The last of the Week 1 NFL games will have ended about a half hour ago when Mike Talas, probably still wearing his Dolphins hat, sidles up to his computer in the office of his South Florida home. Much of his crew of 30 will be doing the same in various outposts around the country.
No time for a post-NFL Sunday hangover.
It’s time for the NFL game charters of FTN Data to watch or rewatch all the games. Not so much watch, actually. Analyze. Catalogue. Mine. Quantify the violent and interconnected interactions of 22 humans that take place in every play.
Then convert them to data points.
Log them into a Google Sheet-infused interface so FTN, where Talas is senior manager of charting, can provide this coded cache to FTN subscribers including hard-core fantasy league players, sports bettors, sportsbooks, Matthew Berry’s Fantasy Life and, as of recently, Football Almanac author Aaron Schatz.
And it’s due by Tuesday morning.
“You have to be a really big football nerd to want to chart football,” FTN Data founder and CEO Kevin Adams said, admiringly.
Talas is, ecstatically. He was a teacher for 17 years, a former head of a high school English department who also taught Cinema Studies. He’s still in the communication and entertainment businesses and film is at the center of it. But now he plies his trade with data points instead of turned phrases. And it is his job now to take the emotion out of the performance. Just facts.
There’s still a story to be told.
“Our charters’ responsibility is to be able to tell the narrative in our data,” Talas told Gaming Today.
Somewhere in that digital narrative will be why the Browns blitz so befuddled Joe Burrow, what Jordan Love saw and Justin Fields didn’t.
Talas (right) describes his crew as “football guys” from disparate backgrounds. There are data guys fresh out of college hunting full-time work, some angling for a career in sports information. There are moonlighting teachers. But there’s a common thread.
“They’re either really big into fantasy and they’re looking for an edge or they just want to sit in air conditioning and watch football as a side hustle.”
“It is a definite football atmosphere. If something happens on Sunday, it’s not out of the realm for there just to be some shop talk. Like, ‘Did you see what Lamar just did?’,” Talas related. “And then when it’s time to get going, guys are grinding and asking questions.”
“I would say, on average, an entire chart takes somewhere around 15 hours to chart for each game total. That’s with every chart: player participation, run, game pass game.”
FTN Data’s Charters Harvest NFL Data the Hard Way
On an NFL weekend with no byes, Talas’s crew produces 80 charts on weeks with no bytes – logging stats for player participation, passing and running games, and line play – with each charter assigned to a particular team. Like a beat writer, they become familiar with individual players, able to identify them by a brand of cleats or length of hair if their numbers are obscured on a certain play.
The NFL disgorges a myriad of its own statistics, but player participation isn’t among them. This makes the FTN chart valuable, the most labor-intensive, and dependent on learned observers to produce quickly. Seventy pieces of information are logged on each play on this chart.
FTN charters subscribe to the NFL’s video “film” feeds through NFL+. The videos become available on a dedicated website about five minutes after a game ends. First comes the broadcast view, essentially what fans watched. About a half hour later a condensed view, without huddles and commercials is available. They often wait for that one.
Charters input what they can from the NFL’s play-by-play stats while awaiting the first film release.
“They’re not just like starting from scratch. They see, ’14:32 in the first quarter as a run play. I need to chart this’,” Talas explained. “And then they’ll go through and they’ll look at motion, they’ll look at play action pass, they’ll look at this or that, run concepts, box counts …”
One charter will log running plays and another passes for each game. It’s still not perfect because camera views are dependent on the network production of the broadcast. Talas doesn’t like producers going seeking art as he’s harvesting data.
“I can tell when there are certain hotshot directors because they’ll wait until the last second before they come to the snap of the play,” Talas laughed. “And that gets really frustrating for us. It’s like yeah, ‘Hotshot directors … Connor is flexing today. … Win your Emmy some other game.”
There will be information gaps in broadcasts even when ‘Connor’ isn’t flexing, making the so-called All-22 coaches’ film so valuable.
“Unfortunately, that doesn’t come out until 48 hours or so after game time,” Talas said. “And so what I have my guys do in the charting system is I’ll have them flag it for review and then I will circle back on it once the All-22 drops and I can go through and kind of get things that guys missed.”
Which can make Tuesday morning deadlines sweat-inducing if there are numerous plays and stats to cross-reference.
All Hail the All-22
“The reason we don’t use it for things like player participation, pass game is because it’s really hard to identify player numbers because it’s so far zoomed out. However, it allows us to see target routes very clearly, and it allows us to see what the defense is trying to do because in a broadcast view, usually the safeties get cut off and anything that happens downfield with corners while the quarterback is dropping back.” – Mike Talas
The coach film is also the only way to produce a new product FTN subscribers had demanded in the offseason: coverage shells. Talas and his band need the All-22 for the proper field view to start counting defensive backs. It makes for a lot of staring at a television, forwarding and rewinding.
“Something that we incorporated this off-season was going through and looking at every single pass play from the entire 2022 NFL season, and we went through and charted what the coverage shell was for each play, whether it was cover one, cover two, cover two/man, cover six, cover nine, all of those types of things,” Talas said.
“That is something that is very frustrating with the NFL and their app because there’s only one place to get the coaches’ film that’s broken down by play and timestamp. Otherwise, they just throw in a jumble of film that’s three different angles for the coach’s film and they don’t give you a timestamp for it. So we do that a little later in the week.”
What was the target route?
Who was the primary defender?
Zone or man?
Did they have help?
Over and over.
FTN is a Rarity in Charting NFL Stats Manually
FTN Data’s excursion into this type of advanced charting began in 2021 with the acquisition of analytics firm Armchair Analysis. Currently, only FTN, Pro Football Focus, and Sports Info Solutions are going to this level, Adams said.
“Sportradar’s not even doing it. They’re out of the charting game,” Adams said. “It gives us a very unique edge to create tools, content, radio shows, and other types of things where it’s FTN Data, much like Pro Football Focus.”
FTN began its collaboration with Football Almanac ahead of the upcoming NFL season.
“They’re probably the most famous advanced football metrics. And now every time someone says DVOA, it is absolutely FTN Data that they’re actually talking about,” Adams said. “That is a very unique niche in that we’re getting a big part of the NFL data community, which has really been dominated by one monster: PFF.”
Adams (right) said his company eschews utilizing the NFL’s Next Gen Stats because of price and inflexibility.
“We could buy the Next Gen feed for like $10 grand a season, or $15, or whatever they sell it for. You can’t do anything with it,” he explained. “You have to just follow their guidelines. And the NFL is very notorious for being very guideline-stringent. They love their guidelines.
“So you can’t repurpose it into tools. You can’t resell it, you can’t do anything with the data. And then they could also cut it off. They could change the price.
“So with the collection of data, it makes it our data. And the NFL legally cannot tell us what we can or cannot do with that data.”
Listen to the full-length, in-depth interview with FTN’s Mike Talas for more stat chat
FTN Corners the NFL Stats Market with Coverage Shells Charting
Talas said certain stats in the NFL geekhead lexicon couldn’t exist his its charters.
Noticeable among them is quarterback pressures, which is defined as a defender advancing within two yards of a quarterback as he’s releasing the ball or 1.5 yards at any point after the snap. The NFL does real data on hurries, which is when a defender is within 1.5 yards of the quarterback at any time preceding a sack or a pass. They seem redundant, but they’re not and statheads want the data.
“There’s all different types of things that the NFL doesn’t give,” Talas explained. “For instance, things like highlight throw, created catch, coverage shells, individual man versus zone, box counts, run concepts. We do quarterback drop-back type, whether he rolls right or left, if he does straight drop back, if he leaves the pocket anytime during the play.
“Did he leave that pocket because of pressure or did he bail early? All of those types of things are things that we provide.”
And things, Adams contends, that a growing group wants.
“Just a box score saying it’s an incomplete pass doesn’t really tell you that much. We want to know if that cornerback is exceptional. We want to know if the wide receiver stinks because he can’t catch an easy ball to his hands. Those are the types of things that the DFS, the betting and the fantasy players want, these advanced stats that are going to give them a little bit of edge.
“And as legalized gambling continues people are going to be searching for that edge more and more and more because they want their gambling dollars to last longer. That’s our goal here is to make you a smarter bettor. A smarter fantasy player. You’re not gonna hear locks or guarantees … but we will make you a lot sharper.”
Computer vision continues to make inroads in sports data mining, but Talas doesn’t think artificial intelligence is yet sophisticated enough to correctly analyze football videos. He paints a scenario where a physical error – like an offensive lineman stumbling at the snap – could confuse computer vision programs about what formation and play was supposed to happen.
“My fear,” he said, “about AI is they wouldn’t be able to pick up on the nuances of things like that, player mistakes.”
“I think that being able to have a human brain, human eyes – while we’re going to deal with errors just like with anything else – being able to distinguish things like human error on a running play or a gator arm on a passing play, those are things that I feel like give us a competitive edge over systems like AI.”
AI can err because it can’t yet account for human fallibility. A human can err because of subjectivity. A fan can be too tough or lenient on their team. There’s a safeguard built in for that, at least for Talas.
“I have some guys that are on their favorite team,” he grinned, wearing a Florida Panthers hat, “some, some that aren’t. …. I’ll just say this: I try to avoid giving myself the Dolphins.”
There are plenty of other games to watch and information to catalog. And the DFS players and the bettors are waiting.