DFS For Dummies: Part II, Understanding Variance & Value in Daily Fantasy

Winning players not only understand variance and value--they exploit it.

This series of articles entitled “DFS for Dummies” details what I learned over the course of the previous 24 months while playing daily fantasy.  I will outline the habits and strategic thinking of successful daily players to help you make your learning curve far less expensive and much quicker than mine.  The series is broken down into 4 smaller articles to keep things focused and to allow the reader to refer back to them as the DFS season progresses; this is the second installment of the larger series.

PART II:  Understanding Variance & Value in Daily Fantasy

Part I of this series “DFS for Dummies” explained how successful daily fantasy players obtain information and begin to research the NFL game slate in any given week.  With that background in mind, we present Part II of the series, whereby we will explore two essential concepts in the daily space:  variance and value.  Each of these terms is generally not well understood by the casual DFS player; the pros, however, use variance and value to gain an edge on the competition.  Here, we will explain these terms, how they are determined, and what you can do to ensure that you appreciate how variance and value can maximize your daily profits.

Variance

Before we discuss the different game options and roster construction strategies in Part III of this series, we should first address a topic that you must fully understand when playing daily fantasy:  the concept of variance.  Variance, in DFS, is the degree of consistency of player scoring from contest to contest (week to week in NFL).  NBA (yes, basketball) fantasy scoring typically has low variance because of the number of ways in which a player accrues points (points scored, steals, blocks, rebounds, assists, etc.); NBA players, by virtue of the scoring system, tend to have extremely consistent fantasy scoring from game to game, something that is rarely true for NFL players.  Let’s take a look at a random 10-game stretch comparing points scored for 3 NBA stars (LeBron James, Kevin Durant, and Kevin Love) versus 3 NFL stars (Peyton Manning, LeSean McCoy, and Calvin Johnson):

Since this is “DFS for Dummies,” the goal is to keep the math as elementary as possible—the chart depicts the performance of each player over a random 10-game stretch last season; on the right are 3 key numbers:  1) mean points scored (per game), 2) standard deviation from the mean, and 3) coefficient of variation.  The mean is nothing more than the average points scored over those 10 games, whereas the standard deviation is a numerical indicator of how far away from the mean the player’s individual scores were over that same 10-game stretch.  Because NBA scores are typically higher than NFL scores, we need to somehow ‘normalize’ the numbers to allow comparison across the two sports--this is where the coefficient of variation (CV) comes into play.  The CV value is derived by dividing the mean by the standard deviation (SD/AVG) and then multiplied by 100 to yield a percentage value that can be compared across the two sports.  You will quickly notice that the CV values for NFL are dramatically higher than those for NBA (over double in some cases); this is variance illustrated in hardcore numbers.  It is worth pointing out that this phenomenon holds true for larger sample sizes, but we only highlight a few players here to keep the dataset as simple as possible.

Why is NFL scoring so variant?  NFL is a highly-variant sport for daily fantasy because the scoring is “event-based,” whereby a touchdown is worth as many points as 60 yards gained (for RB/WR/TE).  Yardage is somewhat predictive, whereas touchdowns are often a crapshoot.  For example, an RB or WR can be tackled at the 1-yard line and a team will bring in a bruising goalline RB to vulture the touchdown, thereby ‘stealing’ points from the player that did the lion’s share of ball handling to that point.  Likewise, a QB can pass for 300+ yards and have his receiver get tackled inside the redzone throughout an otherwise productive game and never actually throw a touchdown pass, thereby limiting his final fantasy output.  

The key to consistent winning in daily fantasy football is not to ‘beat’ variance because that is virtually impossible in the near-term; if you make the correct, value-based roster decisions, you will, however, come out on the right side of variance over the long-term.  To accomplish this, you must endeavor to understand variance, how it affects next week’s salary from site to site, and how to exploit it to your advantage.  On daily sites, salaries are determined by algorithms that take into account multiple factors including recent and long-term performance, historical percentage-owned, and weekly matchup.  A prime example of how variance and player information (Part I of this series) can be used to one’s advantage is Marques Colston of 2013; through 10 games last year, Colston played somewhat injured and only mustered 2 touchdown catches through 10 games.  Despite nearly 70 targets, his touchdown rate was miserably low for being the #1 wide receiver on a potent Saints’ offense.  As the season progressed, Colston’s health improved and he was used more often in the Saints’ offense.  His salary entering Week #11, however, was extremely low throughout the industry because of his lack of touchdowns to that point; if you rostered him that week, you would have been rewarded with a 9-reception, 125-yard, 2-touchdown performance that was good enough for 29 points on FanDuel and 43.5 points on DraftKings!  Remember to try to roster stud players whose salaries are depressed after several bad weeks of performance; when they return to the right side of variance, you will reap the rewards--for GPP’s, this is a strategy that can pay huge dividends!

Value

In daily fantasy circles, the term ‘value’ is perhaps the most frequently discussed topic in any given week.  A beginner on FanDuel or DraftKings will put together a roster that he thinks will score a lot of points; an expert on those sites knows approximately how many points he needs to win and uses value as a means to determine exactly who to roster.  ...but what is value and why is it important?  

In our everyday lives, we seek value—we try to maximize the value of our hard-earned dollars by comparing prices for major purchases, by purchasing generic products when appropriate, and by investing our money into assets that appreciate most consistently.  The last example is perhaps the most amenable comparison to daily fantasy—when we make investments, the expectation is that investment will provide us the highest return on investment (ROI) possible.  In daily fantasy, that ROI comes in the form of ‘points per dollar’ and is a major determining factor for most of our decisions with lineup construction.

“Value” is a qualitative term used to describe the number of expected points per dollar that a player can potentially return in any given week.  Of course, there are 2 components that derive value: 1) site-specific salary, and 2) projected points scored.  On Sunday night, the daily sites will release their salaries for the following week; later that week, FootballGuy David Dodds releases his detailed projections for that Sunday.  With these 2 bits of information, you are now able to generate very specific numbers that can be used to estimate value for the following week--just divide the player’s salary by his projected points scored to determine points per dollar for every player.  If you are lazy or simply don’t have the time...don’t fear...Dodds provides his own ‘value sheet’ to all subscribers nearer the end of the week, when practices are completed and his projections are most accurate!  Many successful daily players put together an Excel spreadsheet to help narrow down selections and pinpoint value every week; see below for an example for quarterbacks on DraftKings from Week 16 of 2012:

 

You should immediately notice several things:  1) Matchup depicts the QB’s opponent and whether they will have home-field advantage (in bold), 2) Site-specific salary is used to help derive a “dollars per point” value, 3) Trusted projections (predicted points scored) from multiple sources (including my own), and 4) Value, as indicated by “dollars per point.”  Just to be clear, I average multiple projections (“Average Projected”) and divide that number by the player’s salary to get the dollars per point figure (I typically use “dollars per point” because it is easier to work with whole numbers).  At that point, I sort the list and color code value, where light green represents the best value and red represents the poorest value.  This is the first step to capturing relative value.  To interpret the spreadsheet, one only needs to look for players with the lowest dollar per point to determine those players with the highest value for that week; in this example, Chad Henne had the best value ($260/point) based on his predicted fantasy output for that week, whereas Drew Brees had the worst value at $479/point.

Deriving value, however, is not as simple as finding the player with the lowest dollar per point number and putting him on your roster because, often times, that player has an extremely low salary and you will be left with excess salary after filling all positions; as a general rule, you never want to leave more than 0.5-1.0% of your overall salary unspent or you will be leaving points on the table.  Thus, it is imperative that you choose your value players judiciously and fill in your roster with quality players, where your excess salary allows.  For cash games and GPP’s, this approach will differ slightly, as you will learn in Part III of this series (“Daily Game Selection & Roster Construction Strategy”).

RECAP

  • Variance is high in NFL.  Don’t get discouraged when you end up on the wrong side of variance; instead, use variance to your advantage to roster players who are in store for big Sundays.  If value is your #1 factor in roster construction, you will not lose to variance in the long run.
  • Determining value is essential for consistent winning in daily fantasy football.  Combine solid projections with player salaries to derive value on a points per dollar scale.
  • Spreadsheets can help sift through the data.  Finding value is simplified by using a spreadsheet methodology to extract value from player projections; spreadsheets should be used as a guide, however, not as a means to determine every player at every position.
  • Never leave more than 0.5-1.0% of you total salary cap unspent.  Leaving money on the table is almost always a mistake.

The next installment, Part III of "DFS for Dummies" will expand upon the use of value as a means for roster construction in cash games and large-field tournaments.


More articles from John Lee

See all