DFS Playbook Strategy: How to Pick a Quarterback
Ever pulled up your Big Ben lineup before afternoon kickoff and wished that instead of lateswapping a player into your lineup, you could lateswap the money you're about to lose back into your account? Or taken one glimpse at your Mariota lineup and instantly wished that the contest had switched to pointperhandoff scoring? Or checked your Russell Wilson lineup and immediately shot a text to Ciara, berating her for not being the only person he's recently begun screwing. (Oh, I'm the only DFS player that has Ciara's number?)
The Daily Fantasy Playbook series will leverage my countless hours of researching and playing DFS to outline strategies you can use to quickly gain a competitive advantage at picking players at each position, starting with quarterback. I'll reveal the keys to making accurate projections, the tactics to minimize busts in cash games, and the strategies that firstplace guaranteed prize pool lineups have been using to win millions of dollars.
DFS Playbook 2017: RB  WR  TE  DEF  K  DFS Strategy Hub
3 Keys to Accurate QB Projections
While taking the time to master bankroll management, contest selection, game theory, etc., will undoubtedly improve your game, DFS ultimately comes down to making the best predictions. There are a lot of quarterback stats and metrics floating around, but making accurate projections comes down to separating the signal from the noise.
1. Weight Efficiency More Than Volume
In 2016, Aaron Rodgers and Drew Brees were both in the top five in pass attempts, so it's not surprising that they both finished as topthree fantasy QBs. But wait—Joe Flacco and Carson Wentz were also both in the top five in pass attempts. Breaking: Joe Flacco and Carson Wentz did not finish as topthree fantasy QBs. In fact, neither one of them managed to finish any higher than 20th. So what gives?
More so than any other position, a quarterback's fantasy production is driven by efficiency as opposed to volume:
Ingame Stat 
Correlation (DK Pts) 
Ingame Stat 
Correlation (FD Pts) 

TD% 
.68 
TD% 
.83 
Pass Yds/Att 
.57 
Pass Yds/Att 
.56 
Comp % 
.40 
Comp % 
.41 
INT% 
.39 
Yds/Comp 
.38 
Yds/Comp 
.39 
INT% 
.33 
Pass Att 
.31 
Pass Att 
.26 
^{Data from 201316 and does not include nonstarting QBs.}
There are two main reasons why efficiency takes precedence over volume for a fantasy QB:
 Passing yards and passing touchdowns are awarded fewer points than rushing/receiving yards and rushing/receiving touchdowns (and no points are awarded to the passer per completion), creating a situation where the average pass attempt is worth less than the average rushing attempt or receiving target.
 Quarterbacks don't have to fend off teammates for precious opportunities like players at the other positions do, which causes the deviation between high and lowvolume players at quarterback to be smaller than that of high and lowvolume players at the other positions.
Since it's tough for a quarterback to separate himself in fantasy by attempting more passes than his peers, he instead needs to do so by being efficient.
Inherently, being an efficient passer means attempting fewer passes than an inefficient passer: a higher completion percentage means fewer incomplete pass attempts; a higher yardsperattempt average means fewer pass attempts are necessary to gain a given amount of yardage; and a higher touchdown rate means fewer pass attempts are necessary to throw a touchdown.
2. Leverage the Predictiveness of Vegas Implied Team Total
Don't buy into the rationale that you should pick a quarterback because he's expected to rack up a lot of attempts as an underdog playing from behind. Same goes for the rationale that you should fade a quarterback because he's not expected to have to throw much as a favorite playing with a lead. Those ideas are flawed because efficiency trumps volume for fantasy QBs,
Those ideas are also flawed because fading quarterbacks playing on favorites means fading quarterbacks whose teams have the highest Vegas implied team totals.
Save for an actual fantasy projection, Vegas implied team total will be more strongly correlated to quarterback fantasy production in a given game than any other stat or metric available beforehand.
Passing now accounts for roughly twothirds of the average NFL team's offensive yardage and touchdowns. So, it should be no surprise that a projection of how many points a team will score is also useful for projecting the individual stats of its quarterback.
Because implied team total is a combination of spread and over/under, it is more predictive than spread or over/under individually:
Vegas Odds Type 
Correlation (DK Pts) 
Vegas Odds Type 
Correlation (FD Pts) 

Implied Points 
.29 
Implied Points 
.29 
Over/Under 
.22 
Spread 
.22 
Spread 
.21 
Over/Under 
.19 
^{Data from 201316 and does not include nonstarting QBs.}
Outside of being part of the impliedpoints calculation, looking at the size of the spread doesn't serve much use when projecting quarterback performance. The spread itself is still useful for differentiating between favorites and underdogs, however, which has implications in cash games and tournaments be discussed later on.
3. Use the Correct Stats and Sample Size When Analyzing Past Performance
We both know it's easy to get caught up looking at game logs, using recent performance to confirm narratives such as "he's been seeing the field well lately," or "his footwork has been messed up right now," or "Tim Tebow stinks." But how much does a quarterback's recent performance really tell us about his upcoming performance? And which stats should we be looking at, exactly?
To find out, I did a regression analysis on the relationship between a quarterback's fantasy points in a given game and his seasontodate averages in various stat categories. I did this for all possible inseason sample sizes, i.e., from 1 game to 14 games.
Two stat categories in particular had a stronger correlation to upcominggame QB fantasy points than the rest:
 Fantasy points per game: Most predictive with a sample size up to 6 games.
 Fantasy points per attempt: Most predictive with a sample size of 7+ games.
^{Data from 201416 and includes the top 30 fantasy QBs from each season.}
The data suggests we'll get diminishing returns if we're looking at a sample size beyond 8–9 games for a fantasy QB.
The fact that the optimal sample size for projecting QBs peaks runs in contrast to the other positions. For nonQBs, predictiveness continues to increase as sample size gets larger. This is something to keep in mind later on in the season (aka Live Event season!).
It's easy to get caught up in recency bias, but there's almost no correlation between quarterback performance from one game to the next. In fact, the most recent game is less predictive for a quarterback than any other position.
Click to become a 4for4 DFS Subscriber now!
3 Tactics for QB Cash Game Lineup Construction
Not too long ago, "pay up at QB" was the accepted maxim in DFS. More recently, the trend has been to bargain hunt; the position is now considered to be brimming with safe plays that can still net at least 10 points on a bad day. But either one of those approaches can be risky as a starting point.
Instead, lineup decisions should be made based on opportunity cost.
1. Aim to Get More Value from Quarterback Than Other Positions
Many DFS players make decisions based on linear value multipliers, i.e., 2 points per $1,000 (aka 2x) on FanDuel or 3 points per $1,000 (3x) on DraftKings. This can lead to suboptimal lineup decisions, however—especially at QB. Due to their linearity, those multipliers fail to take into account that QBs score more points per dollar on average than nonQB positions.
Without getting into why targeting a specific cash line is a slippery slope due to natural variance, let's go over exactly how aiming for a linear multiplier like 2x or 3x makes a lineup vulnerable.
Let's say you locked in Tom Savage at $6,300 on FanDuel in Week 16 last season. "He's playing the Bengals, so it's not a great matchup,” you may have thought, “but he should be able to approach 12.6 points and hit 2x value.”
I just hung the hypothetical you out to dry, because Savage only reached 1.29x value. But no matter how many points Savage scored, it would have been a mistake to target only 2x for him.
Nearly half of the QBs on that Saturday slate reached at least 2.7x. And nearly half of all QBs on FanDuel reach at least 2.4x value every week. By targeting 2x value—or any linear equivalent—with no regard for position, you would fail to account for each fantasy position's unique scoring expectation. You would also fail to account for salarybased expectation, i.e., the cheaper a player is, the higher his value multiplier needs to be. For example, 27 points at a $9,000 salary is more valuable than 12 points at a $4,000 salary even though both are technically 3x value.
To create a nonlinear value multiplier that targets the same 150 DraftKings points/120 FanDuel points as 3x/2x, I ran a series of regressions to find the relationship between salary, position, and fantasy points:
QB cash game target score for 150 DraftKings points (3x) = 16.5 + (salary * 0.00087)
QB cash game target score for 120 FanDuel points (2x) = 10.4 + (salary * 0.00118)
For example, take Week 1 of last season: At a FanDuel salary of $9,000 would indicate that Aaron Rodgers needed roughly 21 points—2.34x—to hit value; at $5,000, Dak Prescott needed 16.3 points to hit value, but that results in a much higher target multiplier of 3.26x.
DK Salary ($) 
DK Pt Cash Target 
DK Pts/$1,000 Value 
FD Salary ($) 
FD Pt Cash Target 
FD Pts/$1,000 Value 

9000 
24.3 
2.70 
9000 
21.0 
2.34 
8000 
23.5 
2.93 
8000 
19.8 
2.48 
7000 
22.6 
3.23 
7000 
18.7 
2.67 
6000 
21.7 
3.62 
6000 
17.5 
2.91 
5000 
20.9 
4.17 
5000 
16.3 
3.26 
Essentially, the minimum number of points you would target from a QB is 20.9 on DraftKings or 16.3 on FanDuel. Then for every $1,000 increase in salary, 0.87 points (DraftKings) or 1.18 points (FanDuel) is added.
Again, the key takeaways are:
 More value is required from QB than any other position.
 The lower a QB's salary, the higher a value multiplier he'll need to hit. (This is true for all positions.)
2. Allocate Salary According to Opportunity Cost
Identifying top plays is only part of the puzzle in the DFS. Any pick you make will only be as valuable as how strong it makes your lineup as a whole. Therefore, you need to take opportunity cost into account. For instance, is paying up for Tom Brady worth it if it means you can't fit David Johnson?
DraftKings Salary Allocation: Beware of Overspending
DraftKings seemingly decreased their QB salary ceiling starting in 2015. This resulted in roughly 8 out of every 10 QBs being bunched in the $5,000–$7,000 range each week. This can exploited by aiming to use the cheapest viable QB option on the slate. Paying up at QB should essentially be viewed as a luxury. Clear value plays at other positions need to be available to offset the opportunity cost of paying up at QB.
Historically, there's been no difference in consistency rate between a QB on DraftKings priced as high as $7,500 and one priced as low as $6,000. There has been a major difference, however, between the value of a topdollar RB or WR and the more inexpensive options. This suggests paying up at those positions should be prioritized over paying up at QB.
DK Salary ($)  QB Cons %  n (QB)  RB Cons %  n (RB)  WR Cons %  n (WR)  TE Cons %  n (TE) 

9000+  72.7%  11  44.7%  47  
8500–8900  16.7%  6  71.4%  7  52.5%  61  
8000–8400  50.0%  14  40.0%  15  23.8%  42  75.0%  4 
7500–7900  37.7%  61  46.8%  47  30.0%  110  28.6%  7 
7000–7400  31.8%  110  30.8%  65  32.3%  130  50.0%  8 
6500–6900  32.4%  136  33.3%  69  29.7%  155  31.6%  19 
6000–6400  33.6%  137  24.2%  99  33.7%  169  46.7%  15 
5500–5900  27.9%  165  24.6%  134  24.4%  180  40.0%  35 
5000–5400  18.2%  379  28.1%  139  22.6%  190  14.6%  48 
< 5000  25.7%  709  23.6%  911  28.2%  649 
^{Data does not include nonstarting QBs, RBs/WRs projected under 9.0 points, and TEs projected under 7.5 points by 4for4 in a given game.}
DraftKings salary is less predictive of fantasy points for quarterbacks than for any other fantasy skill position.^{1} As long as risky options (such as underdogs, which we'll get to next) are eliminated, building around the cheapest viable QB on DraftKings is a sound approach.
FanDuel: Inexpensive Options can be Deceiving
Since QBs tend to score significantly more points per dollar than nonQBs on FanDuel, the cheapest QBs on a given slate will tend to have the best pointperdollar projections.
But more often than not, those cheap QBs end up busting.
In fact, QBs priced between $6,000 and $6,900 on FanDuel have historically been more inconsistent than players at any other position or salary range:
FD Salary ($)  QB Cons %  n (QB)  RB Cons %  n (RB)  WR Cons %  n (WR)  TE Cons %  n (TE) 

9000+  46.7%  192  51.2%  123  40.4%  114  
8500–8900  33.5%  209  43.4%  145  45.1%  175  11.1%  9 
8000–8400  34.5%  258  36.3%  146  37.0%  211  58.5%  41 
7500–7900  32.9%  356  25.3%  182  36.4%  269  33.3%  36 
7000–7400  29.7%  306  34.1%  293  28.5%  389  30.6%  36 
6500–6900  22.1%  312  27.8%  399  26.6%  537  30.1%  28 
6000–6400  24.1%  220  31.2%  337  30.9%  560  31.8%  170 
5500–5900  34.5%  55  29.6%  240  31.6%  383  31.0%  248 
5000–5400  25.0%  68  22.7%  141  31.2%  237  32.0%  197 
4500–4900  39.7%  68  32.3%  96  32.6%  89 
^{Data does not include nonstarting QBs, RBs/WRs projected under 9.0 points, and TEs projected under 7.75 points by 4for4 in a given game.}
QBs priced in the $6,000s and low $5,000s have collectively failed hit value consistently enough to offset the amount of raw points more expensive options produce. There will undoubtedly be exceptions from time to time, but the best approach on FanDuel will generally be to not punt QB.
On the other hand, QBs priced $9,000+ have been extremely consistent. It won't always be necessary to pay $9,000+, however. since QBs become more consistent than RBs and WRs starting in the mid$7,000s. Due to positional scarcity, it usually makes more sense to pay for consistency at RB and WR first if necessary.
3. Leverage Vegas Odds to Maximize Consistency
You don't want to be one of those DFS players that thinks only in terms of points per dollar. Expanding your thought process to include probability—like we just did when reviewing positional consistency rates—can provide a major edge.
That's what cash games are all about: maximizing the probability that your lineup scores a sufficient amount of points to be profitable each week.
Since cash game target scores are adjusted for salarybased expectation, they are extremely useful for comparing the consistency rates of differently priced QBs with all the various possible Vegas odds.
HomeRoad and FavoriteUnderdog Splits
Classifying homeroad and favoriteunderdog splits in terms of their four possible combinations rather than as two separate splits helps to maximize predictiveness. While doing so is a small and perhaps obvious detail, it can dramatically increase your cash game lineup's floor.
The consistency splits indicate that home favorites are the optimal QB play by a large margin:
^{Data from 201316 and does not include nonstarting QBs.}
Using the cash game target scores is critical here because traditional pointperdollar splits would have failed to demonstrate the true value of a home favorite. For example, looking only at average points per $1,000 of favorite QBs (2.94) versus underdog QBs (2.82) would have indicated just a minimal fourpercent edge for the favorites—the cash game target scores reveal the edge is much larger.
The discrepancy between favorite and underdog QBs is so large that underdog QBs should arguably almost never even be considered in cash games.
Maybe that sounds extreme, but according to the consistency rates, only 1in3 players on FanDuel and 1in4 players on DraftKings will hit cash game value. Given those high bust rates, you need to be extremely aggressive in eliminating potential land mines. (A player hitting value on FanDuel has a higher probability than one hitting value on DraftKings because it's easier for a lineup to score 120 on FanDuel than 150 on DraftKings. As I mentioned earlier, targeting a specific cash line is a slippery slope.)
Vegas Implied Team Total
The largest jump in QB consistency rate occurs when that QB's Vegas implied team total crosses the 22point benchmark:
Data from 201316 and does not include nonstarting QBs.
Because the average consistency rate of QBs is 32.2 percent on FanDuel and 24.1 percent on DraftKings, an implied team total of 22 also marks the breakeven point where QBs begin to perform with aboveaverage consistency. An implied team total of 26+ is ideal, however—that's where all QBs start to match the consistency rate of the average home favorite.
Click to become a 4for4 DFS Subscriber now!
3 Strategies for Picking a Quarterback that Wins GPP Tournaments
Note: Data in this section was compiled from firstplace lineups in the 2015–16 DraftKings Millionaire Maker and FanDuel Sunday Million GPPs.
On average, the QB slot has been responsible for the thirdlargest scoring contribution to firstplace lineups in both the Millionaire Maker and Sunday Million over the past two seasons—and not far off from matching the top RB and WR:
Rank (DK)  Position (DK)  DraftKings Pts  Rank (FD)  Position (FD)  FanDuel Pts 

1  WR1  35.9  1  RB1  31.1 
2  RB1  35.5  2  WR1  29.9 
3  QB  32.0  3  QB  29.0 
4  WR2  29.7  4  WR2  24.7 
5  RB2  26.2  5  RB2  21.3 
6  TE  25.3  6  WR3  18.7 
7  WR3  23.1  7  TE  18.4 
8  FLEX  22.5  8  D/ST  18.4 
9  D  15.2  9  K  12.7 
^{WR1 determined by highestscoring WR in lineup, WR2 determined by secondhighest scoring WR in lineup, and so on (not determined by salary)}
Three main strategies can be employed to ensure you're picking a QB with tournamentwinning upside:
1. Identify Quarterbacks That Fit the HighUpside Profile
Because every team's quarterback will be good for 25+ pass attempts in most weeks, you could technically make (at least a poor) case for playing almost any QB in a tournament. This is especially true if you were going to stack the QB with his receivers (more on that coming up in a few moments).
But certain types of QB picks simply don't carry enough upside.
It's essential to narrow down your player pool of QBs to those that check at least some of the boxes of a highupside play.
DraftKings HighUpside QB Profile
Salary:
 At $6,359, the QB slot in firstplace lineups ranked fourth in average salary, behind WR1 ($7,994), RB1 ($6,850), and WR2 ($6,666).
 A large gap in average salary exists between QB and lineup slot with the nexthighest average salary (RB2, $5,044).
 Despite just over onequarter of all QBs being priced in $6,000–$6,900 range over last two years, firstplace Millionaire Maker lineups used one in that range nearly half the time.
 Despite half of all QBs being priced in the $5,000–$5,900 range over the past two years, only 35 percent of firstplace lineups used one in that range.
Scoring:
 QB in firstplace Millionaire Maker lineups scored 35+ points roughly onethird of time, 30+ roughly twothirds of the time, and 25+ roughly 8.5of10 times.
 Among QBs to score 25+ points over the past four seasons, roughly threequarters either passed for 300+ yards, threw for 3+ touchdowns, or both.
 Home favorite QBs accounted for the largest percentage of QBs to score 25+ points (43%, significant because it greatly exceeds home favorites' onethird share of the overall population).
 Road underdog QBs made up the secondhighest percentage of 25point scorers (25%, significant because it falls below the 33 percent expectation).
Projection:
 A 4for4 projected ranking of QB1–QB5 was most common among 25point scorers (31%),
 4for4 projected ranking in QB6–QB10 range (23%) and QB11–QB15 range (22%) weren't far behind QB1–QB5; middle salary tiers have been good place to find lowowned QBs without sacrificing value.
FanDuel HighUpside QB Profile
Salary:
 Average salary in firstplace Sunday Million QB was $8,109, second to only WR1 ($8,409).
 Sweet spot was $8,000–$8,900 (47 percent of firstplace lineups in past two years used QB in that range, but just 22 percent of QBs overall were in that range).
Scoring:
 QBs in firstplace Sunday Million lineup scored 35+ points 15 percent of the time, 30+ points 44 percent of the time, and 25+ 79 percent of the time.
 Over past four seasons, 83 percent of QBs that scored 25+ on FanDuel threw 3+ touchdowns, but just 69 percent hit the 300yard mark (due to FanDuel's touchdowndependent scoring system relative to DraftKings).
 Home favorites made up largest percentage of QBs to hit 25point benchmark (44%); road underdogs secondmost common (25%).
Projection:
 4for4 weekly positional ranking of QB6–QB10 accounted for significantly larger share of QBs that scored 25+ points (24%) than QB11–QB15 (17%, explains why firstplace lineups in Sunday Million paid up more at QB than in Millionaire Maker).
2. Exploit the LittleKnown Edges to Quarterback Stacking
Even if you're a DFS neophyte, you probably know that stacking is a good tournament strategy, and data backs it up. Pairing a quarterback with one of his teammates is a strategy that the majority of firstplace Millionaire Maker and Sunday Million lineups have used, although Sunday Million lineups (94%) have done so more than Millionaire Maker lineups (72%). Roughly twothirds of QB stacks in both sites' firstplace lineups have been twoman stacks, the majority of which contained a WR.
The data revealed two uncommon strategies that firstplace winners have been using to gain an edge:
 Shy away from stacking QB with an expensive WR: I don’t think I’ve heard anyone talk about this yet, but of WRs priced $8,000+ to appear in the firstplace lineups, only 28 percent on FanDuel and 17 percent on DraftKings were paired with their own QB. The average price of a WR in a QBWR stack was a modest $7,469 in the Sunday Million and $6,085 in the Millionaire Maker.
 Stack the QB with an opposing passcatcher: A passcatcher from the opposing team was included in nearly onethird of QB stacks in both sites' firstplace lineups. Including opposing passcatchers in a QB stack is an underrated strategy—especially given the relatively strong correlation between opposing passing games.
3. Fade the Chalk
In both sites' firstplace lineups, the QB slot had the lowest average ownership:
DK Pos 
DK Own % 
FD Pos 
FD Own % 

RB1 
20.7 
RB1 
21.2 
WR1 
18.6 
WR1 
19.7 
WR2 
15.4 
RB2 
13.2 
RB2 
12.4 
WR2 
10.5 
WR3 
12.0 
TE 
10.1 
D 
12.0 
D/ST 
10.0 
FLEX 
10.8 
WR3 
8.3 
TE 
10.0 
K 
7.5 
QB 
7.7 
QB 
7.4 
^{WR1 determined by most expensive WR in lineup, WR2 determined by secondmost expensive WR in lineup, and so on.}
QB ownership is easy to exploit.
There tends to be a dozen or so viable QBs each week, but the field tends to disproportionately favor the top one or two chalk options. This creates a situation where a number of QBs end up with low ownership that doesn’t truly reflect their odds of being one of the week’s top scorers. In other words, the odds of chalk QBs are overstated while the odds of all other viable QBs are understated.
You don't need to deliberately be contrarian at QB—you simply need to exploit ownership inefficiencies by fading the chalk.
How to Find Value Outside of the Chalk
I saved this section for last because all of the other valueseeking strategies and tactics we went over can be employed to identify strong QB options not among the chalk, such as a QB projected to be lowowned that also:

Is projected to be efficient.

Has a high implied team total.

Has had a poor fantasy performance recently that can't be attributed to anything other than natural variance.

Is a home favorite.

Is projected outside the top 10.

Is teammates with a cheap WR in your lineup (preferably one that is also projected to be lowowned).
As a final note, remember to avoid recency bias at all costs. As mentioned earlier, the correlation of QB fantasy points from one game to the next is almost nonexistent. It may feel uncomfortable to fade a popular QB option that is coming off a good game, but the better move will almost always be to play a viable option projected for low ownership that is coming off a poor game.
QB Daily Fantasy Playbook (Recap)
QB projection strategy:
 Weight efficiency more than volume.
 Leverage the predictiveness of Vegas implied team total.
 Use the correct stats and sample size when analyzing past performance.
 Fantasy points per game: Sample size of up to 56 games.
 Fantasy points per attempt: Sample size of 7+ games.
QB cash game strategy:
 Aim to get more value from QB than other positions.
 Allocate salary according to opportunity cost.
 DraftKings: Beware of overspending.
 FanDuel: Inexpensive options can be deceiving.
 Leverage Vegas odds to maximize consistency.
 Target home favorites and fade underdogs.
 Target Vegas implied team totals of 26 or more, and fade anything under 22.
QB tournament/GPP strategy:
 Target QBs that fit the highupside profile.
 DraftKings: 25point/300yard upside, $6,000–$6,900, home favorite, ranked in top 15.
 FanDuel: 25point/3TD upside, $8,000–$8,900, home favorite, ranked in top 10.
 Exploit the littleknown edges to stacking.
 Shy away from stacking the QB with an expensive wide receiver.
 Stack the QB with an opposing pass catcher.
 Fade the chalk.
Photo by Kevin C. Cox/Getty Images.
Click to become a 4for4 DFS Subscriber now!
Footnotes
1. The correlations for DraftKings salary to points scored are: RB .33, WR .32, TE .26, QB .25., DST .18. (back)
Related Articles