Pro Football Focus' Matthew Coller | Giants Huddle

[MightyGiants]

Dispells many of the myths that have floated around our forum.   Turns out Tom Coughlin was a big fan of PFF

On this episode of the Giants Huddle podcast, John Schmeelk is joined by PFF's Matthew Coller to discuss his book, Football Is a Numbers Game: Pro Football Focus and How a Data-Driven Approach Shook Up the Sport.

[MightyGiants]

"The more they got the information, the more that Tom Coughlin wanted to go to (former Giants senior director of football information) Jon Berger and say, 'Hey, can you get PFF to give us this? Can we find out that?' (It got) to the point where PFF assigned one person just to answer Giants questions," Coller elaborated on the Giants Huddle podcast.

Coughlin and the Giants were so pleased with the help provided by the PFF employee that they invited him to Super Bowl XLVI on their behalf.

"In the Super Bowl, the second time they went to the Super Bowl, they gave that guy a ticket to the Super Bowl because they felt he was so valuable in helping them out (and) answering whatever Tom Coughlin's questions were," Coller said.

https://giantswire.usatoday.com/2024/06/23/did-ex-new-york-giants-coach-tom-coughlin-use-pff-win-super-bowl-xlvi/?taid=66781442ffc9eb0001c1e7f3&utm_campaign=trueanthem&utm_medium=trueanthem&utm_source=twitter

[uconnjack8]

Is this considered using analytics?

[MightyGiants]

Is this considered using analytics?

PFF generally does fall under the analytics umbrella.  To be fair, the PFF grades are just one small part of what PFF does (and not the part the teams are as interested in compared to the rest of what they do). 

[Bob In PA]

PFF generally does fall under the analytics umbrella.  To be fair, the PFF grades are just one small part of what PFF does (and not the part the teams are as interested in compared to the rest of what they do). 

@uconnjack8 and Rich: The difference, IMO, between actual analytics and PFF is that analytics analyze data, and data is FACTS, whereas (to a noteworthy extent) PFF "data" is a mixture of fact and opinion (or evaluation), so it's not IMO true data. Bob

[DaveBrown74]

@uconnjack8 and Rich: The difference, IMO, between actual analytics and PFF, of course, is that analytics analyze data, and data is FACTS, whereas (to a noteworthy extent) PFF "data" is a mixture of fact and opinion (or evaluation), so it's not IMO true data. Bob

Completely agree. With PFF there is an element of subjectivity. With pure analytics, hard data are being analyzed in a variety of ways. The conclusions then drawn the work done may be up for subjective interpretation and contextual explanation, but the actual data itself is not.

[nb587]

Did the Giants go backwards in the analytics department when they hired Gettleman?  Didn't seem like he had much use for analytics?  (But that could just be my anti Getty bias)

[DaveBrown74]

Did the Giants go backwards in the analytics department when they hired Gettleman?  Didn't seem like he had much use for analytics?  (But that could just be my anti Getty bias)

Gettleman was overtly anti-analytics. He scoffed at it and used to mock it regularly. When he finally acquiesced and they brought in some sort of small group of individuals to begin to try to get the Giants more up to speed on analytics, he referred to them as "computer folks."

[Hadron]

Yeah. I remember that press conference where Gettleman was mocking analytics. Shurmur was sitting there, half asleep, but even he gave Gettleman a look of 'seriously bro?' or something of the ilk.

[DaveBrown74]

Yeah. I remember that press conference where Gettleman was mocking analytics. Shurmur was sitting there, half asleep, but even he gave Gettleman a look of 'seriously bro?' or something of the ilk.

Yup. He mocked analytics, and he also scoffed at the concept of positional value with comments like "a good player is a good player", etc. He further pushed back on the comment that RBs have a shorter shelf life than other players, and he used Jonathan Stewart as an example of an RB who is "still Jonathan Stewart after all these years." Stewart, then 31, had 17 yards on 6 attempts for us that season and never saw an NFL field again after that.

[MightyGiants]

@uconnjack8 and Rich: The difference, IMO, between actual analytics and PFF, of course, is that analytics analyze data, and data is FACTS, whereas (to a noteworthy extent) PFF "data" is a mixture of fact and opinion (or evaluation), so it's not IMO true data. Bob

@DaveBrown74

I purchased the book about PFF that was mentioned in the video (since I am still laid up from my injuries, I have more reading time on weekends).

To Bob's comments, PPF does two things.  There is factual data (like player participation or where they lined up) that is completely objective.   That is the data NFL teams are buying.  Interestingly enough, the Giants were the first NFL team to work with PFF.  Sadly, in the beginning, since PFF was originally UK based, the Giants had to email PFF to call them because the cheap-ass team's switchboard didn't allow international calling.

Regarding the grades, which I assume Bob is discussing, they (like any grade on an NFL player) are subjective. That said, not all subjective measures are the same in terms of quality.  First, any system to measure players has subjectivity in terms of how to value different performances.   Second, there is some subjectivity in terms of how a player performs in a given play.   To those points, I agree.

However, a subjective system that creates a consistent set of rules that can be taught to others, which creates repeatability, has value over a random subjective opinion.  Considering that former NFL coaches have either worked with or others worked for PFF to tweak the grading system also adds to the value of such grades.  Graders (despite suggestions posted here), have to read and follow a 300-page graders guide. Once a grader grades a game, it's then reviewed by position-specific graders to improve "accuracy."

So, it's technically correct to say PFF grades are subjective, but I think that by saying so, one really needs to appreciate that subjective has significant degrees of quality