I won the science fair in fourth grade. I built…well…my dad built a water distillation unit and it put to shame a bunch of papier-mâché volcanoes. Since those fateful elementary school days I haven’t done any physical science experiments but I’ve done a bit of experimenting in the space of daily fantasy sports.
Now, I’m not claiming that I’m akin to Bill Nye the Science Guy, but I enjoy cooking up experiments to test new strategies and analyze macro trends. The problem is, all of these experiments have been founded on my own hypotheses. I’ve been testing or tracking trends and analyzing the results on my own accord – simply drawing from one paradigm, rather than shifting it or pulling from others.
I wanted to change that.
Tap into different minds and new paradigms and create a unique, but viable strategy for large field GPPS on DraftKings.
By tasking smart people to provide me with individual criteria they felt would be successful or trends they wanted to have tracked and then meshing them all together to create GPP teams.
If this seems like me throwing together a crazy hodge-podge of ideas, it’s because it is. But let me pose it to you as a strange analogy – I’m like a toddler trying to jam every shape I have into one of those plastic cubes with holes cut out for particular shapes. If after a few tries, the shape, or the trend doesn’t fit, I’ll pick up a new one and try again until it all falls into place.
Of course, the analogy falls a bit short because the toddler controls his destiny and even if I do come up with an “optimal” set of trends I might not win – but I’m still going to try.
Week One (which was actually week two of the NFL season)
I asked DailyRoto’s own, Mike Leone (@leonem), Drew Dinkmeyer (@DrewDinkmeyer), Colin Drew (@drewby417) and the mythical DFSalbatross (@DFSalbatross) to each provide me with an individual criteria for this week.
This is what they came up with:
- Colin Drew – pairing of two pass catchers with a quarterback priced at less than $7,000.
- Mike Leone – wide receivers from the $7,000 range.
- Drew Dinkmeyer – running backs priced at $4,500 or below with an average of two receptions per game.
- DFSalbatross – tight end from a game with a total equal to or greater than 45.
- ME! Logan Hitchcock – WR in the flex at less than $5,000.
Despite how difficult it seems I was able to create four lineups. I put them in the $3 entry, $250K Play-Action on DraftKings.
Before I analyze the individual trends and the lineups as a whole, first let me point out a few of the flaws of the experiment that I’ll be cleaning up before heading into next week.
- Alleged Miscommunication
This is proof of the criteria that Mike Leone passed to me. Therefore, when I began to mesh and build the lineups for this week, I filled all the WR slots on DraftKings with wide receivers in the $7,000 range. Mike though, says that this isn’t what he meant. Apparently he only wanted me to pick one WR from the $7,000 range. Can you fault me for using three? The guy said wide receivers! Plural! This didn’t make lineup construction impossible, but it did make it much more restrictive, keeping me from building further iterations using the trends and shrinking our potential for data collection. Mike has since admitted his mistake and will be tweaking his trend for next week.
- Sample Size
Four lineups on one day will never be enough to truly analyze these trends as well as they could be analyzed. I know that, but yet, I knowingly entered only four lineups in the $3 GPP instead of expanding my entries and increasing my potential for data collection by entering more teams at a smaller entry level. Next week, I’ll be moving to the $1 GPP and creating more teams, hoping to draw even more information on the trends and the success of the merger.
Tracking the success of these trends is somewhat unfair – especially since they tie to each other and restrict the player pool at each position (particularly if a trend affects more than one player on the roster) but, I’m going to track them anyway. Using a made up statistic called APPA (average points per player affected).
For example, Colin Drew’s trend affects three players on each team. By totaling their points and finding the average, I can see how large a footprint his trend left on a team on average.
Here are the findings:
|<$7,000 QB + 2 Pass Catcher Pair||15.85|
|<$4,500 RBs with > 2 catch avg.||14.01|
|WRs between $7,000-$7,900||13.11|
|TE in game with total > 45||19.48|
|WR in Flex < $5,000||15.23|
But…as you are well aware about the nature of daily fantasy sports the players that garner the highest salaries, are often better players. Therefore, it’s unfair to evaluate the trends solely based on points produced as some of them are eating more of the total salary then the others (looking at you, Mike Leone).
So, DFSalbatross did some weird math trick thing (this bird is smart) and produced this formula to neutralize salary based on what we’re looking to achieve out of a player:
TRUE EFFECT = APPA – (3*(AVG. SALARY OF THOSE AFFECTED)/1,000 + 6)
And….one more table.
|<$7,000 QB + 2 Pass Catcher Pair||5.53|
|<$4,500 RBs with > 2 catch avg.||7.08|
|WRs between $7,000-$7,900||-3.57|
|TE in game with total > 45||11.23|
|WR in Flex < $5,000||9.23|
I don’t have anything to compare these to yet, so my analysis on their success will be non-existent, but I think you can make a note of how they compared to each other this week. I won’t mention any names.
Determining the success of a GPP lineup can only really be measured in two ways – by point total and by the cold hard cash it brings back (or for the purpose of this experiment, total ROI).
Again, this week I have nothing to compare it to – but here are the baselines.
|Week||High Score||Low Score||Avg. Score||Cash Line||Total ROI|
*Week one of this experiment is actually week two of the NFL season.
It’s too early to jump to conclusions and despite the horrific result for Mike Leone in week one and the overall detriment he was to the team, I won’t push him out. I’ve given all of the contributors the ability to tweak or switch their criteria for week three and will update all evaluation metrics next week.