Daily Fantasy Football Player Selection Strategy by Contest Type – DFS Army NFL Training Camp Series

basic-training

I’m back with the next installment in our “Training Camp” series. So far we’ve covered how to interpret the Vegas Lines, Bankroll Management, and Player Correlations. In this column, we are going to tie everything together and look at how and when to apply the lessons we’ve learned as they pertain to the following three contest types – Cash, Hybrid, and GPP. Before we get into the fantasy football player selection strategy side of things it’s important to define the three contest types we are looking at.

DEFINING THE THREE CONTEST TYPES 

Cash Contests – In DFS the term “Cash Contest” generally refers to contests which pay out a large percentage of the field ,  (I’m adding something to the definition here that you won’t see anywhere else …. yet, and do produce an additional reward for an outstanding or well above the mean score.  Double ups and  50/50’s are the most common “cash contests” per this definition.

GPP  (Guaranteed Prize Pool) –  For the purposes of this column and going forward in my Vegas Lines columns I’m using the term GPP to define larger contests with very top heavy payouts.  Qualifiers and promotional contests like the Draftkings Millionaire Maker are great examples of GPP’s. I generally consider a top heavy payout to be anything over 10% of the prize pool going to 1st place. Consider that in the Millionaire Maker Contests I’ve seen as much as 30% of the prize pool, in a contest with hundreds of thousands of entrants, going to the 1st place winner. Similarly, in live final qualifiers, you see as much as 50% or more of the pool going to a single player.

Hybrid Contests –  The are contests that are not top heavy like our “GPP” definition, but actually do incentivise outstanding scores, unlike our “Cash” game definition. Examples of Hybrid contests include tournaments with flatter prize structures that payout over 20% of the field, head to head contests and leagues. You may be surprised to see Head 2 Head contests listed here but when executed properly, a head to head strategy will reward outstanding scores.

WHY USE THREE DEFINITIONS?

You may be wondering why defining these three contest types is important. After all, shouldn’t we strive to build the best lineups, score the most points and win some money? Unfortunately, it’s not that simple. Each contest type rewards different things and therefore in the pursuit of an edge on the competition we can use our definitions to narrow down our lineup construction methodologies for each contest type and hopefully increase our ROI. Think of it like this, we know that it’s important to find low ownership players in top heavy GPP’s. We know that fading the chalkiest plays in GPP contests is the right thing as far as game theory goes. Even knowing all of this, most people are too scared to do it. We see that spectacular matchup and we want a piece of that player. Separating your entries into these different categories really allows you as a player to get past the fear of the fade. If Julio Jones has a cherry matchup against a porous defense in the highest total game of the week, we naturally want to roster him all over the place. Breaking up your thought process into three completely different contest types allows us to play Julio in our Cash and Hybrid lineups. Using him there assuages the fear of not using him in our top heavy GPP entries.

Psychological Rule of Thumb #1 – When you can mentally separate the different strategies and approached to the different contest types, it can help eliminate the fear of going against the grain. 

STRATEGIES FOR EACH CONTEST TYPE – CASH, HYBRID, & GPP

CASH LINEUP CONSTRUCTION

Earlier we defined “Cash Contests” as mainly consisting of 50/50 and Double Ups. The defining characteristic is that these contests DO NOT reward players for scores well over the mean. In a cash lineup, the ultimate goal is to produce a score that beats the mean score of that contest. Beating the mean by one point is just as good as beating it by 50 points. There is also a harsh penalty for coming in even slightly below the mean. The penalty being that you lose all of your money. Losing all your money sucks so our goal is to do everything in our power to avoid that situation. How do we use this knowledge to assist with fantasy football player selection? First and foremost we need to understand that ownership percentages, or knowing what your opponent is doing, is completely irrelevant when constructing cash lineups. We aren’t trying to beat the field here, we’re simply trying to beat the mean score.

Cash Game Rule Of Thumb #1 – Play the chalk. Don’t worry about what your opponents may or may not be doing. 

Another key strategy in cash games is to diversify risk by avoiding players that are correlated. If you look back at the Training Camp column on player correlations we defined a number of situations in which players performances are strongly correlated to one another. The most obvious and robust correlation is pairing up a QB with their WR1. Using a lineup with players whose scores are highly correlated works really well in GPP and Hybrid contests but it’s the opposite in cash games. Instead, we want to build a lineup that can withstand a poor performance or two and still reach the cash line.

Cash Game Rule of Thumb #2 – Avoid using highly correlated player combinations.

In a cash lineup, predictability and floor are the most important factors. We know from statistical studies that the QB and RB positions are the most consistent, while WR/TE, D and K are the most volatile. We need to use this knowledge to our advantage when crafting our cash lineups. Lets look at the RB position. We generally see two different types of Running Backs in the NFL. You have your three down backs that are heavily involved in the gameplan regardless of game flow considerations. Matt Forte, in his tenure with the Bears, is a perfect example if this type of player. Forte was heavily utilized in the passing game and on the ground regardless of game flow. He generally produced consistently solid games because RB fantasy scoring is extremely correlated to opportunities. RB’s that get opportunities in all situations are the ones we want for our cash games. Unfortunately, these types of RB’s generally come with high salaries.

Cash Game Rule of Thumb #3 – Pay up for consistent RB’s in Cash Lineups

Football is a game of injuries and no position is more prone to injuries than the RB position. When backup RB’s are thrust into a starting role the salary algorithms for the DFS sites are not always able to price in the additional production. The best of these situations occur when the backup is clearly going to get the lion share of the opportunities.

Cash Game Rule of Thumb #4 – Don’t shy away from value. Use those replacement RB’s!! 

On the flipside, we know that on a week to week basis the WR position is very volatile. That’s great for contests that require massive scores to win, but in cash games, we want to pay for consistency and avoid volatility. In essence, we want players here that get a decent percentage of the targets from their team and are less dependent on TD’s for their scoring. Think of a player like Julian Edelman as the perfect example. Edelman doesn’t necessarily score a ton of TD’s but he’s consistently good for 7-10 receptions and 70-100 yards per game. In cash games, we are generally spending up for game flow proof RB’s so we need to save a bit on the more volatile WR position. The key is to find players with relatively high floors. We want to stick to chalky players, ideally playing at home in games with high Vegas totals.

Cash Game Rule of Thumb #5 – Target value WR’s with high floors that are not dependent on high variance situations, like scoring  TD’s, to reach value.  

The QB position is actually one of the most consistent as far as fantasy scoring goes. Conventional wisdom says you should pay for stud QB’s in cash lineups. It’s a tough one to argue with but I’ll add a caveat. QB’s that average more than four runs per game tend to be the most consistent fantasy scorers. I’m ok with saving some salary in a cash game by going with a player like Tyrod Taylor at QB. He may not bring the Cam Newton 40+ point upside to the table, but the floor is safe.

Cash Game Rule Of Thumb #6 – Pay up at QB and lock in those consistent points

The most volatile, and least predictable positions in fantasy football are Defense and Kickers. Defensive scoring is often predicated on high variance situations like special teams TD’s, pick 6’s and fumble recoveries. In cash games its best to stay chalky with the defense and pay as little as possible salary. Same can be said for the kicker position. I’ll rarely pay over min salary for a kicker in a cash game.

Cash Game Rule of Thumb #7 – Spend as little as possible on the high variance positions of K and DST. 

Our last position to consider is the flex. Both Draftkings and Fantasy Draft feature a flex position instead of a Kicker. We’ve already covered the fact that RB’s are more consistent than WR’s. It’s generally better to use an RB in the flex spot unless an unusually great value pops up at one of the other positions.

Cash Game Conclusion:

The most important thing to keep in mind when crafting a cash lineup is to focus on the floor of the lineup. We want to construct a lineup that can withstand a scenario where we were wrong about a particular game and still cash. Therefore we avoid using correlated players and keep things as chalky as possible.

HYBRID LINEUP CONSTRUCTION

If you have followed this blog for the past few years or have been a reader of my weekly “Vegas Lines” columns or strategy content you know that I’ve generally preached a hybrid approach to DFS. I tend to create a series of  hybrid lineups, between 10-30 per week and use them in all different types of contests. I’ve always focused on head to head matchups as the primary driver of my bankroll and I’ve been fortunate to have maintained a consistent 61% win ratio across head to head contests going back  through 2012 when I started getting heavily into daily fantasy football. That’s great of course but then I noticed something odd. My 50/50 win rate (54%) was lower than my H2H win rate which seemed odd at first considering I use the same lineups for both types of contests. As a prolific researcher and stats geek, this discrepancy bothered me and I wanted to figure out what the deal is. Most people approach head to head contests in the same way as 50/50’s. Generally, these types of contests are referred to as “Cash” contests which generally defines games that pay out around half of the participants. I actually think these two contest types are quite different. You may be thinking of head to head contests as a mano a mano situation. Realistically though doing H2H the right way involves making a lineup and entering it in a bunch of head to head contests. In a single head to head contest, you never know what type of score your opponent will put up. This is different from a 50/50 or Double up where you know the cash line will be the mean score. If I take a single lineup and enter it in 50 head to head contests the result should be a bell curve of scores:

bellcurve

Most of the opposing scores will fall within the “Average Performers” range. Some will be very high and some will be very poor. There is safety in this bell curve. If you wind up with a sub-par lineup you probably won’t lose all of your money. It would have to be a complete bomb to produce a total loss. Lets compare this to the cash contest return from the previous section. We know that in a cash contest a score 1 point below the mean will equate to a total loss of your entry fees. A score of 1 point above the mean will double your money. If we take the same $100 we put into the cash lineup and instead put it into 20 $5 head to head contests our results are completely different. Scores of 1 point above or 1 point below the mean will probably result in a break even situation. In this same situation, if we can put up a score that is one standard deviation unit above the mean we should win around 75% of our entries.  This is how Hybrid Contests differ from cash contests.  They reward scores that are well above the mean.

Which types of contests are Hybrid’s and how do we attack them? Aside from Head to Head contests, the hybrid approach to lineup construction applies to leagues, Triple and Quadruple ups, and some of the new flatter payout structure GPP tournaments that I’m seeing featured in the lobbies of the big sites. As a rule of thumb for identifying hybrid GPP’s, I’d say any contest that pays out over 25% of the entrants and pays 10% or less of the prize pool to 1st place.

How To Attack a Hybrid Contest

Now that we’ve identified what a Hybrid contest is we need to understand how to attack it. We’ve established that Hybrid lineup construction methodology differs from Cash in that we get a tangible benefit for scores well above the mean. Look at the bell chart above. Our goal in our hybrid lineup is to attain a score that is one standard deviation unit or more above the mean. This brings us around to player correlations. In my Player Correlation Training Camp Series Column, we established a number of lineup construction “Rules of Thumb” as they pertain to player stacks. I’m not going to retype it all here but what I will do is explain the logic. Since we are targeting scores at least one standard deviation unit above the mean, we can say with some certainty that all of the players on our fantasy team need to put up quality scores if we are going to cash. This is where stacking comes into play. I start every hybrid lineup with a QB-WR1 Stack. That is the baseline.

I start every hybrid lineup with a QB-WR1 Stack. That is the baseline. The logic is simple. We are building our lineup for upside. We know that if our WR1 puts up a dud we probably aren’t going to cash. We know that when a WR1 does well, their QB is correlated and should also do well. In essence, we want to construct lineups with players that are correlated across the board.

Another difference between Hybrid and Cash lineups is how we look at the RB position. In my Interpreting the Vegas Lines Training Camp Series column we discuss how to interpret the Vegas Lines to predict game flows. This is going to be key in our Hybrid lineups. Since we are sacrificing consistency for upside we want to save salary by using players that may be game flow dependent. I love to use the two down RB example here. We know that two down RB’s tend to do well in games where their teams are favored at home because the assumption is that they will be protecting a lead or running out the clock with a big browser RB in the second half of the game. Similarly, a 3rd down scat back type player like Danny Woodhead tends to excel in games where his team falls behind and goes into the hurry up offense. Both player types tend to have depressed salaries on the DFS sites because they don’t produce consistently. They are game flow dependent. In Hybrid games, unlike cash games, it is ok to roll the dice with game flow dependent players. We want to try and interpret the Vegas Lines, predict game flows and be right about it.

Hybrid Lineup Rule of Thumb #1 – Stack Correlated Players – (QB & WR1)  (D + K)  (D+RB)

Hybrid Lineup Rule of Thumb #2 – Take some risk with high variance game flow dependent players

In the “Cash Lineup” section we discussed how QB and RB are the most consistent positions while WR, TE, D and K are the most volatile. This is another situation in which we need to flip the script as far as how we construct our lineups. We can take more risk on volatile players in our Hybrid Lineups so it makes sense to spend up on WR’s and TE’s while saving cap on RB’s and QB’s.

Hybrid Lineup Rule of Thumb #3 – Spend more for WR’s with massive upside, Save on the QB and RB positions

Should we worry about ownership percentages when constructing Hybrid Lineups?

One of the trickiest things about daily fantasy sports, in general, is understanding when to fade the chalk. Chalk plays are the most obvious plays of the slate. Fading the chalk is not necessary however in our hybrid lineups because we aren’t trying to beat the entire field. Instead, our goal is to produce fantasy scores in the top 25% while simultaneously protecting our floor. That doesn’t mean that we need to stuff every chalk play into our Hybrid Lineups, instead, it’s a general understanding that there is no need to deliberately look for low ownership situations just for differentiation. The idea is to construct highly correlated lineups grounded in our Vegas Lines and value models that should combine high upside with a high floor.

Hybrid Lineup Rule of Thumb # 4 – Don’t get cute!

Here’s the deal. Every week we play in tournaments and we see someone select a player in a terrible situation, that winds up going off for a big day. Let’s call this one the Tolbert / Kuhn rule. Tolbert is the goal line vulture on the Panthers. Kuhn was a goal line vulture on the Packers. These guys would get two or three carries in their best game strictly around the goal line. Is it possible that they convert two of them for TD’s? Sure it is. I don’t give a shit. These aren’t the types of players we want in our hybrid lineups. I’m not spending my hard earned money banking on players that get two or three opportunities per game. If you feel the urge to get cute, see the next section.

GPP LINEUP CONSTRUCTION METHODOLOGY

 For the purposes of this column, GPP contests are basically top heavy tournaments with over 2000 entrants. By Top Heavy I am referring to contests that pay out over 25% of the prize pool to the top 10 finishers. Some of these contests pay out as much as 50% of the pool to the top 10 finishers. Qualifiers and 10x contests fit this category as well. If you look back on the bell curve chart above, in a GPP we are looking for results that fall two or more standard deviation units above the mean. When you enter this type of GPP keep two things in mind. First, the goal is to win. We plan our lineup with the understanding that it will take outstanding performances from every spot on our lineup to take down a 1st place finish. The second thing to understand is that you probably aren’t going to win so be prepared to kiss your entry fee goodbye. There’s a reason that I’ve been preaching the 80/20 bankroll management allocation standard for as long as I’ve been writing my columns. Winning a GPP is a massive long shot and most people will never get there unless they play with discipline.

GPP Lineup Rule of Thumb #1 – You need studly performances from all of your players to win. Use this absolute knowledge to your advantage when constructing lineups! 

As someone that has been writing about daily fantasy football for many years, I often get tweets or messages in our DFS Army slack forums from people asking me what I think of their lineup. I generally ask what type of contest they are putting it in. When they say GPP, my standard answer is “looks good to me”. There’s no real point in critiquing a GPP lineup. Last season someone won the Draftkings Milly Maker with a Pittsburgh stack that included Ben, Brown, Martavis, Bell and Pittsburgh D. I believe it was the week that the Steelers were playing at home against Denver, the #1 D in the league. Normally I wouldn’t recommend stacking four offensive players from a single team and that team’s defense. Still, for whatever reason, that stack won the biggest GPP and turned a truck driver from Pittsburgh into a millionaire. Amazing stuff. If he would have shown me that lineup in advance of the contest I’d probably be chuckling to myself about that Steelers homer choosing a ridiculous lineup full of home team players in an awful match up. That homer had the last laugh.

Does this mean we should simply pick a random team of players in terrible match ups and toss it into a GPP? Now that I think of it that may not be the worst idea in the world but realistically no. There actually is a logic to GPP lineups which we will cover after another anecdote.

Just this afternoon I was talking to a buddy of mine that I often talk fantasy sports with. He was talking about a survivor pool he is in and was asking which game in week one was the chalk game. In a survivor pool, the chalk game is generally the one with the biggest spread that week. My buddy was looking to fade the chalk game because he knows, from years of playing in survivor pools, that at some point a chalk play will bust and take out 75% of the field. It’s not difficult to look up the Vegas lines and play the heaviest favorite, particularly in the early weeks of the pool.  The same can be said for GPP entries. If you know that 40% of the field will be on a particular player, it can make sense to fade that player. Generally, there is a good reason that player is highly owned. They are probably a great value in a great spot. Still, we have to remember that the goal in a GPP is to win. Lets say that the 40% owned player will produce a spectacular result 7 times out of ten. Those other three times, when that player produces a dud, 40% of the field is immediately knocked out and your win odds go up exponentially. That doesn’t mean you should always fade the chalk in a GPP however. Chalk is chalk for a reason. Still, if you think two players are in good spots, but one is going to be 40% owned and the other should only be 10% owned, you should take the lower owned player simply for differentiation.

GPP Lineup Rule of Thumb #2 – Don’t just fade the chalk to be differentiated. Make sure you are pivoting to a player with similar upside but with much lower usage. 

I have a really cool example here that pertains to week one of the 2016 season. With the recent news that Tony Romo will be out for a while, tje idea of rolling out Dak Prescott in cash and GPP is everywhere. He is a min salary QB that has looked spectacular in the pre-season. Most people have ignored the news that the Broncos are rolling out Trevor Siemian as their starter in week one. Siemian is also a min salary player but he hasn’t stood out as much as Prescott in the preseason. Anytime a QB is at min salary they are worth a look. In this case, we have that situation on a player that is likely to be very low owned. These are situations to exploit in GPP lineups.

GPP Lineup Rule of Thumb #3 – Create correlated lineups

GPP Lineup Construction Rule of Thumb #4 – Stack up high salary WR’s with their Cheap salary QB

In my previous Training Camp column on Player Correlations, we went over different positions and how they correlate to one another. In that article, I defined some player correlation rules of thumb. I won’t rehash everything here other than to say that a GPP lineup should use players that are highly correlated. The one certainty we know of in GPP’s is that you need every player on your team to produce a quality fantasy score. If your stud WR puts up a dud you aren’t winning shit. Therefore, particularly if he is connected to a low-cost QB, you should always pair up that combo. Basically, if you plan to use Dez Bryant at stud level salary, you better get Dak in there as well.

GPP Lineup Construction Rule of Thumb #4 – Don’t be stupid

There is differentiation and then there is stupid. I remember a couple of seasons ago, Riley Cooper was playing the outside WR spot for the Eagles. The guy was lucky to average 2-3 targets per game. One week he got lucky and was caught two of three passes thrown his way, both went for long TD’s. The next week people were rushing to roster him and calling him an ideal GPP play. First of all, let’s not forget that Cooper is a racist. Screw him. Secondly, a guy that gets lucky with a 2 TD on 3 targets game is far from an ideal GPP play. He is a stupid play. Let the suckers grab low volume players. We want guys in good spots being overlooked because a similar player is in a slightly better spot.

CONCLUSION

The topic of lineup construction methodology is one of the most fascinating yet overlooked areas of Daily Fantasy Football strategy. As players, we need to train ourselves to look at each contest type through the lens of what it takes to be successful in that specific type of contest. In cash, the goal is to beat the mean. We build lineups that can withstand the shock of a player or two underperforming. There is no reward here for a top notch score. Keep it chalky, play the best players in the best situations. Floor is king in cash games. In Hybrid we consider both floor and ceiling. We build correlated lineups for high upside but there is no need to fade the chalk. We’re looking to put up a score between 1 to 2 standard deviation units above the mean. In GPP’s we are looking for 2+ standard deviation units over the mean so this is where we can take some chances. Fade chalky plays for similar low ownership situations. Above all other things, in all three formats, don’t take players that suck.

Daily Fantasy Football Training Camp Series

Part One: Stacking – DFS NFL Lineup Construction Methodology

Part Two: Vegas Lines and Player Selection – Advanced DFS Strategy

Part Three: Player Selection Strategy by Contest Type