David Does Daily (Fantasy): Historical Scoring Data By Position

This is the fifth installment in David’s diary about playing one-day fantasy football for the first time. All other posts are listed in this blog section.

Last week, I read a couple Daily Fantasy strategy books written by Jonathan Bales, based on the recommendation of reader Tony:

Yes, the titles sound kind of self-helpish, but I found them very enlightening. Bales doesn’t make a bunch of generic, unsupported claims about daily fantasy strategies. He makes specific claims and backs them up with actual data.

So, I did two things:

  1. The data in Bales’ books applies to DraftKings, so I pulled some similar numbers from our database, but using the FanDuel scoring system. Specifically, I checked out scoring variance by position. That data is posted below, after my weekly recap.
  2. I tried to apply some of the advice and strategies I learned from the books.

And … it was my worst week yet, so I’m throwing the books away.

(I’m kidding. I actually think the books have good information in them. I’m pretty sure I can chalk the bad week up to variance and other factors, not the book-related choices I made.)

Week 4 Results Recap

Note: The links in the Lineup Notes field take you to that specific lineup.

I changed the format of the chart a little this week. I entered about a dozen head-to-head matchups using two lineups, and it just clutters up the chart to list all of those individually. So I’ll be grouping the contests by lineup.

Bankroll Entering Week 4$236.72
DayContestFeeLineup NotesResultWon
Thu50/50 League$2#1: use TR Picks / Rotowire + NF optimize97.82 (85 of 100)--
Thu50/50 League$2#2: use TR Picks / Rotowire opt + no dupes from #194.78 (80 of 100)--
Thu$30K NFL Snap$2#2: use TR Picks / Rotowire opt + no dupes from #194.78 (11707 of 16460)--
Thu$30K NFL Snap$2#1: use TR Picks / Rotowire + NF optimize97.82 (10818 of 16460)--
Thu$2500 NFL Mini Dive$1#1: use TR Picks / Rotowire + NF optimize97.82 (1466 of 2169)--
Sun50/50 League$2H2H #1: Pay for QB/RB/Graham. Possession WR122.84 (46 of 100)$3.60
Sun50/50 League$2H2H #2: Sub Gates for Graham, more $ on WR96.84 (86 of 100)--
SunHead-to-head (x8)$8H2H #1: Pay for QB/RB/Graham. Possession WR122.84 (5-3)$9
SunHead-to-head (x4)$4H2H #2: Sub Gates for Graham, more $ on WR96.84 (0-4)--
SunSalary Cap 60k$2GPP #1: CHI QB/WR + GB RB, pay for DEF90.04 (90 of 100)--
SunSalary Cap 60k$1GPP #2: NO QB/TE + DAL RB, pay for TE113.6 (9 of 20)--
SunSalary Cap 60k$1GPP #3: ATL QB/WR + MIN RB, pay for WR107.22 (74 of 100)--
Total Net Winnings-$16.40
Sign Up Bonuses Earned$1.40
Bankroll After Week 4$221.72

Overall, it was a pretty terrible week from a bankroll perspective. None of my Thursday lineups won. None of my tournament lineups won. In fact, out of seven different lineups I entered, six of them threw up goose eggs.

The lone bright spot is that the lineup that cashed (and turned a profit) is the one that I actually spent a bunch of time constructing. I essentially spent all morning Sunday doing research and setting up that first lineup, then whipped out the other four Sunday lineups in the last 15 minutes before the day’s slate of games started.

So, on my to-do list for next week (again): stop waiting til the last minute.

The TeamRankings Private Tournament

As I do every week, I also entered the TeamRankings private tournament using the screen name teamrankings, as opposed to my personal account, tr_david.

This week, rather than use the same strategy I used the first few weeks (remove players from teams where we had a strong Under pick or where we we strongly favoring their opponent against the spread, then try to optimize a lineup using the remaining players), I simply used my personal top 50/50 lineup. That lineup finished 52nd out of 167 players, easily in the money.

If you include the results of the TeamRankings private tournament, my top lineup went 2 for 2 in 50/50’s and had a 5-3 record in head-to-head contests, for a profit of $12.60 from $20 in entry fees. Not bad.

Historical Scoring Data By Position

The main strategies I used in constructing my top lineup were derived from reading the Bales books, and also from diving into the TeamRankings database to take a look at  how FanDuel scoring varies by position. I’ll post that historical data first, and then review the main findings.

I didn’t have time to do an exhaustive review of scoring data, so I had to focus on one topic. Bales published data about which positions had more or less week-to-week fantasy scoring consistency using the DraftKings scoring system. I decided to take a look at that same issue, but using FanDuel’s scoring system. I focused on only the top players, in an effort to figure out at which positions it made sense to spend money for “studs”.

To do that, I pulled the top 50 player-seasons for each position, from 2003 to 2013 (those are the years we have full player data for). For example, the top 5 tight end player-seasons in that time period are:

  1. 2011 Rob Gronkowski
  2. 2011 Jimmy Graham
  3. 2013 Jimmy Graham
  4. 2009 Dallas Clark
  5. 2004 Antonio Gates

I then took those 50 player-seasons and looked at the scores in individual weeks to come up with an average scoring distribution for the top players at each position. I did not include games from Week 16 or 17, to avoid the impact of players resting in advance of the playoffs. Here’s what the distributions look like:

FanDuel Weekly Scoring Distribution By Position, Top 50 Player Seasons

There are some obvious trends here, so let’s discuss.

Data Takeaways & Top Lineup Strategies

As mentioned above, I used this info (most of which simply confirmed things Bales wrote about) to craft my top lineup this week. Here are the main strategies I used in that lineup, and why. Remember that it was a head-to-head and 50/50 lineup, so consistency and a high floor are the goals.

Pay Good Money For The Most Consistent Positions: QB and RB

In a 50/50 or a head-to-head matchup, you don’t want your expensive players dropping the ball (literally and figuratively). You’re not looking for boom-or-bust plays, you want consistency. Statistically, you’re more interested in a high median than a high mean.

As you can see from the chart, not only do the top wide receivers have a lower average score than quarterbacks or running backs, but they have a long tail on the right side of the distribution. That may imply that their points tend to come in bunches — a few high scoring games mixed in with mostly lower scoring games.

The QB and RB curves, on the other hand, aren’t as skewed. If anything, the QB graph actually has a long tail to the left, meaning that occasionally a stud QB will lay a stinker, but for the most part they are consistently in double digits.  In fact, 47% — roughly half — of the weeks in this sample had scores between 15 and 25 points, and 74% of weeks scored 15 points or more. The next closest is running backs, with 66% of weeks scoring 15 points or more.

I also took a look at the standard deviation of the weekly scores of the top players at each position, and how it compared to the average (STDEV/AVG is called the coefficient of variation). Quarterbacks were the most consistent by this measure:

  • Tier 1: QB 0.35
  • Tier 2: DEF 0.41, RB 0.43, K 0.43
  • Tier 3: WR 0.52, TE 0.56

So it seemed to me like focusing on paying for high-scoring players at QB and RB was a smart idea. Obviously I wanted to keep value in mind as well, but all else being equal, I’d rather go with the expensive QB and cheap WR, rather than the expensive WR and cheap QB.

Don’t Waste Money On Kickers

There’s just not much advantage to gain by using by using a top tier kicker. Remember, we’re looking at the 50 best player-seasons here, yet the most common result is between 5 and 10 points. Finding a cheap kicker with a good matchup seems like a much better strategy than paying top dollar.

But DO Spend Money On The Very Top Tight Ends

Based on the chart, the same anti-kickers argument looks like it might apply to tight ends. After all, the most common scoring bin for the top 50 tight end seasons was also 5 to 10 points. The difference, though, is that the tight ends have a long tail to the right.

I don’t have extensive data to back this up, but I believe that tail is partly because the very top tight ends (think Jimmy Graham) are so much better than the 50th ranked tight ends that the distribution gets skewed. For example, if you look at only the top 5 tight end season listed above, their most common weekly score was between 10 and 15 points. So I believe paying a bit more for a top tight end may be worth it.

Focus On Possession Receivers Rather Than Big Play Receivers at WR

Actually, this one came directly from one of the Bales books, not from the data I pulled. Bales suggested that on a partial PPR (point per reception) site like FanDuel, focusing on receivers whose value comes more from their high number of catches than their yardage or touchdowns could be a smart play. The idea is that touchdowns are a much rarer even that receptions, so the points will be distributed very unevenly — maybe 0 points from TD’s for several games, 6 the next, maybe 12 in another game. On the other hand, the number of receptions a player makes should be more consistent — like 2.5 points in one game, 6 the next, 3 the game after that.

So when deciding between wide receivers this week, when there were close calls I went with the players who had more receptions and fewer touchdowns projected.

All in all, I think these strategies seem solid. They worked out well in the top lineup I made, and I’ll continue to apply them in Week 5.