’re looking for in drafts and stacks.
- Draft your first QB before Round 10. Advance rates plummet when we try to go bargain shopping in the double-digit rounds. It’s not that it’s bad to take a QB2 in this range, but our QB1 should be coming from either the elite tier or the second tier.
- Draft 4-7
- The exact number of RBs to take largely depends on when you take your backs. If passing on the position early, it makes sense to draft more RBs. On the contrary, if you take RB early, it makes sense to draft fewer RBs because we’ll be relying on the scores of our Round 1 and Round 2 RBs most weeks.
- Draft 6-10
- Just like the RB position, deciding how many to select depends on when you take your wideouts. If spending early-round draft capital on multiple WRs, the data suggests 6 or 7 WRs is viable. On the other hand, if you take RB early, you’ll want to build your WR room with a build that likely has 8+ WRs.
- Despite the notion that we can just make up for a lack of quality with quantity at WR, the data suggests we should be leaving the first few rounds with at least one WR. Robust RB builds struggled, likely because those teams lack an elite WR that actually has meaningful upside that the double-digit round WRs simply don’t have. Read this article to understand that in more detail.
- Draft 2 or 3
- Consider the possibility of 4 TE builds based on last season’s data, but only do this if you’re passing on the elite options.
- Double elite TE seems like a fun strategy, but the advance rates are terrible when selecting 2 TEs before Round 5.
- In most scenarios, if you draft one TE early, you likely only need 2 TEs. On the other hand, if you pass on TE in the first 5 rounds, 3TE builds seem to be optimal.
*Data in this article obtained using Rotoviz’s Underdog Advance Rate Explorer Tool and Underdog Fantasy