Recently I was inspired to do an in-depth research project on what the building blocks were in order to take down a cool mil. That’s right I am talking about the DraftKings Millionaire Maker (Milly Maker). I have played DFS for many years now and have accomplished almost everything I set out to with DFS. Been to a live final tournament, won a big tourney, played the weekly grind to make money regularly at DFS, but there has been 2 things that have so far eluded me. First is a seat at the World Championship – nearly made it several times. And lastly is a top 5 finish in the Milly Maker – best I have done is 14th. The Milly Maker is a beast of a contest. You are playing with about 250,000-300,000 other players for a cool 1 million dollar top prize and it drops off steeply after 1st. The epitome of top heavy. My inspiration for this research came from an article Adam Levitan, a popular DFS analyst, wrote over the summer on the same topic. I was interested in his findings but sought to understand them at a more fundamental and granular level so I could start to use the knowledge in constructing the perfect GPP lineups. For those who are unfamiliar with the term GPP – this is commonly used to reference big prize tournaments with lots of players. While it technically just stands for Guaranteed Prize Pool it is typically the big tournaments that sites like DraftKings denote as GPP.

About Me:

For those of you new folks who don’t know me, my name is Hunter and I have been around with Kevin since this was just year 1 of a blog he had of daily fantasy football strategy.   I used to do a bit more writing back in those days with Hunter’s Corner and various columns on Defense analysis, Vegas line breakdowns, and general player value analysis, but I have not done much writing in the last several years.  Mainly for selfish reasons due to time constraints and wanting to spend my time working on my game and becoming a better DFF expert. I am not sure if I will ever consider myself an expert until perhaps I take down a milly or win first at a live final, which I have been to a couple but have yet to win one outright.  But I can say I am certainly experienced at the game. Note: I said expert and not professional – I do not make a living at this and I work with a reasonable budget on a weekly basis. Typically in the couple hundred dollar range. I am a fan at heart and I play DFS because I truly love football, I love daily fantasy and I love the comradery of our DFS Army group here.  The money I win playing DFS each year is just a bonus.

ANALYSIS:

Here are the 4 topics I set out to understand better about the winning lineups of the Millionaire Maker contests. Unfortunately I only had the results lineups for the last 2017 Season so my analysis is a relatively small sample size, but there is still much to learn, and by analyzing the top 20 lineups in each contest and comparing to the general field we can increase the sample to a more respectable number.

1. Most Used/Interesting Stack Combinations (QB/WR, QB/TE, QB/WR/TE, RB/DST)
2. Ideal GPP Salary Allocation Per Position (Is there a price correlation to winning)
3. Ideal Ownership target % per Position (Is there ideal ownership % by position)
4. Ideal Average Ownership % for Lineup (Is there ideal ownership % for lineup)

Below are my findings based on those topics:

But before I reveal the results a disclaimer: These are my own interpretation of the results of personal analysis I did with data at my disposal, which I obtained from playing in these contests last year and saving the results. Your interpretation may vary. I am also not suggesting anyone make any financially impacting decisions based solely on my research. Use this as you would any other piece of information you find on the internet. There is no substitute to doing your own research and understanding all the data for yourself. All that being said…

Most Used/Interesting Stack Combinations

I wanted to see what lineup construction stacks were most commonly used in winning lineups. Here is a snapshot of the results which I will discuss key takeaways below.

Stack frequency in top 20 Milly Maker lineups

A couple definitions:
– For the REG STK, DOUBLE, TRIPLE, QUAD, and NO STK – I am referring to a QB stacked with either 1 (REG), 2 (DOUBLE), 3 (TRIPLE), 4 (QUAD), or 0 (NO) players or DST from the same team as that QB.
– NOTE: this is different than a “GAME STACK” which typically means a QB and ANY player from that same team and also ANY player from the opposing team.
– Anywhere you see “o” – I am referring to opposing team position like oWR means opposing WR.
– NAKED QB or NO STK means that the QB is not in a lineup with anyone else from that same team.
– The table on the right is the same information as that on the left, but I am specifically highlighting where the TOP 20 used a larger percentage of that stack vs general public (Green) and where they used less (Red). Obviously if there is still a large percentage of LUs using a stack regardless if it is less or not it is still worth consideration.

KEY TAKEAWAYS:

1. Over 90% of all top 20 lineups stacked the QB with at least 1 player from the same team.

In comparison to the overall field where 82% used a QB stack of some sort it is clear that it is a good strategy to always stack. Nothing new here but good to see validation.

2. 76% of all top 20 lineups stacked their QB with a WR from same team.

In comparison to the overall field where 67% used a QB stacked with WR it seems to indicate that the higher placing lineups are more heavily QB/WR than other stack combos, but again all Stack combos where used more for the top 20 lineups.

3. A rather large nearly 40% of top 20 lineups used a double stack and single stacks come in just above at 44%.

In comparison to the overall field where only 25% used a QB stacked with 2 players from same team this is maybe the most telling comparison of all. Fire up those double stacks! QB/WR/WR, QB/RB/WR, QB/WR/TE being all about even but the WR/WR combos having a higher differential from the overall field than the rest.

4. Over 22% of top 20 lineups use some form of Game Stack with the Double Game Stack (QB+2 same team+1 opposing team) being the most common at 12%.

In comparison to the overall field where less than 18% used a game stack and an even bigger margin of less than 5% using a double game stack vs the top 20 rate of 12%. Clearly the higher 40% double stack of the top 20 is coming into play here but I found it interesting that out of all double stacks done more top 20 lineups added in another player from opposing team VS the general field. 30% of “top 20” lineup double stacks were game stacks and only 20% of the general fields double stacks were game stacks. In general though you have a higher percentage chance of having a double stack that is not a game stack placing vs one that is. What this tells me is that if I am doing 4 out of 10 lineups as double stacks then I should at least have 1 or possibly 2 of them as full double game stacks.

5. Over 50% of top 20 lineups used opposing WR, RB, or TE to one of their WRs even if that WR was not in a stack with QB.

I suspect that this occurrence is more of just a random one that people are typically picking players from games with high vegas O/U lines more often than explicitly trying to match up a player with someone on opposite team. Either way there does not seem to be a difference in the percentage of top 20 lineups VS the field. But the general statistic does reinforce the game stack theory above.

6. Triple Stacks while rare in top 20 at under 6% were more common in top 3 lineups at a surprising 11.8% with a third of those being game stacks.

This statistic I am a little worried to trust due to the limited sample size of basically 48 lineups falling into this top 3 category, but that does basically equate to 2 weeks out of 16 that won due to triple stacks and at least 1 lineup each week having triple game stack. That is a significant indication if it were to hold true over more than 1 season. Without more data we can only speculate though that triple stacks might be something that gets your into the top 3 when an otherwise double stack wouldn’t.

7. 0% of top 20 lineups had a QB matched with the opposing DST.

Very occasionally does this situation workout where the DST gets some Punt/KO return TDs and possibly turnover returned for TDs and then the opposing team is playing from behind and has to throw a lot. But apparently it is even more rare that people play it and it makes it into the top 20.

8. Very few lineups had a QB matched with an opposing RB. It was most common to have QB matched with opposing WR and then only somewhat common to use the opposing TE.

TE and WR were used more frequently in game stacks from top 20 lineups than the general field. RBs less.

9. Stacking the RB with the same teams DEF was in roughly 23% of lineups.

This was 5% more prevalent in Top 20 lineups than the general field. 23% vs 17%. This tells me I should try to slightly increase the number of these lineups I do this with. If I was doing 2 out 12 before I would bump it to 3 or 4 out of 12.

10. A surprising 3-6% of these top lineups used 2 WRs from the same team without matching them with their QB.

A higher percentage (6%) of these occurred in the top 3 lineups vs top 20 but perhaps these are due to stacking 2 opposing WRs against a QB stack.

11. For Double Stacks there is a fairly equal distribution between QB/WR/WR, QB/WR/TE, and QB/RB/WR with the latter having the slight edge.

There is not much to say here other than all of these were used about 5% more frequently in Top 20 lineups than in the general field.

12. Less than 10% of all top 20 lineups used a QB naked without stacking with a single player from same team.

This was far less than the general field at about 18% of the lineups of the field using a naked QB. I am 100% fine with limiting my exposure to Naked QB stacks to less than 5% every week.

13. The QB/DEF stack was the least popular stack in the top 20 at under 8%

It was slightly more prevalent by only 1% in top 20 lineups than by the general field, so I would say not a significant enough factor to change any behavior based on.

 

Ideal Salary By Position Analysis

When I started this analysis I wanted to see if there were any patterns to the salary ranges used for winning lineups. I suspected that it was likely fairly randomly distributed and for some of them that was the case, but there were a few surprises that I did not expect. Here is the data results in a couple visual images and my follow-up take on the data.

Above you can see the salary allocations for QB and RB. Let’s take a look at QB first:

KEY QB TAKEAWAYS:

1. The Median QB Salary for top 20 lineups was $6400
2. There was a significant portion of QBs taken at exactly $6400 or $6700 at nearly 23% of all top 20 lineups.
3. Over 81% of lineups spent $6800 or less on QB.
4. None of the top 20 lineups spent over $7900 on QB 

KEY RB TAKEAWAYS:

For RBs, I separated them into the highest priced RB in the lineup from the lowest priced RB in the lineup in order to see how the salaries compared with other lineups. I was interested to see if there were more Studs and Duds lineups or if it was more evenly distributed.

1. The Median High-Priced RB Salary for top 20 lineups was $8100
2. There is over $400 difference between the average of $7676 and the median of $8100 which means the majority of lineups fall above $7800 with a handful of much lower salaries to bring the average down.
3. There is a larger cluster of lineups with high-end RBs in the $7900-$8200 range and another cluster in the $8700-9200 range making up nearly 46% of all top 20 lineups.
4. Only 5% of lineups spent over $9200 on a RB
5. The Median Low-Priced RB Salary for top 20 lineups was $4800.
6. Nearly 50% of all low-end RBs salaries fell between $3900 and $5000, with the most popular being $4500-$4800.
7. There is a 10% cluster of low-end RBs at $5900-$6000.
8. Less than 5% of lineups had all RBs priced over $7400 (all studs)

Above you can see the salary allocations for WR:

KEY WR TAKEAWAYS:

1. The Median high-end WR salary for top 20 lineups was $7300
2. A significant portion at almost 15% occured at $7800-7700 salary. Other notable cluster of 20% at $6500-$6800 and of 9% at $8400-$8500
3. 65% of lineups spent between $6500 and $8500 for a high-end WR and 90% spent at least $5900 or higher on a WR.
4. No top 20 lineup had a WR priced higher than $9400
5. 95% of lineups also had at least 1 WR at $6000 or less and 60% spending $4400 or less.
6. 47% of low-end WRs used in top 20 lineups were between $3700 and $4400 salary with another 13% between $4900-$5000
7. The 3rd/4th WR used in most top 20 lineups was either another low-end WR range as stated above OR a WR in the $5900-$6800 range.
8. Only 6.2% of top 20 lineups spent the minimum $3000 on a WR.

 

Above you can see the salary allocations for TE, FLEX, and DST. Let’s take a look at TE first:

KEY TE TAKEAWAYS:

TE is an interesting position and I was wondering how often a top 20 lineup used a top 3 TE by salary VS a middle pack TE or punt option.

1. The Median TE Salary for top 20 lineups was $3900 which was higher than the general field median of $3600.
2. The Average TE Salary of $4425 was over $500 more than the Median of $3900. Meaning there were a decent amount of High-end TEs taken but the bulk were $4000 or below.
3. 20% of the top 20 lineups used a TE priced above $6800 which was far greater than the field at only 5%.
4. 74% of lineups used a TE at $5000 or under.
5. There was a large price range of TEs between $5100 and $6600 which were only used in about 5% of top 20 lineups. I am calling this the DeadZone for TEs.
6. There was a significant amount of top 20 lineups using min-priced TEs at over 25% using TEs priced $3000 or below which was also a decent amount higher than the field of 16%.
7. General rule of thumb was either go with a top priced stud, go with a min-priced flyer (maybe injury replacement), or go with TE in the $3500-$4000 range.

KEY FLEX TAKEAWAYS:

For FLEX spot I wanted to know what percentage of position was most used in flex and how much people paid for that position. The fact that the player was in the flex wasn’t as important since many people would use studs or or whatever in the flex – so the salaries are worked into the numbers for RB, WR and TE above. So I am mostly disregarding the price point for the FLEX spot and using the relative position salary analysis instead based on what position player I plan to use in my FLEX.

One Big Difference between the top 20 and the field was the usage of the FLEX position. The top 20 lineups used RBs over 62% of the time and the field was 41% for RB using them less than WRs. The field also used a TE almost 5% more frequently than occured in top 20 Lineups.

1. RBs were used in the FLEX in over 62% of top 20 lineups VS 41% in lineups across the general field.
2. TEs still were used in the FLEX in about 10% of the the top 20 lineups, so not an insignificant number, but less frequent than the general field of about 15%.
3. WRs were used in the FLEX in about 27% of top 20 lineups which was far less than the general field of almost 44% of all lineups.
4. It appears that the biggest salary cluster for the flex position was between $3900 and $5600. This isn’t a huge surprise due to salary cap constraints and the prices at other positions used. 

KEY DST TAKEAWAYS:

For DST I was interested in if winners were just going with the favorite DST (highest priced) or if they were digging deep for cheap defenses.

1. The median salary for a DST on top 20 lineups was $3100.
2. There is a fairly even distribution of lineups using DST in the $2900-$3800 range, but that basically covers a large amount of DSTs each week and does not tell us much.
3. It is worth noting that about ⅓ of top 20 lineups used a DST priced at $2800 or less and another ⅓ used DSTs at $3500 and higher. So really that does not tell us much other than there are decent clusters of lineups at the salaries at $2800-$2600 and $3500-$3800.

  

Ideal GPP Lineup Salary Allocations

I decided to use the Medians and the Outlier Salaries listed above to come up with a few “Ideal” Salary allocations by position. In these samples I am using a RB in the Flex for all of them seeing how that was over 62% of the winning lineups. These are just samples to play with and as you can see from the above that there are really a lot of possibilities, but these are in-line with the majority of winning top 20 lineups.

Above you can see some differences between what the Median Salary was by the general field (ALL) and the Top 20 lineups. The couple of big outliers are that the top 20 lineups used far more High/Low priced RB combos VS the field. The high priced RBs typically ran at least $800 more than the fields high-priced RBs and the low were a few hundred less. The TE that were used were typically a bit more expensive by about $300 and the DST were closer to the upper range at $3600.

 

Ideal Ownership Percentage Analysis

I think we all suspect or assume that in order to win a huge GPP tournament we have to get a “little crazy” with our lineups. What I mean is that we have to find that rare low owned guy that goes off for a big game. Some people take this idea and run with it slapping in a bunch of long-shot guys and fail miserably. What I wanted to see was if there was a general pattern to the ownership percentages by position and as a whole for the winning top 20 lineups.

Above are the average and median ownership percentages for each position.

As you can see the RBs overall were the one position that was most highly owned. That is most likely due to the limited number of viable RBs each week, but I expected the same to hold true for QB and TE as well. Above I don’t see anything that sounds out too crazy to me other than the bigger differences between average and median for WR and RB – which just means that in top 20 lineups we are more likely to find a handful of very high priced RBs and WR than the very low priced options.  It was somewhat interesting that top 20 lineups actually had a bit higher ownership percentage of players than the general field, which reinforces my assumption that people real do play to many crazy plays.

 

In this table above I break down the percentage of the top 20 lineups which had at least one player that was either greater than or less than the given ownership percentage by position.

KEY OWNERSHIP TAKEAWAYS:

1. Over 92% of top 20 lineups had at least 1 player under 5% owned. The vast majority of these were WRs with about 70% of lineups using a WR that was under 5% owned.
2. 89% of top 20 lineups also used at least 1 player that was over 20% owned and 44% used a player over 30% owned.
3. 64% of top 20 lineups used a RB that was over 20% owned and 37% of the lineups used a RB over 30% owned.
4. It seems logical that using a WR over 25% owned may have a lower top placement occurrence. RBs seem far safer to use up to about a 32% ownership before they start to have diminishing returns.
5. Only 8% used a QB over 20% owned, so a rule might be to stay away from highly owned QBs, TEs and DSTs.
6. Finally the average ownership percentage of the entire lineup for top 20 lineups was 12.8%. If you find that your lineup is significantly over or under this average percentage then you may want to consider replacing some players.