VonKarman Posted December 11, 2019 Share Posted December 11, 2019 7 minutes ago, Leader said: Edit add on: Btw....the post that I made didnt deal with "the catch probability of each throw" (as you mention above) - it was the time from snap to release of the pass by the QBs listed....and I do think that time can be calculated. Not with the degree of accuracy they pretend to do it with. Quote Link to comment Share on other sites More sharing options...
Leader Posted December 11, 2019 Author Share Posted December 11, 2019 1 minute ago, VonKarman said: Not with the degree of accuracy they pretend to do it with. It's probably easier than pie with an electronic digital stop watch. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Running play? Reset to zero...... Thats how it would be done. Quote Link to comment Share on other sites More sharing options...
VonKarman Posted December 11, 2019 Share Posted December 11, 2019 9 minutes ago, waterfish_21 said: How do statistics work. You need a big enough statistical population to begin with. How are you gonna get any meaningful info when each throw&catch is unique? Quote Link to comment Share on other sites More sharing options...
deathstar Posted December 11, 2019 Share Posted December 11, 2019 Weird debate in here. Is it 1980? Quote Link to comment Share on other sites More sharing options...
AlexGreen#20 Posted December 11, 2019 Share Posted December 11, 2019 3 minutes ago, Leader said: It's probably easier than pie with an electronic digital stop watch. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Running play? Reset to zero...... Thats how it would be done. Who's clicking the watch? Accurate to a hundredth of a second? Human and laser times are usually off by like .15 seconds in track. Who's doing this for 16k throws? Quote Link to comment Share on other sites More sharing options...
VonKarman Posted December 11, 2019 Share Posted December 11, 2019 Just now, Leader said: It's probably easier than pie with an electronic digital stop watch. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Running play? Reset to zero...... Thats how it would be done. Nope. There are two/three main reasons for that. 1- How many frames per second do you have in your tape? 24?, 60? You need at least 100 fps to be able to have a 0.01 uncertainty 2-How do you determine in which exact frame the snap starts without a doubt? There's probably. a 2-3 frame uncertainty when the ball is snapped and the same happens when the ball is thrown. 3-Who makes those decisions? Are you sure these companies develop the technology to do it or do they have a guy doing an internship doing it. The uncertainty analysis is not as simple as some of you may think. 2 Quote Link to comment Share on other sites More sharing options...
waterfish_21 Posted December 11, 2019 Share Posted December 11, 2019 Just now, VonKarman said: You need a big enough statistical population to begin with. How are you gonna get any meaningful info when each throw&catch is unique? Whoa hold your brakes! Who brought statistics/applied math/probability into this. Were not talking about actual statistics. I'm talking about the sports term of a stat, like "QBR", or "quarterback rating" or "yards per rush". The stat that they created and named "Catch probability" is not actual catch probability applied through statistics. They use variables they gather and put it in an algorithm and post it for everyone to see. We can all make our own version and post it as well. You guys are digging way to deep into this. I wish they would show how long their database goes back and their algorithm so there could at least be discussion about this, but we won't. Quote Link to comment Share on other sites More sharing options...
Norm Posted December 11, 2019 Share Posted December 11, 2019 7 minutes ago, Leader said: It's probably easier than pie with an electronic digital stop watch. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Pass release - click. Make note of the data and reset to zero. Snap - click. Running play? Reset to zero...... Thats how it would be done. Lol guys. There's no way that's how this is done in 2019. I need to find a video and show you what they do but I don't know exactly how to go about it. Quote Link to comment Share on other sites More sharing options...
VonKarman Posted December 11, 2019 Share Posted December 11, 2019 Just now, waterfish_21 said: Whoa hold your brakes! Who brought statistics/applied math/probability into this. Were not talking about actual statistics. I'm talking about the sports term of a stat, like "QBR", or "quarterback rating" or "yards per rush". The stat that they created and named "Catch probability" is not actual catch probability applied through statistics. They use variables they gather and put it in an algorithm and post it for everyone to see. We can all make our own version and post it as well. You guys are digging way to deep into this. I wish they would show how long their database goes back and their algorithm so there could at least be discussion about this, but we won't. Yeah, in if we do that we shouldn't be taken seriously, the same way we do not take NextGenStats seriously. Quote Link to comment Share on other sites More sharing options...
Leader Posted December 11, 2019 Author Share Posted December 11, 2019 Just now, AlexGreen#20 said: Who's clicking the watch? Accurate to a hundredth of a second? Human and laser times are usually off by like .15 seconds in track. Who's doing this for 16k throws? It seems apparent somebody is actually doing it no? Unless (again) you feel these stat platforms (PFF / NextGen) are putting out bogus data. Technically its possible. It can be determined with an electronic digital stopwatch. With human assistance, the data CAN be captured electronically. Might be the most boring job in the world - but it actually wouldnt be that difficult. In the course of a normal game, there are what (?) two snaps per minute - on average (?) and not all of them would be pass plays. Quote Link to comment Share on other sites More sharing options...
AlexGreen#20 Posted December 11, 2019 Share Posted December 11, 2019 9 minutes ago, Leader said: It seems apparent somebody is actually doing it no? Unless (again) you feel these stat platforms (PFF / NextGen) are putting out bogus data. Technically its possible. It can be determined with an electronic digital stopwatch. With human assistance, the data CAN be captured electronically. Might be the most boring job in the world - but it actually wouldnt be that difficult. In the course of a normal game, there are what (?) two snaps per minute - on average (?) and not all of them would be pass plays. Yes, they're putting out bogus data. If the entire thing isn't bogus than the accuracy that they're claiming certainly is. Quote Link to comment Share on other sites More sharing options...
Leader Posted December 11, 2019 Author Share Posted December 11, 2019 (edited) 9 minutes ago, AlexGreen#20 said: Yes, they're putting out bogus data. If the entire thing isn't bogus than the accuracy that they're claiming certainly is. Fine. If you're actually interested to know, I suggest you contact either PFF or NextGen on Twitter and ask them. It's very likely they'll tell you. I've provided a simple (and possible/probable) method it could be done. If thats what they're doing - dont know - but its a simple procedure that doesnt require some elaborate design, lasers or alike. Just a stopwatch, two fingers, good eyesight, quick reflexes and a pad to make note of the data. It could be done during the game live - or leisurely with an All 22 film that can be stopped, started, rewound and such. Done. Edited December 11, 2019 by Leader Quote Link to comment Share on other sites More sharing options...
Fl0nkerton Posted December 11, 2019 Share Posted December 11, 2019 ur tellin me there aren't football scientists working for espn Quote Link to comment Share on other sites More sharing options...
waterfish_21 Posted December 11, 2019 Share Posted December 11, 2019 13 hours ago, Norm said: I'd sure hope they are ripping the video from a22 site then using their own players for easy rewinds already. Then for those timings they watch the timestamps on the video for the time to throws. Right? That's what I'd do. I remember in college having to write a program that broke a video down by frames. It could do things like calculate time between frames, wasn't 100% accurate but was considered pretty close. This was used on videos on youtube. Used a bit of open source code as well, very simple stuff. The hardest part, not difficult, just time consuming and most likely inaccurate/inconsistent, would be finding the frame at which you start and stop the calculation. But PFF supposedly has people that do this stuff all day anyways, so who knows. 1 Quote Link to comment Share on other sites More sharing options...
AlexGreen#20 Posted December 11, 2019 Share Posted December 11, 2019 7 minutes ago, Leader said: Fine. If you're actually interested to know, I suggest you contact either PFF or NextGen on Twitter and ask them. It's very likely they'll tell you. I've provided a simple (and possible/probable) method it could be done. If thats what they're doing - dont know - but its a simple procedure that doesnt require some elaborate design, lasers or alike. Just a stopwatch, two fingers, good eyesight, quick reflexes and a pad to make note of the data. It could be done during the game live - or leisurely with an All 22 film that can be stopped, started, rewound and such. Done. You think a person with a stopwatch is accurate to a hundredth of a second? You would be lucky to get within a tenth of a second. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.