Photo: Imagn Images

In my last two articles I talked about what types of games to play and how to manage your bankroll. In this article I am going to lay out a game plan for setting the optimal lineups at one of the best daily sites, Fanduel. If you think you are going to be a successful daily player just because you are a great season long player you are going to be in for a rude awakening. You need to refine your skills around setting your lineup and understanding how different positions have different ceilings and floors and what games you need to take risks in and which ones you should play it safe in.

It may seem like I am beating this one to death but the first thing to consider when setting your lineup is the type of game you are playing. It may seem simple to just start taking the best values when choosing lineups but you really need to start by determining what type of game you want to play, and that usually comes down to how much risk you are willing to take. As mentioned in my previous articles the two most common types of games are the large guaranteed prize pools (GPP) and head to head (H2H) games both of which significantly vary in terms of risk. As such the strategy between the two couldn’t be more different.

One thing that makes these two types of games so different is the score that it takes to win. In GPPs you typically have to finish in the top 10% to even get any money and have to finish in the top 1% before you get a significant payout. I can’t even tell you exactly what a winning score is going to be in a GPP because it varies so much. Some weeks a score of 210 won’t even take down first place if your opponents have both Josh Gordon and Alshon Jeffery going berserk and other weeks 146.4 points might be worth $1 Million. That said over the long run a score of approximately 180 will earn you a very nice profit and will put you in the running for a big payout.

Let’s compare that to what it takes to win in a H2H league which is fairly predictable. The graph below shows actual results from thousands of games at varying buy in levels and represents your chance of winning based on your points scored.

The first number that stands out is you have about a 50/50 shot of winning with a score of 113. Compare this to the points needed in the GPPs above and you will quickly see that you need to take a lot more risk in the GPPs.

The second thing that stands out is there isn’t any magical score that is going to win you a game but the curve is the steepest between 100 and 120 points and if you get to the top of that curve you are winning about 65% of the time and will start to earn a pretty consistent return on your investment. We can set that as our goal but it won’t be easy to achieve.

All of these stats are for contests with a salary cap of $60,000 so if we set our goal for GPPs at 180 points and our goal for H2Hs at 120 points we are looking at a cost of approximately 500 and 333 Dollars Per Point (DPP), respectively in each type of contest. What’s the best approach to allocate our money in order to reach these targets?

At Fanduel a starting lineup consists of 1 QB, 2 RB, 3 WR, 1 TE, 1 DEF, and 1 PK. We can choose any combination of player to fill those slots. However, we can quickly eliminate a lot of players based on the great projections here at Footballguys. If we limit our player pool to the guys that are most likely startable players we can start to figure out how to best allocate our resources.

Narrowing our population to these startable players is important because if we base our population on the entire pool the results won’t be representative of the players we are picking from during the season. Many of the players outside of our player pool will have different average scores, standard deviation of scores, and in turn different Coefficient of Variations (CV). CV is simply the players standard deviation divided by their average score.

The question is how do we define “startable” players?

To answer this question I used data from the great “On the Daily” series from Maurile Tremblay and my own personal experiences. I determined that there were actually two thresholds that needed to be met based on each players DPP value. The first threshold would be how that player’s value ranks at his specific position.

The second threshold needs to be created for stud players. This is the case because if you pick only the best values from your value chart you will most likely end up not spending all your money and therefore not maximizing your potential points. This allows us to pay a little more for good values on studs. For our purposes a stud is defined as top 5 most projected points at QB, TE, PK, and DEF and top 10 at RB and WR.

Here is how to read the table for the QB line. If the player is rated in the top 5 of all players or if he is a stud player and he ranks in the top 10 in value he should be in consideration for your lineup. Since there isn’t really much value related to studs at defense or place kicker I have left their stud threshold to be the same so they need to rank in the top 4 or 5 respectively to be considered.

If we use the guidelines above and apply it to the 2013 weekly rankings from David Dodds, Maurile Tremblay, and Sigmund Bloom to create a “startable” player pool we find that we can expect the following DPP at each position using all of the player’s results from their actual games. 

What this means is if you spend 8,000 on a startable QB you would expect them to score 20.1 point. This is calculated by taking the amount spent of $8,000 and dividing by the actual DPP from the table ($8,000/398DPP). On the other hand, if you spent 8,000 on a TE you should only expect about 13.2 points. 

There are two really important things in this chart. First of all the total actual DPP of 512 points from this group of players gives you a projected score of about 117 which from the chart above will give you a very good win percentage of 58% in a head to head league and a steady return on your investment even considering the rake that Fanduel charges. On the other hand it won’t get you too far in GPPs.

The second thing is that all positions are not priced equally as QBs provide a lot better DPP than the other positions and TE and PK provide very poor DPP. Your first thought may be that you should try to maximize the amount of money that you spend on QBs to maximize your DPP.

Unfortunately, maximizing our DPP spend across positions isn’t so simple. We need to first take into consideration the risk of each position and we also need to consider how else we can spend the savings.

There are a lot of ways to determine the risk involved with each position but today I wanted to break down each position by how often each player reached specific point thresholds using real data from last season.

For example, we determined a player would need to produce above333 DPP to be a good start in a GPP and would need to produce above 500 DPP to be a good start in H2Hs. We will call these GPP and H2H. There are two more categories that are pertinent to measure the downside of a player. I will call the first one “Poor” which is anything over 1,000 DPP and “Average” which is a value between 500-1,000 DPP.

The table below shows how often each of the startable players fell into each bucket and it reads as follows. Startable defenses will score below 1,000 DPP (Poor) 22% of the time. They will also score between 500-1,000 (Average) 36% of the time. Finally, they will score over 500 DPP (H2H) 42% of the time and above 333 DPP (GPP) 19% of the time. Remember, GPP value is a subset of H2H value so the first three columns should total 100% not all 4.

Here we can see that 43% of all startable players achieved H2H value last year and of those players only 18% achieved GPP value. To put that in perspective on an average week only 1.6 players on your team will put up a GPP worthy performance. That’s why we need to take a lot of risks in GPPs. 

Now that we know how often each position scores in each bucket we need to understand the opportunity cost of spending more or less on each position. In order to answer this question let’s split each position into three separate groups: bargains, studs and everyone else. We will use the same definition of stud as above.

Bargain status is a little more difficult to categorize due to the pricing structure but I decided that the following looked most accurate to me. Please note that in some cases players can be both a stud and a bargain player. Think of running back like Christine Michael who is playing for the first time for an injured Marshawn Lynch. His price will be very low but he will most likely be ranked as a stud in the rankings. This is especially possible for defenses who are so matchup dependent.

Now let’s split the chart into the three categories of studs, bargain and other? Finally, let’s add one more column called “Gap” which is simply the difference between H2H and Poor? This gap score is important because it shows us the difference between the odds of reaching our goal and a game that kills our chances of winning.

Let’s look at how we can use the data in the table to help us determine our optimal lineups in GPP’s. Remember, for GPP’s we are really only concerned about upside and protecting against downside since we need the highest score possible.

GPP’s

QB – Quarterbacks have by far the highest amount of GPP success at the bargain level but they also have one of the largest drops when it comes to comparing that success at the stud level. In fact they go all the way from a 43% success rate at the bargain level to 23% at the stud level. That said even at 23%  QBs are still one of the most successful values at the stud level.

This value comes from the way that Fanduel has priced quarterbacks as we saw in the table above. They simply cost the least in terms of DPP. So even though  QBs have lower upside in points scored since they are priced much more favorable they become an attractive option even at the stud level. However, keep in mind that since they are even more successful at the bargain level it may be worthwhile to cut your spending at quarterback if you have other intriguing studs to play.

RB –Running backs are very similar to quarterbacks in that they have the biggest success rate drop as we go from bargain level to stud level. In fact by the time that we start comparing the positions at the stud level RB falls to the worst odds of GPP success. We certainly do not want to pay for stud  RBs in GPP’s when we can be so much more successful with bargain players.

The reason we see these big decreases in value at QB and RB is because both are the positions that get the most consistent workload. Before each game you often know approximately which player from these positions is going to get the majority of the opportunities in each game and they often get quite a few opportunities. Since the number of their opportunities doesn’t vary as much neither does the standard deviation of their scores which results in a lower CV. Because of this it is hard for them to reach the ceiling needed to get GPP value.

WR – WR is the opposite of QB and RB. As we move from bargain to stud level there is no change of success rate and WR is tied with QB as the most successful GPP position. This happens because at the wide receiver position the CV is much larger. This gives the stud wide receivers the opportunity for a huge upside game that can help you win your GPP. However, since the low end WR opportunities aren’t nearly as predictable for the bargain players it makes it difficult to predict their scores. We see this to be the case in the number of poor performances by bargain wide receivers. Remember in GPP’s we have no opportunity for error so a poor performance is going to kill our chances for a win as fast as a GPP level performance will help it. These two cases make it very clear that paying for stud wide receivers is the way to go in GPP’s.

TE – TEs are similar to WRs because there doesn’t really seem to be much difference between studs and bargains at the GPP level. However, TE success rates are much lower than WRs because of how expensive the baseline DPP’s are for TEs. But one area that is clearly affected is the number of poor performances by the stud players. As mentioned above these poor performances can hurt our odds substantially, as such I would tend to lean towards spending less on tight ends.

DEF and PK – Here we see that actually spending on a stud player is slightly more favorable than spending on a bargain player. But remember this is only for studs that are rated as very top values. In essence if there is a stud and a bargain both with similar values we should go ahead and spend a little more for the stud since they have both a slightly higher floor and a slightly higher ceiling.

Head to Head

In head to head matchups we are looking for consistent performance. We need good scores but they don’t have to be nearly as explosive. A couple of good performances will put us in good shape for a very good week. However, we do want to limit downside as poor games can quickly derail a great week. As such the difference between this floor and ceiling is really important in head to head games so we will primarily look at the gap column here.

QB - Quarterbacks are the most consistent position at both the stud and bargain level. In both cases they are the most likely to provide H2H value, over 65% in both cases, and the least likely to fall flat on their face with under 10% poor rate in both cases. This makes a QB a very flexible position where it makes plenty of sense to spend on a stud or go a bargain route. In fact for  QBs you can even go the route of other  QBs if necessary. However, because of how successful the quarterbacks tend to be a default standard would be to spend more on  QBs to lock in more points.

RB – The gap at running back is 24% at the stud level and 15% at the bargain level. This isn’t a big improvement but does favor going towards a stud over a bargain player. This difference is due primarily to the fact that stud running backs so rarely have a poor performance. This again is based on the principle that the opportunities that the running back is going to get are a lot more predictable from game to game which makes their CV much smaller and they won’t bust nearly as often. Similar to QB I think you can really choose from either of the three player pools but I would tend to break ties towards the more expensive players.

WR - At wide receiver we see the largest increase in the gap as we move from bargain players to studs. This is consistent with the findings mentioned above in the GPP table. Just like at the GPP level bargain studs don’t have many more H2H level games but they do have a lot more poor games. In order to avoid these poor games we should pay more for WRs.

This may be a little surprising considering that WRs are typically one of the most unpredictable positions and in H2H’s we want as much consistencies as possible. The problem is the bargain WRs are more unpredictable than the studs so it makes sense to eliminate some of this unpredictableness and take the lesser of the two evils.  That said since WR is a little more volatile I would be happy to take my best bargain player to combine with my studs.

TE – Tight end is the position that most stands out. Based on the stats you should go with the bargain tight ends as often as possible. Their H2H success rate is one of the best as is their bust rate. But when we look at the studs list we see that they are one of the most consistent busts and very rarely provide our target value. In fact their gap goes from the second largest at 43% in the bargain bin all the way down to the worst at 3% for studs.

A lot of this has to do with how TD dependant tight ends are and how difficult it is to predict TD’s. If a tight end is priced lowly he can reach this goal on just a few red zone targets but if he is priced higher it is extremely difficult. There is one exception to that rule and his name is Jimmy Graham. Last year Graham actually provided H2H value more often than the bargain players but unfortunately he provides poor value at almost the same rate as the other stud players. If you are going to spend for a tight end it should be for Graham but even then it isn’t a slam dunk. Hopefully Rob Gronkowski will stay healthy this year and fall into this category as well.

DEF and PK – For defense and place kicker I am going to simply reiterate what I said above. Spending slightly more on these positions is desirable as long as their values are similar. If they aren’t good values you shouldn’t even be considering them.

To summarize in both H2H and GPPs I would focus on the studs and scrubs method because the “other” players typically don’t score as well and don’t offer as much safety as these studs or scrubs. As for each individual position see the chart below for a quick reference point.

Remember these are just general guidelines and your weekly projections should help lead you to the right split in any given week. For example, if Jamaal Charles is set to play the worst run defense in the league I wouldn’t bench him because he doesn’t fit the proper profile. But if you want to be a long term winner in the daily game most of your lineups will reflect something similar to the layout above.

Follow @SteveBuzzard