Startable Quarterback Percentage (SQB%)
Greg Smith outlines his Startable Quarterback Percentage (SQB%) metric and explains how it applies it to late-round quarterback drafting in 2QB and Superflex fantasy leagues. …
As an outspoken proponent of late-round quarterback drafting — even in 2QB and Superflex leagues — I wasn’t surprised when our recent guest post at Rotoworld was met with a little push-back from the general two-quarterback fantasy crowd after they clicked through to my LRQB-centric rankings. Home leagues will admittedly feature drafters who overvalue the QB position, and the fear of missing out on quality passers is a natural reaction to the pressure applied by those more casual drafters. My 2QB home leagues are no different. I routinely pick quarterbacks earlier with my local friends than I do in analyst drafts. Still, I virtually never draft signal callers in the first two or three rounds, and my aversion to high-priced quarterbacks has served me well.
Implementing the late-round quarterback strategy correctly takes some amount of in-draft practice. Check out Sal’s team (@2QBFFB) in this recent mock as an example of the plan in action. The more you mock, the better your feel for the pace of QB drafting will become. Above all, though, accepting LRQB into your fantasy football heart requires a leap of faith. Initially, it may be difficult to avoid bleeding from the eyes in shame when you compare your lineup with Matt Ryan & Alex Smith to an opponent’s with Andrew Luck & Carson Palmer. Because of those instinctive reactions, helping others accept the logic and methodology behind this quarterback-agnostic approach is often a struggle. I wanted real data to support my position, so I amassed some historical fantasy stats and created what I’m calling Startable QB Percentage (SQB%).
Laying the Foundation
If you’re a regular visitor to TwoQBs, you’ve probably stumbled across one of our QB Cards, pages which compile weekly finish data and other stats from the past four years for currently-relevant fantasy quarterbacks. Startable QB Percentage takes that data and relates it to quarterback ADP, creating a map between player-agnostic draft slots and performance.
To make the QB Cards, I looked at every individual quarterback season since 2012 and determined how often those passers had top-10, top-20, and sub-24 weekly finishes among their peers. Let’s look at the data from the aforementioned Carson Palmer’s QB Card as an example:
Looking at his excellent 2015 season, we see Palmer finished as a top-10 quarterback in 7 of his 16 games (43.8%) and a top-20 quarterback in 14 of 16 games (87.5%) from an ADP of QB19. Let’s compare those values to Ryan Tannehill’s 2015 campaign:
Tanny was drafted seven spots higher (QB12), but only posted top-10 performances in 3 of 16 games (18.8%) and top-20 performances in 11 of 16 games (68.8%). I’ve cherry-picked these example players, but my goal in concocting SQB% was to consider all quarterbacks at their respective ADPs, whether they proved to be over- or undervalued.
Throwing Away the Names
The next step was taking each slot in the yearly quarterback ADP and combining those players’ QB Card stats into one cluster for that draft slot. For example, here’s a look at all the QB1s by ADP since 2012:
|Year||QB ADP||Player||GP||Top-10 #||Top-10 %||Top-20 #||Top-20 %||Bust #||Bust %|
Between 2012 and 2015, the first quarterback drafted in fantasy drafts could be expected to finish approximately 54.2% of his games in the top-10 and 81.3% in the top-20 at his position on a weekly basis. That certainly sounds pretty good for a 2QB league, but what about all the other quarterback draft slots? Let’s dive in and take a look…
Startable Quarterback Percentage by ADP
I started by looking at each ADP slot individually, but before we get into the results, a quick aside on weighting the results for reliability:
In addition to considering the raw percentages for top-10, top-20, and sub-24 QB finishes at each draft position, I adjusted those percentages to account for the number of games played. I multiplied the finish percentages by the fraction of possible games played. In the QB1 example above, Peyton Manning (2014) had the same raw percentages as adjusted percentages because he played the full 16 of 16 games. On the other hand, Andrew Luck (2015) saw his adjusted SQB% values fall because he only appeared in seven contests. Here are Luck’s top-10 numbers side by side:
Pure Top-10% = 0.542
Adj. Top-10% = 0.542*(7/16) = 0.237
Ultimately, I chose to discard the adjusted numbers because they didn’t tend to reshape the data in a meaningful way. Due to the somewhat random nature of injuries, it turns out most ADP slots are similarly subjected to the negative effects of these adjustments over the course of four years. There is an exception for quarterbacks with very late ADPs, as they inherently play fewer games. Adjusting for their lack of service time makes their bust rates (sub-QB24 finishes) easier to stomach because the expectations are lower or nonexistent. Nevertheless, I opted to ignore the adjusted values and focus on pure percentages for the purposes of this piece.
“Back to the lecture at hand,” said the good Doctor. After creating four-year averages for each average draft position, I plotted the three SQB% values together:
As you can see, the plots are all pretty noisy, but some trends do appear. On the whole, fantasy drafters are adequate at identifying who the best quarterbacks are. We are also generally able to tier the position well through QB25 to account for declining top-10/top-20 percentages and increasing bust percentages.
Still, when I imagine a trendline through each plot between QB4 and QB25, I don’t infer a steep value drop among quarterbacks. On average, QB25 has posted more top-20 weekly finishes than QB4, QB5, and QB6 over the past four years. I’m cherry-picking again, of course, and those higher-ADP quarterbacks justify their costs by busting less often and delivering more elite top-10 finishes, but these plots show how baseline-startable quarterbacks are readily available outside the top-15 at the position.
Buckets of SQB%
Rather than simply imagining those trendlines in the plots, I resolved to iron out the data to make the overall patterns across the chart easier to identify and digest. To achieve smoother plots, I tiered the quarterbacks into buckets, with each bucket amassing an equal number of ADP slots.
I originally divided the data into five ADP slots per bucket. Here are the results:
|ADP Range||Bucket||Top-10 %||Top-20 %||Bust %|
That’s more like it. Once again, we see a clear top tier, then a slow degradation in quarterback value until QB25. After the 25th quarterback is drafted, the position tends to become a crapshoot, with players posting top-20 finishes less than half the time and more bust weeks than top-10 weeks.
Remember these results are averaged out over many players and multiple years, though. Outliers will pop up from time to time. Take Tyrod Taylor’s 2015 season as an example. He was QB29 in 2015 ADP, but he never finished below QB24 in his 13 games, while finishing in the top-10 five times (38.5%) and the top-20 nine times (69.3%). That drastic overperformance is why he’s the TyGoat, but if he posted the exact same figures this year from an ADP of QB10, it might be considered a letdown relative to the baseline data above.
Once I got this far, I started to second-guess my method for tiering the quarterback ADP into equally-distributed buckets. Lumping QB1 through QB5 together doesn’t necessarily line up with 2016 drafting tiers, as there is a clear top-4 at the position before a drop-off to QB5 in draft cost. I had always planned on examining quarterback groupings according to the upcoming season’s ADP, but that way of thinking is slightly flawed because ADP tiers are known to shift between seasons and I was pulling data from the past four years.
I bounced my concerns off of the other TwoQBs co-founders, Joshua Lake and Sal Stefanile, to get their thoughts. They echoed interest in the idea of applying the historical data to this season’s ADP tiers, but also encouraged me to try one other equally-distributed bucket size. I felt I had zoomed back enough with 5-QB buckets, so my next effort used buckets containing three quarterbacks each:
|ADP Range||Bucket||Top-10 %||Top-20 %||Bust %|
This data doesn’t reveal much more than the 5-QB buckets, so let’s jump straight to the custom-sized buckets based on 2016 ADP gaps. This table includes the quarterbacks related to each bucket as they tend to be tiered in our 10-team redraft ADP (all offseason mocks included), but keep in mind the SQB% numbers are based on 2012-2015:
|ADP Range||Bucket||Top-10 %||Top-20 %||Bust %||Related QBs|
|QB 1-4||1||55.6%||82.5%||10.3%||Cam Newton
|QB 5-7||2||41.0%||69.7%||18.5%||Drew Brees
|QB 8-9||3||44.4%||80.6%||13.0%||Carson Palmer
|QB 10-12||4||37.1%||73.0%||14.0%||Philip Rivers
|QB 13-20||5||31.2%||65.9%||20.9%||Jameis Winston
|QB 21-24||6||32.7%||61.0%||23.4%||Ryan Tannehill
|QB 25-27||7||21.4%||60.0%||30.3%||Ryan Fitzpatrick*
|QB 28-35||8||20.8%||47.2%||35.4%||Carson Wentz
Robert Griffin III
|QB 36+||9||12.3%||31.9%||55.8%||All other QBs|
*Note: Most of our ADP data comes from before Ryan Fitzpatrick re-signed with the Jets, and his own ADP is deflated as a result.
What the 3-QB and custom-sized buckets have in common is a slight dip in value for the second tier of quarterbacks. We can’t blindly apply an overvalued tag to all the quarterbacks who happen to fall in this year’s QB5-QB7 ADP range, though. This season could certainly play out differently than the past four, especially considering the community’s likely overreaction to 2015’s LRQB takeover in quarterback valuations for 2016.
Nonetheless, it is interesting how passers in that second group haven’t fared especially well relative to the third and fourth tiers. The data can’t teach us which specific players to avoid, but it can teach us to avoid taking the bait on quarterback runs in the early rounds. In any given season, there are typically only three or four signal callers who are legitimately elite. If you miss on those guys and it spurs you to reach for QB5 or QB6, you might be spewing value. The charts above show how the players in the 7 to 12 range of quarterback ADP are largely safer investments at their costs.
If you take a late-round quarterback approach and pluck passers from the deeper buckets, you lose some upside at the position, but top-10 weekly finishes are still more common than bust weeks all the way down to QB24. Consider the counteracting upside you stand to gain at running back and wide receiver while you avoid quarterbacks with your early picks, and the LRQB strategy can easily pay for itself.
Anecdotes, Not Antidotes
Wrapping up, I want to reaffirm the compiled and generalized nature of these statistics. The average top-10 finish rate of 40.3% for QB11 between 2012 and 2015 doesn’t imply this season’s QB11, Eli Manning, is destined to deliver on that level. His QB Card shows that level of production is within his range of outcomes (see his 43.8 top-10% in 2014 and 2015). On the other hand, the card also shows Eli’s potential downside (6.3 top-10% in 2013). The truth is likely somewhere in the middle, perhaps around the 40.3 Top-10 SQB% for QB11, but Manning’s season could go sideways in a variety of ways.
Rather than try to apply Startable QB Percentage in a targeted way to individual quarterbacks, we should use it to assess the quarterback position as a whole. The results reveal reliability deep into the passer ranks, and they can inform our draft strategy for the position. Late-round quarterback drafting isn’t a cure-all. You still need to evaluate players properly and navigate your draft with a finely-tuned gauge for the pace of quarterback pick frequency.
Anecdotally, though, I vouch for the viability of waiting to draft passers, and SQB% backs me up. While I don’t have the data to create Startable Running Back Percentage or Startable Wide Receiver Percentage, I’ve experienced more noticeable value drop-offs in those positions through my many years of drafting 2QB and Superflex leagues. My love for LRQB isn’t simply based on these great disturbances in the drafting force, though. Studies on quarterback production from 1QB formats give credence to the strategy.
Furthermore, the notion that late-round quarterbacks are safer than late-round running backs and wide receivers is highly intuitive. We know quarterbacks — even the mediocre ones — are going to touch the ball a large number of times in every game they play. That simply isn’t the case with rushers and receivers available in the later rounds.
Compare the volume outlooks of Teddy Bridgewater (130.2 in 2QB ADP) to Tevin Coleman (123.2 in 2QB ADP) and Will Fuller (126.0 in 2QB ADP). Teddy is surely the safest bet among the three to produce consistently, and he’s drafted the latest on average. For those who would tout Coleman and Fuller as higher-upside players, remember it wasn’t long ago when Bridgewater was a high-profile prospect himself. If his upside relative to cost is on par with Coleman and Fuller, and his floor is stable merely because he plays quarterback, then Bridgewater is strictly more valuable a pick in the late rounds of a two-quarterback draft.
The arguments against the case of Teddy v. Coleman & Fuller are built right into the premise, unfortunately. First, how can we know if Teddy’s upside is on par with the defendants? We simply can’t, not for sure. Our goal is to narrow the possible range of outcomes for each player and hope to adjust correctly when we miss. Second, if all quarterbacks have inherently stable floors, why prioritize any single player like Bridgewater? This helps all passers, including those who are clearly better and more costly than Teddy. So why not draft quarterbacks earlier and lock in a better floor with higher upside? That’s the rub in 2QB and Superflex leagues. There will always be many different ways to win.
Late-round quarterback drafting works for me, but that shouldn’t stop you from trying to find fantasy value in other ways. When I look at the SBQ% plots above, I see an advantage to be gained by devaluing signal callers. Others will see an advantage to be gained by drafting the elite top-10 SQB% of top-tier quarterbacks. Ultimately, though, whether we’re drafting quarterbacks or any other positions, we need to be correct in some evaluations and pick the right players. Which brings us once again to Startable Quarterback Percentage’s greatest weakness: it tangles a great number of unrelated quarterback seasons together and provides no context for individual performance. That’s where the QB Cards come in. Having a microscopic view helps when you need to untangle a macroscopic mess. Dive into the data, extract your evaluations, and develop a strategy that works for you. Happy drafting.
Latest posts by Greg Smith (see all)
- Week 2 Rankings & Game Flowbotics - September 11, 2019
- Week 1 Rankings & Game Flowbotics Preview - September 3, 2019
- 2QB/Superflex Rankings Methodology and Cheat Sheets - August 23, 2019