In light of the recent scrutiny surrounding daily fantasy sports sites, I wanted to shift the focus from the legality of the sites to the popularity of such games. So, to explain why I took the time to even complete the project I am about to share with you, let me first provide some background.
To start, I am a junior basketball player for New York University. Unfortunately, in the beginning of the season I was sidelined with a foot-injury that would not allow me to play for 2 months. As a result, I missed my team’s first-semester games. We won all five of our games (all non-conference) and we are currently ranked 12th in the nation.
During warm-ups in one of our games, one of my teammates posed the question to me, “What do you think my value would be on DraftKings for tonight’s game?” I assured him that his value would well exceed $10,000 but I had no solid statistical information that supported my answer. So, I decided to take his question and run with it.
The following project is based off of data I collected from NYU’s official athletic website and RotoWire’s database that shared DraftKings salaries for the night of December 16, 2015. (http://www.rotowire.com/daily/nba/optimizer.htm?site=DraftKings)
When laying out the following information regarding salary value in respect to each player, let me be the first to tell you that each and every one of my teammates have equal value to our team’s success. Of course, due to the nature of the sport, one player is going to score more than another but you must understand that basketball is a team game. Each one of my teammates is vital to our team, whether it is getting in for one minute and blocking a shot, checking into the game for 10 minutes to lock down an opponent’s best player, or being supportive on the bench.
Now, diving into what I found.
I downloaded all of the individual box scores into EXCEL and utilized DraftKings scoring method as a function to calculate my teammate’s overall fantasy output for the season. (Pictured below are both DraftKings official system for scoring a player’s nightly performance and a breakdown of NYU players’ box scores with their fantasy output bolded in the far right column.)
I then separated the data to specifically analyze players’ performances on a game-to-game basis. In order to narrow the focus of the project, I only analyzed players who had played in more than 75% of our team’s games. In retrospect, this step was somewhat unnecessary to do since I already had the aggregate fantasy output, but my goal was to dissect the numbers and see the highs and lows over the small sample size (5 games). The aim was to uncover any breakout performances in a certain game that yielded a monstrous fantasy output.
1 Ross Udine
2 Jay Murphy
3 Doug Gertner
10 Tony Bai
11 Max Ralby
13 Evan Kupferberg
20 Brad Lahens
25 Patrick Burns
34 Costis Kontikas
Chart of All Players in Each Game
In terms of determining a salary value, which was the main objective of this project, I collected: (1) the entire list of NBA players playing on the night of December 16, 2015; (2) their salaries according to DraftKings; and, (3) the average amount of fantasy points they had earned over the course of the 2015-2016 season. Why did I pick that specific date? December 16th was a Wednesday, which is typically the day of the week where a majority of NBA teams play. Meaning that there would be a larger sample size, or pool of players. By having a larger sample size it allowed for me to more accurately cross-reference the data between DraftKings and NYU.
One point to note is that DraftKings calculates fantasy points simply based off of box scores, and employs an arbitrary scoring system. For example they value an assist greater than a rebound, as well as a block or steal at twice the value of an actual point scored in a real basketball game.
Both DraftKings and FanDuel sites use algorithms and run various calculations in order to set a player’s “salary” for the night. Although I do not know the exact formulas or algorithms that the two sites use, I took two approaches to try and discover what the values of my teammates would be on a given night.
In my first attempt, I plotted all of the data I had gathered regarding average fantasy points and nightly salaries. I used average fantasy points as my explanatory value, or “X” value; and, used the NBA players’ nightly salaries as my responsive variable, or “Y” value. Ultimately, I charted average fantasy points against salaries.
What I originally thought I would find was a scatter plot that depicted a relatively linear line, but that was not the case. Since I calculated the correlation between the two variables (X & Y), I was also able to figure out the formula for the best-fit line.
y = $4,562.45 + (0.91851948 x)
However, that formula did provide the answers I was searching for. Due to the fact that the scatter plot (charting NBA data) in reality represented more of an exponential curve, determining the best-fit line formula for a linear correlation did not meet my standards in answering the original question. As for what exactly I am referring to, I attached a table that shows these results when applied to my teammates.
The formula does not account for the salary to be exponentially increasing as the average fantasy point total increases as well. The increase in value calculated based on a linear correlation formula will only increase as so much as the total fantasy output increases. For example, if Doug Gertner increased his average fantasy output by roughly five points then he his salary would uniformly increase by five dollars ($4570 to $4575). However, according to DraftKings Doug Gertner’s 7.85 fantasy points would yield a salary worth $3,100. Yet, if he increased his average fantasy output by five points to 12 fantasy points, then the exponential correlation would bump up his salary by $500 to $3600. This latter example leads me into my second approach.
With this approach, I used the scatter plot depicting the exponential relationship between DraftKings salaries against average fantasy points as my basis for finding NBA players whose average fantasy outputs matched with that of my NYU teammates. The scatter plot is pictured blow.
The second approach I followed produced much more satisfying results. Cross-referencing a scatter plot that is charted based on DraftKings’s real data samples and exponential relationship between both variables (fantasy salaries against fantasy outputs), allowed for me to be much more accurate in calculating what my teammates’ salary would be if DraftKings included a section for D-III college basketball. Although DraftKings may have bigger problems in NYC at the moment with New York Attorney General Eric Schneiderman ordering a cease and desist of daily fantasy sports businesses, I am here to help when they decide to expand their business to D-III basketball if they are able to get past Schneiderman.
Regardless, the results for each of my teammates is as follows:
1 Ross Udine (31.7 avg. fantasy points) = $6,900
2 Jay Murphy (11.5 avg. fantasy points) = $4,100
3 Doug Gertner (7.85 avg. fantasy points) = $3,100
10 Tony Bai (12 avg. fantasy points) = $3,600
11 Max Ralby (23.3 avg. fantasy points) = $5,000
13 Evan Kupferberg (41.1 avg. fantasy points) = $8,500
20 Brad Lahens (10.7 avg. fantasy points) = $3,100
25 Patrick Burns (23.55 avg. fantasy points) = $5,000
34 Costis Kontikas (31.65 avg. fantasy points) = $6,300
Again, let me reiterate, these numbers do not accurately represent the true value that my teammates bring to our team and collective success. The salaries reflect only the product of an exponential relationship based on a hypothetical total (average fantasy points) generated by DraftKings nightly algorithm.
As seen by the comparable players listed under each NYU player’s name, I only selected NBA players that play the same, or similar, position. Since Jay Murphy and Max Ralby also are primary and secondary ball handlers within our team’s offense, they take on some of the point guard’s responsibilities, so I included comparable players like TJ McConnell for Jay and Tony Parker for Max. I followed that same methodology when drawing comparisons for Patrick Burns. Due to Patrick being an absolute man-child on the floor and being able to play the wing and post positions, I included one SF and two PFs that were producing comparable fantasy outputs. And, when it came to players like Ross Udine and Doug Gertner I only included one comparison because there was one player at each of their respective positions that had nearly the exact same output (difference being less than 0.1).
An aspect of the data that is a bit hidden is the discrepancy in minutes played in an NCAA college basketball game (40 minutes) versus that of an NBA game (48 minutes). For that reason, I believe college players are put at a disadvantage for achieving extremely high outputs. Furthering that point, since DraftKings salaries seem to exponentially grow in relation to production and fantasy scoring output, the more time players have to produce the greater the fantasy output would be and the more expensive the salaries would become.
I believe that my teammates’ fantasy outputs are very conservative estimates because NCAA basketball games are 40 minutes in length rather than 48. However, going off of the listed salaries for my teammates, guys like Brad Lahens, Tony Bai, and Doug Gertner are wildly undervalued. The full impact those players have on our games does not always show up in the box score.
At the end of the night, regardless of what DraftKings arbitrarily values, each of my teammates are excellent players and each play a key role in our team’s success. There is no way NYU would be ranked in the top 15 in the country, if we were missing just one player or one coach from the equation. What makes basketball such a great sport is that it’s a team game, and we win as a team and lose as a team. All these fantasy numbers and salaries are simply hypothetical – meaning all numbers should be taken with a grain of salt.