Call For Research: Ted and Analytics

[Maxie the Taxi]

^^^Don't they call it "confirmation bias" or something?

[pbmax]

Yes. But there are 100 times more Randall's than either Gunter or Allison. If you're looking for a classic SPARQ failure, it's not Randall. It's Jeff Janis. Nobody's arguing a pure athlete without natural football instincts is a sure bet to make it in the NFL. We're saying it's the best way to cull through the thousands of players in college and find the players worth pursuing by scouts.

Agreed, however, I'd argue he's a product of scouting too. And unlike Randall, who I tend to be skeptical is ever gonna amount to more than a journeyman, I still think one of these years, something will click for Janis, and he will be a star in this league - hopefully with the Packers.

Data based analytics are not a replacement for scouting. No one hires an analytics person then turns the scouting offices into a rec area. They help anchor scouting with measurements and grades that can cross college conferences and opponents, and even time.

They let you remember more lessons from the past, not fewer. They also help to bust myths. There probably is a huge advantage to being a tall QB. But there are levels of performance that make the limit worth ignoring. Analytics can help you figure out where that is with less subjectivity.

But mostly they help you at the margins, later in the draft, to find the overlooked talent that is still available because the database doesn't forget who got what grades. And the database also incorporates the grades the scouts gave the players.

I doubt Randall was simply an analytics find. I bet it took scouts who thought his skills translated to CB to make the case that he was worth a first rounder regardless of his measurables. Ted and Ron took flyers on great athletes long before data became a thing. Remember Ted said he likes his guys to remind him that the idea is not just to have measurables, its to take good football players. The data just helps you sort the tradeoff you are making.

[vince]

Exactly.

Another important thing about analytics is that it's an evolutionary process. The world is complex and simple formulas incorporating a few data points will never accurately reflect reality.

As actual results are measured against what the analytic models used have predicted, the models must be updated and additional data points may be added to continuously improve their predictive value.

In other words, if your initial models don't work Ron, you need to update them and likely enhance their sophistication through the incorporation of additional data and performance metrics to make them increasingly reliable.

It's not analytics that don't work, it's the people who don't know how to use them to enhance (not replace) their existing experience and judgment for better decision-making.

[texaspackerbacker]

I was joking.

Did you watch the youtube? Or the movie? It's great.

I thought you might be hahahaha. I just watched the little clip you posted.

[Maxie the Taxi]

I can almost guarantee you that the Randall and Rollins picks were the result of scout "reasoning" rather than analytics.

Randall wasn't even ranked by pSPARQ analytics as a CB, but as a FS. As a free safety he ranked #11 among prospects, just four spots better than Jermaine Whitehead who is currently on our Practice Squad.

Rollins was ranked #67 among all prospects tested at CB.

If you rank Randall's pSPARQ score among the CB's tested, it turns out he would have been #23 among CB's, certainly not 1st round draft material.

[texaspackerbacker]

Does his wife know?
He better hope not. Mrs. Janis:



(There is supposed tb be a picture of her here with the buck she shot.)

https://www.instagram.com/p/BL9jtVkjdlv/

Nice! (the girl, but the buck too) ........ amazing how you can find just about anything on the internet these days.

[texaspackerbacker]

I can almost guarantee you that the Randall and Rollins picks were the result of scout "reasoning" rather than analytics.

Randall wasn't even ranked by pSPARQ analytics as a CB, but as a FS. As a free safety he ranked #11 among prospects, just four spots better than Jermaine Whitehead who is currently on our Practice Squad.

Rollins was ranked #67 among all prospects tested at CB.

If you rank Randall's pSPARQ score among the CB's tested, it turns out he would have been #23 among CB's, certainly not 1st round draft material.

Seriously? I'd sure like to hear the Ted "reasoning" for those picks.

If I was a GM, I'd go with college on field performance ahead of both scout "reasoning" and analytics - for everything except maybe RBs that is.

[Maxie the Taxi]

Moreover, if TT had relied on pSPARQ analytics in 2015 he would have most likely drafted Ronald Darby with his first round choice, a choice Tex might have liked a whole lot more than Randall.

[Maxie the Taxi]

Seriously? I'd sure like to hear the Ted "reasoning" for those picks.

If I was a GM, I'd go with college on field performance ahead of both scout "reasoning" and analytics - for everything except maybe RBs that is.

Tex, scout reasoning IS based on college on field performance, but there is a HUGE amount of subjectivity involved, comparing performance from program to program, level to level and conference to conference for instance.

[vince]

Seriously? I'd sure like to hear the Ted "reasoning" for those picks.

If I was a GM, I'd go with college on field performance ahead of both scout "reasoning" and analytics - for everything except maybe RBs that is.

All aspects and dynamics of college on-field performance as well as scout "reasoning" can be and are incorporated into performance models. The magic of automation and the vast availability of data enable the development of algorithms that measure complex relationships and weigh hundreds of variables together.

Character ratings, quality of college competition, scheme contribution considerations, and much, much more are incorporated.

Players wear body telemetry instruments that track all kinds of physical activity and stress measurements, creating mounds of data that are being incorporated and corollated with strength training data, nutrition data, medical histories, as well as a variety of additional performance metrics to offer insight into how practice schedules, nutrition plans, strength training goals, and ultimately on-field performance may be able to be optimized.