Fantasy Ultimate: Reflections on our first go

Posted on October 29th, 2012 by Mark "Spike" Liu
This past weekend there was a massive Fantasy Ultimate contest using player-level stats. 1832 people created their teams of 7, and to our knowledge this was the first contest of its kind in our sport. Frankly, we’re as surprised as anyone that we (The Ultimate Page, Ben Demboski, and Leaguevine) were able to pull this off so I figured I’d write about its origins, the challenges we faced, and also the future outlook for this project. 

Origins of this project

In late August 2012, Lou Abramowski, Leaguevine advisor and creator of The Ultimate Page on Facebook, began searching for developers to build a Fantasy Ultimate application for Club Nationals that would live on top of the Leaguevine API. Given that the tournament was two months away, many thought this was overly ambitious but he pushed hard and recruited a handful of talented people who were interested in contributing to this. 

This would be a 3rd party application built on top of the Leaguevine API. Leaguevine’s involvement with it would simply be to make sure the API worked and that we exposed the right granularity of data to make this easy to build.

There was some initial talk about this on the Leaguevine-developer’s group but it soon quieted down. Lou continued pushing to make this happen by creating wireframes for the project and delegating various other tasks. By the time mid-october came along and the initial prototype still didn’t exist, everyone who knew about this project (including myself admittedly) was skeptical this would happen in time. Everyone was still hopeful, but clearly skeptical.

Enter Ben Demboski, co-founder of Gatherball. The week/weekend before the tournament Ben went to work on building the Fantasy Ultimate site and the Monday morning before the tournament he made a massive code push that convinced everyone that this thing was truly going to happen. 

The next two days, several of us devoted 100% of our time to getting it ready for deployment on Wednesday so people could draft their teams before the tournament started on Thursday. Lou rounded up tons of sponsorships, Ben made huge code improvements based on feedback from the Leaguevine-developer’s group, Leaguevine’s designer Andrew McIntee put in some work for free, I worked on getting the site deployed and making sure the stats system was working, and countless Ultimate players worked on finding folks who would be at Nationals who could take stats to populate this system.

We launched Wednesday mid-day and in under 24 hours, we had 1800 people create their fantasy teams. Thousands of others signed up just to see what all the excitement was about. These people were overloading our servers and we struggled to keep them up on Wednesday, having to refactor code, provision better databases, and increase the number of web servers to try and keep it online. We didn’t yet have Google Analytics set up, but we could sense from the Leaguevine API usage that this thing was going nuts, as this fantasy-ultimate site was making an average of 10 requests/second to our API over the course of a full day with spikes being far higher.

During the tournament (i.e. Challenges)

Now that we had tons of users and excitement, all we had to do was make sure all of the teams took real-time stats to populate the system. 

This never happened.

When I got to Nationals and began talking to teams, all of them knew about fantasy and most had assigned someone to take stats for their team. Finding this one helper to do this was hard for the teams to pull of because there were very few friends, family, or fans at Nationals. Almost everyone there was a player and players had to focus on winning games, not taking stats.

Nonetheless, most teams did find a stats keeper. The problem was that USAU was also short staffed and didn’t have enough volunteers to take the stats they always take for Nationals events. Because of this, USAU handed each team a stats sheet and asked them to take pencil and paper stats for all their games. The teams gave these stats sheets to the stats keepers they lined up, and these stats keepers ended up not taking any real-time stats on Leaguevine.

After the first round or two of having very few real-time stats for teams, it looked like fantasy-ultimate was doomed. 

To salvage this situation, we talked a bunch with USA Ultimate about getting images of these completed paper score sheets and manually entering all these stats onto Leaguevine. After some initial skepticism, USAU jumped on board and was very helpful at providing these sheets for us to use. 

While we were getting these images, a number of amazing volunteers stepped up to manually transcribe these stats sheets into stats onto Leaguevine. Keep in mind here, that Leaguevine has not yet been set up to easily accept post-game stats so these volunteers were simulating games using the Leaguevine-ultistats interface.

Every night after games, we would take photos of about 30 stat sheets, and email them to these awesome volunteers (namely Zach Rabin and Zack Purdy) who would turn those into stats on Leaguevine. 

Through this scrappy work, we managed to get significant statistical data onto Leaguevine which could populate this popular fantasy-ultimate application.

Future Outlook

This application got a huge amount of exposure within the Ultimate community in a very short time and it is clear people want more of this. Tournament-level Fantasy Ultimate is interesting, and various tournaments could use this to generate exposure for their events.

One thing that is clear, though, is that manually transcribing data from paper stats sheets is not the way to go and will not work long-term. It will be crucial for tournaments using this system to ensure that there are people at every field entering real-time stats into Leaguevine. This is the only way this can scale since there are only so many Zach Rabins and Zack Purdys out there who are willing to do the dirty work.

Assuming tournament organizers are able to get people to use Leaguevine-ultistats or other 3rd party stat keeping apps built on Leaguevine, there are a plethora of cool things that can be done with the fantasy-ultimate site: Positive and negative players, taking detailed stats into account (completions, completion %, etc.), and taking defensive stats into account (D’s, defensive efficiency, etc).

Beyond just tournaments there is a lot of potential in Leagues as well. We are working on a partnership with Major League Ultimate that would result in a full season’s worth of stats in 2013. This quantity of stats data spanned over many weeks opens the possibility for more traditional fantasy leagues that have drafts.

The future is definitely exciting for Fantasy Ultimate and it has the right leadership in Lou Abramowski and Ben Demboski to make it amazing. If you are interested in joining them in building this thing, please join the Leaguevine-developer’s group and introduce yourself.

Thanks to everyone involved for making Fantasy Ultimate happen and thanks to the Ultimate community for making it so popular!
  • UltiVillage ran a couple of fantasy competitions based on UPA Nats a number of years ago. Don't remember the exact details, but I believe it was a very similar setup.
    3:50 p.m. on October 31st, 2012
  • Yeah, I vaguely remember those competitions. If I remember correctly, they ran both a team pick-em and a player pick-em competition in different years. I think the player pick-em was only applicable to players who were in the finals. Hopefully we can find enough stats keeping volunteers to bring the community lots more fantasy contests!
    4:48 p.m. on October 31st, 2012
  • Thanks for the nice summary, Mark! And great work by Ben, Lou and everybody else pulling this off! As an outside observer, my main concern about this project has always been the actual stat-keeping. I thought that this fantasy-ultimate project would give a big push to the development of the stat-keeper software, because lots of people will provide feedback about their real-world experiences in using it. Indeed, stat-keeping turned out to be the major bottleneck, but unfortunately, the positive side-effects of having lots of people using it have not materialized. I think that in that area, there is still ample room for improvement.
    10:06 p.m. on November 13th, 2012
(you must be logged in)