Article in Companies, Startup, Programmable Web, Technology, New York City categories.
Fantasy Football Without the Fantasy: NumberFire
The NumberFire team has been busy making an app meant to render this experience infinitely better.
The NFL is a cash cow. The league’s ticket and merchandising sales are through the roof, and its TV ratings have only gotten higher with each passing year. With all of this interest, fantasy football has only grown larger as well. While it’s difficult to get an exact number, approximately 50 million people play fantasy football in America each year, a staggering amount. While football fans have been busy building and managing their virtual teams the NumberFire team has been busy making an app meant to render this experience infinitely better.
In fantasy football, participants form online leagues with their friends, colleagues, or strangers. The leagues - hosted on sites like ESPN.com, Yahoo, and NFL.com - hold, on average, between 8 and 12 teams. Each league holds a draft before the season, during which participants select somewhere around 15 NFL players each for their teams. Players are chosen primarily based on their projected statistics for the upcoming season. The players that people think will have the best stats are drafted first. Team owners then choose 9 of those 15 players to start - only these starting players can score points for the team - and leave the rest on the bench, and players can be added and dropped from the team throughout the season. For each game, teams are required to start a quarterback, two running backs, three wide receivers, a tight end, a kicker, and a whole team’s defense. Teams then face each other, and points are awarded based on statistics like yardage and touchdowns, field goals for kickers, and interceptions and sacks for defenses. Teams face each other head-to-head each week, and the team with more points wins. Records are accumulated, and the season concludes with playoff rounds and a champion is crowned. Just watch The League on FX if you have any other questions. And even if you don’t, watch The League - it’s great.
Over the years, league members adopt different strategies aimed at forming the best possible fantasy team, and most include different biases toward and against players they’ve come to love or hate. NumberFire, created by Nik Bonaddio, is attempting to take those biases out of the fantasy equation. With his company’s system, he’s trying to apply the principles of Wall Street analytics to the competitive world of fantasy football. His aim is to base all fantasy maneuvers on mathematical fact instead of the opinions standard to sports reporting. I spoke to Nik recently to discuss how his system works, how his company can grow in the future, and some tips for fantasy football domination.
Jeremy Rothschild: How did you come up with the idea for NumberFire?
Nik Bonaddio: It came from my playing fantasy football for 20 years or so. Being the engineering type, I’d get frustrated reading articles by [ESPN.com columnist] Matthew Berry or the writers on Yahoo, and it would just be opinion, opinion, opinion. I’m naturally skeptical of when people say something without any facts to substantiate it. I want to see logic, but all I was seeing was opinions. It didn’t make sense to me. Sports are about numbers. We have stats for everything in every sport now. Why are all of these columns about fantasy players based with opinions? If numbers are available for comparing players, it doesn’t make sense to do anything but look at that data. I thought that there’s a niche here for people like me and a big group of fantasy football players. Everyone wants to make the right trades, and pick up the right guys for their teams. It made sense to build that system.
JR: What are some ways that fantasy players can use the site?
NB: We project the stats for every player and every team. You have certain guys on your team, and you want to see who to play that week. We’ll provide accurate projections instead of an opinion. Our forums have a question and answer format, with narrative questions intended for certain people. You can ask certain questions and tag in ways to get answers. For instance, I’m a Steelers fan. So I may be following topics like [wide receiver Hines] Ward, [quarterback Ben] Roethlisberger, and other Steelers. I’ll have a newsfeed for the players that I follow. If someone has a question about one of the players that I follow, I’ll be able to answer that question. This creates a community of experts about certain players, and everyone can get their questions answered by those who know more about them. Those answers are distributed within the site. So the site becomes half crowd-sourced with these kind of answers, and half totally data driven with our algorithms for projections. People want more information, and they want answers to their questions. We have the data analytics and the social tools to do both.
JR: What are some of the factors that you take into account to make your projections?
NB: The majority of it comes from historical data. If I wanted to know how [Giants QB] Eli Manning would do this week, I would look into the database and find internal statistics. The system looks for similar players from similar teams that have faced comparable defenses. Our algorithms see what happened in the past, and it gives out scores based on accuracy. This turns into a giant weighted average and an aggregation of lots of different factors, and our specific projection is based on all of that.
There have been so many NFL games to draw data from that history will repeat itself. We’re using the logic that you can figure out how a player will do by looking at what statistically similar players have done in the past against similar defenses. Of course, we have offsets for weather conditions like rain that make certain stats less attainable, and also injuries. We have formulas for players that are listed before the game as questionable or doubtful. This is all completely data driven by our algorithms, and there’s no opinion analysis. If you wanted to, you could find a fantasy opinion from anywhere. Everyone has their own biases, and we want to make it analytical and mathematical. We tend to be right about 60- to 65-percent of the time. Our projections are bereft of any bias or opinion.
JR: Do you personally make the projections?
NB: The algorithm does. My role is to start the script that starts the algorithm. We have a stat guy that creates these algorithms. There’s no human editorial process to it. It’s automatic and mathematical.
JR: Have you encountered any backlash when projections are off? You couldn’t have seen [quarterback] Tom Brady (left) exploding like he did in week 1 against the Dolphins.
NB: We sometimes get emails with complaints, and it happens. It’s not fun. Everyone thinks they’re the expert in the fantasy world. If you get a projection wrong, you’ll get hate mail. It’s just the way it is, and you deal with those expectations. I think of it like being the house at a casino; gamblers may win a hand or two, or even a bit more, but the house will always win.
We’ll be wrong sometimes; nobody can predict Brady for 500 yards, in the same way that no one can predict Donovan McNabb throwing for 30. We can just do our best to make an accurate prediction over the long run, so users can see we make rational and scientific sense versus what else is there.
JR: What does the premium section of the site offer that the other parts don’t?
NB: The free part is like vanilla ice cream. It’s good. You can definitely have a good dessert with it. The premium is your extras; the hot syrup, caramel, jimmies. It’s the added stuff. The free part is all fantasy-based, player analytics. For the premium, we have a handicapping section, where we use the same logic for picking games instead of for fantasy teams. We also offer our fantasy tools. We have a roster analyzer, where you tell us who’s on your roster, and we’ll analyze it, look for weak areas, and recommend ways to improve it. It gives you a stronger sense of your strengths and weaknesses. We’re working on another tool, a trade analyzer where you put in a trade offer and ask if it's fair or not.
JR: Let’s talk fantasy. Which players this season will gain more points for a team than people expect? What strategy do you use to find those players?
NB: I try to apply stock market principles to fantasy sports. I’m always looking to buy low and sell high, based on what’s happened in the past. For Cam Newton and Steve Smith [Carolina Panthers who had high scoring weeks in the first week of the season], I’d be wanting to sell high. Smith wont put up 20 points per week (a high number). Guys like [Chiefs wide receiver] Dwayne Bowe and [Browns running back] Peyton Hillis won’t score four points per week (a very low number), on the other side of it. What you’ve got to do is buy low and sell high, without over-analyzing too much.
JR: And what about the opposite scenario - a player that you expect to disappoint, relative to how highly he was drafted?
NB: [Patriots WR Chad] Ochocinco. People thought that he was going to go to New England and become the new Randy Moss [New England’s top wide receiver for years], and so far that hasn’t happened. It’s hard to say since it’s only one week. People thought he’d take over for Deion Branch as a top Patriots receiver, but that didn’t happen at all. We saw lots of Branch and very little Ochocinco on Monday. It looked like the Patriots designed formations that just didn’t include him as part of them. I wouldn’t drop him yet but it’s certainly troubling.
JR: What type of players are the most valuable on a fantasy team by your calculations? What might be the unexpected key to success?
NB: The key to all of fantasy football is separation. Guys like Tom Brady and Aaron Rodgers are obviously great, but there’s not that big of a difference between the amount of points they get and what the other top guys get. A replacement for them wouldn’t result in that big of a points drop off a lot of the time. The same can be said for some of the popular running backs and wide receivers out there, since there are so many viable replacements. You need to look for separation. I love Antonio Gates (above) in particular because he gives you such a big difference at the tight end position, since there really aren’t any reliably great tight ends besides him. The difference between a top QB like Rodgers and a guy that isn’t quite at the same level is much smaller than the gap between Gates and the mid-level tight ends. There’s way more top tier wide receivers that are interchangeable than there are tight ends, so that makes Gates so valuable. There’s a huge difference between Gates and the second or third best tight end, let alone the sixth or seventh best.
JR: With that in mind, to those fantasy players that drafted QB Peyton Manning, who was injured last week, who would you suggest as a replacement?
NB: [Bills QB] Ryan Fitzpatrick (left), he’s a good pickup. I’d stay away from Cam Newton, since one week doesn’t say too much about a player. I’d look at [Chiefs QB] Matt Cassel. He’s a lot better than (his poor outing in) week one showed. I think someone might drop him, or be willing to trade him on the cheap. Matt Schaub and Josh Freeman, too. The key is obviously keeping up in the news, but also knowing the values of players. Newton’s will never be higher, and Cassel’s will never be lower. I’ll have higher economic value trading for Matt Cassel now than I will later.
JR: Which site do you think is best for league hosting? ESPN? Yahoo?
NB: To me they’re all pretty much the same. I have problems with each one. None are perfect, and they all seem like they’re just there to be there.There hasn’t been any innovation. I’ve always played on Yahoo. I’m a Yahoo guy. It’s more that I’ve always just been there than loving it. I’m interested in FanDuel, which has whole leagues in the course of just one week, but I’ve never really used them that much. People are creatures of habit with where they host their league.
JR: Are there any fantasy writers that you read regularly?
NB: Yes and no. I do check out whatever anyone else is talking about, but again, any time I read these articles, it almost makes me angry because it’s so opinion based. Why aren’t statements qualified? Just being a writer for ESPN doesn’t make me think, “Oh, okay, he must be right.” Show me the substance. The writers are good at what they do, and they don’t stink, but it’s not the kind of information I use to make decisions. I wouldn’t buy a stock because my friend likes it. I want to see the company’s financials and their reports more than just word of mouth.
JR: How do you see the company growing in the future?
NB: Right now we’re in a good spot, because people want this information. We got to 10,000 users quickly. We have a lot of really good partnerships with [popular sports blogs] SB Nation and Bleacher Report. Short term, we’re gonna branch out into other sports, particularly March Madness and baseball. Long-term we want to work with different media sources, like ESPN and Sports Illustrated. We want to give them access to our technology to help their writers write better articles. We hope it can empower writers to have data with what they write, so that I’m not looking for statistical information when I’m reading an article. If we go to ESPN and say that we’ll help them write these articles, and give them context, to us that’s a pretty strong selling point. ESPN makes a ton of money, and we see an opportunity to be a business that’s visible and successful within that.
JR: Any advice you’d give to startups?
NB: If you’re gonna do this, make sure you’re doing it for at least three to four years. If you go into it thinking “I’m gonna sell,” that’s not the right approach. It only makes sense if you’re really committed to giving a large chunk of your life to it. It’s a very slow process, and it’s always longer than you think it’s going to be for all aspects. One of the things about TechCrunch and these tech rags is that they show all of these success stories and for every single success there are 15,000 founders that are struggling anonymously. People are seduced by the crazy stories and good news without knowing how much of a grind it is. Be sure that you’re ready for what’s coming. It’s a haul. There are no shortcuts.
JR: You’re a Steelers fan. What happened in Week 1? (Note: the Steelers were humiliated, 35 to 7, by their rival, the Baltimore Ravens)
NB: I wish I knew! Sometimes, you just go into a game, get punched in the mouth, start feeling bad about it, and it’s just a ball rolling downhill. You get involved in those games sometimes where nothing right happens. You lose your focus and it’s just not your week. It doesn’t mean you’re necessarily bad.
JR:I couldn’t set the Vegas betting line high enough for the Seahawks game this Sunday.
NB: If I’m [Steelers coach Mike] Tomlin, I’d say they embarrassed you. You need to get your respect back. The league is laughing in your faces. You need to go in and just decimate Seattle. Run up the score. Hurt them. I certainly hope that happens.
JR: What's your Super Bowl pick?
NB: We liked the Packers and Patriots before the season. It’s not exactly going out on a limb, but they look like the two best teams.
JR: If you had to pick one bar to go to in New York City for one night, what would it be?
NB: I wish I had a good answer; I’m not a huge nightlife guy. I had a pretty crazy college experience. I’m an old man in the startup world. The Meat Packing District, as much as I don’t like it, when I’m in the mood for going there, I have a good time. I like Provacateur in the Meat Packing District to go all out.