1 00:00:00,120 --> 00:00:06,680 Speaker 1: It's time to get inside the Giants Huts. Giants Giants Bubbling, 2 00:00:07,040 --> 00:00:09,720 Speaker 1: give me some job. Part of the Giants Podcast Network. 3 00:00:09,760 --> 00:00:12,360 Speaker 1: Let's roll. Welcome to another edition of the Giants Little Podcast, 4 00:00:12,400 --> 00:00:14,520 Speaker 1: brought to you by Citizens Official Bank of the Giants. 5 00:00:14,560 --> 00:00:16,960 Speaker 1: I am John Schmelk, coming to you from our podcast studio, 6 00:00:17,040 --> 00:00:20,079 Speaker 1: presented by Hackensack, Brady and Health Keep Getting Better. Very 7 00:00:20,120 --> 00:00:23,160 Speaker 1: excited to be joined by the host of Prime Vision 8 00:00:23,239 --> 00:00:26,280 Speaker 1: and Amazon Prime Thursday Night Football, Sam Schwartztein. 9 00:00:26,320 --> 00:00:28,720 Speaker 2: Sam, what's going on, man, I'm excited to have you. 10 00:00:28,760 --> 00:00:31,240 Speaker 2: I've been watching some Giants football, watch that Giants preseason 11 00:00:31,240 --> 00:00:33,920 Speaker 2: game about three times already, so looking forward to talking. 12 00:00:34,280 --> 00:00:37,800 Speaker 1: Yeah, and Giants' final preseason game against the Patriots will 13 00:00:37,800 --> 00:00:39,919 Speaker 1: be on Thursday Night Football and Amazon Prime. There will 14 00:00:39,960 --> 00:00:41,920 Speaker 1: not be a Prime Vision broadcast that day, but we 15 00:00:41,960 --> 00:00:44,440 Speaker 1: are going to talk about that game moving forward. Here. 16 00:00:45,000 --> 00:00:47,640 Speaker 1: Do you do any prep work in the preseason getting 17 00:00:47,640 --> 00:00:50,760 Speaker 1: ready for the regular season during these preseason broadcasts, Sam. 18 00:00:50,840 --> 00:00:54,520 Speaker 2: Absolutely, we treat this like a regular game that we're 19 00:00:54,520 --> 00:00:56,320 Speaker 2: going to have. We actually are probably more prep for 20 00:00:56,360 --> 00:00:58,680 Speaker 2: this game than we will be for our Week three 21 00:00:58,800 --> 00:01:01,560 Speaker 2: game because of how much time we're gonna how much 22 00:01:01,640 --> 00:01:03,480 Speaker 2: lead of time. I have more games to watch, and 23 00:01:03,920 --> 00:01:07,039 Speaker 2: there's no short week here for watching for our games, 24 00:01:07,080 --> 00:01:08,840 Speaker 2: so normally we have the short week we have to 25 00:01:08,880 --> 00:01:11,039 Speaker 2: catch up injuries and things can change. What prep I've 26 00:01:11,040 --> 00:01:13,199 Speaker 2: done on Friday for what ends up happening on Sunday, 27 00:01:13,480 --> 00:01:16,440 Speaker 2: then preyer for Thursday, and so we retreat this. We're 28 00:01:16,440 --> 00:01:17,800 Speaker 2: in game modes starting this week. 29 00:01:18,240 --> 00:01:22,160 Speaker 1: Do you take anything out of these preseason games from 30 00:01:22,240 --> 00:01:27,080 Speaker 1: a data perspective, trend perspective or do you just watch 31 00:01:27,120 --> 00:01:29,640 Speaker 1: it and then crinkle it up, throw it away and 32 00:01:29,680 --> 00:01:31,680 Speaker 1: start from scratch with game one regular season. 33 00:01:32,360 --> 00:01:34,720 Speaker 2: I'm trying to see, because we have no data on rookies, 34 00:01:35,240 --> 00:01:37,840 Speaker 2: I try and see what a rookie tendency might be 35 00:01:38,480 --> 00:01:41,240 Speaker 2: like Jackson Dart throwing under pressure. That was a great 36 00:01:41,319 --> 00:01:43,959 Speaker 2: moment for him. I will want to see that continue, 37 00:01:44,080 --> 00:01:47,120 Speaker 2: and you can't replicate that in practice. And since we 38 00:01:47,200 --> 00:01:50,080 Speaker 2: don't have data coming into these guys in the season, 39 00:01:50,280 --> 00:01:52,200 Speaker 2: I can't really project out what he'll be like in 40 00:01:52,240 --> 00:01:54,680 Speaker 2: Brian Dables offense from what he did in Ole Miss. 41 00:01:54,680 --> 00:01:57,360 Speaker 2: Everyone can know that one word plays in Ole Miss 42 00:01:57,400 --> 00:01:59,360 Speaker 2: are gonna be very different than what Brian Dables calling. 43 00:01:59,480 --> 00:02:03,360 Speaker 2: So That's my focus is how are the rookies translating 44 00:02:03,400 --> 00:02:06,120 Speaker 2: abdul Carter? Are they translating what's their game? Like? A 45 00:02:06,160 --> 00:02:08,600 Speaker 2: is aera rush inside versus outside? Got his pressure rate? 46 00:02:08,680 --> 00:02:10,799 Speaker 2: He did great? That's what I'll try and see. 47 00:02:10,960 --> 00:02:13,600 Speaker 1: Will you take any college data for the rookies. Let's say, 48 00:02:13,600 --> 00:02:15,480 Speaker 1: you know, week one, week two regular season, you don't 49 00:02:15,480 --> 00:02:17,440 Speaker 1: have any real rookie dadda yet? How much will you 50 00:02:17,520 --> 00:02:19,839 Speaker 1: use some of the stuff from college, even if it's 51 00:02:19,840 --> 00:02:22,840 Speaker 1: just tracking data stuff like that to try to be 52 00:02:22,919 --> 00:02:23,800 Speaker 1: a little bit predictive. 53 00:02:24,960 --> 00:02:27,000 Speaker 2: I actually spend less time on that because of scheme 54 00:02:27,040 --> 00:02:29,919 Speaker 2: difference and the rules differences are so you know, up 55 00:02:29,960 --> 00:02:33,760 Speaker 2: tempo offense is transfer portal. It's so different that you know, 56 00:02:34,200 --> 00:02:36,840 Speaker 2: there's some guys who probably spent more time on their 57 00:02:36,919 --> 00:02:39,240 Speaker 2: NFL team than they did with their their college football 58 00:02:39,280 --> 00:02:41,240 Speaker 2: team last year because of the transfer portal. They might 59 00:02:41,280 --> 00:02:44,560 Speaker 2: transfer it in the fall. So I look at the 60 00:02:44,639 --> 00:02:48,679 Speaker 2: NNGS athletics score from next Gen Stats, try and see 61 00:02:48,680 --> 00:02:52,440 Speaker 2: if I can find appical comps for a player. Did 62 00:02:52,440 --> 00:02:54,440 Speaker 2: they have somebody similar on the team that was comped 63 00:02:54,440 --> 00:02:56,680 Speaker 2: to the same athletics score. And that's why I'll try 64 00:02:56,760 --> 00:02:59,120 Speaker 2: and project out what I think they'll do in the NFL, 65 00:02:59,200 --> 00:03:01,440 Speaker 2: just because the game is so different. 66 00:03:02,160 --> 00:03:04,280 Speaker 1: Yeah, and look, it'll be the last Jests rehearsal coming 67 00:03:04,320 --> 00:03:06,800 Speaker 1: up Thursday night for the Giants and Patriots. We're excited for. 68 00:03:06,840 --> 00:03:08,960 Speaker 1: It should be an excellent game. People should go check 69 00:03:09,000 --> 00:03:10,959 Speaker 1: it out. The only place you can find on televisions 70 00:03:11,280 --> 00:03:13,880 Speaker 1: is on Amazon Prime, So people should go check that out. 71 00:03:14,000 --> 00:03:16,120 Speaker 1: All right, Sam, I'm gonna start here for giant fans 72 00:03:16,160 --> 00:03:19,160 Speaker 1: that maybe are not familiar with Prime Vision, they have 73 00:03:19,160 --> 00:03:21,320 Speaker 1: not checked it out yet, they have not tuned in. 74 00:03:21,840 --> 00:03:24,040 Speaker 1: Why should people tune in? What are they getting if 75 00:03:24,040 --> 00:03:26,680 Speaker 1: they go to that alternate program with Prime Vision. 76 00:03:27,560 --> 00:03:30,400 Speaker 2: Look, if you're a hardcore fan or you are trying 77 00:03:30,400 --> 00:03:32,040 Speaker 2: to get into the game and learn more about the 78 00:03:32,080 --> 00:03:35,360 Speaker 2: game and try and get that what we call edutainment factor, 79 00:03:35,400 --> 00:03:38,560 Speaker 2: prime Vision is the broadcast for you. We broadcast the game. 80 00:03:38,600 --> 00:03:41,080 Speaker 2: We switched over to the behind the quarterback point of view, 81 00:03:41,080 --> 00:03:43,320 Speaker 2: that high sky camera where you can still see all 82 00:03:43,320 --> 00:03:45,560 Speaker 2: twenty two players and every single play throughout the play, 83 00:03:45,840 --> 00:03:48,160 Speaker 2: but now you're seeing it from the perspective of the quarterback. 84 00:03:48,440 --> 00:03:50,000 Speaker 2: We also have player name tags, so if you're a 85 00:03:50,040 --> 00:03:52,320 Speaker 2: fantasy football fan, and you want to know where Malik 86 00:03:52,360 --> 00:03:54,839 Speaker 2: Neighbors is lined up. You can find him on every 87 00:03:54,840 --> 00:03:58,200 Speaker 2: single play because we'll have him tagged above his head. 88 00:03:58,480 --> 00:04:01,040 Speaker 2: And then we integrate some AI into the broadcast so 89 00:04:01,040 --> 00:04:02,400 Speaker 2: that you can watch the game like a pro. I 90 00:04:02,440 --> 00:04:04,880 Speaker 2: played football at Stanford. I played it for eleven I 91 00:04:04,880 --> 00:04:08,600 Speaker 2: played football for eleven years in through college. I played center, 92 00:04:08,680 --> 00:04:12,240 Speaker 2: and I was trying to identify blitzers, and so you're 93 00:04:12,320 --> 00:04:15,360 Speaker 2: one Alex Strand. My boss said, hey, let's combine what 94 00:04:15,440 --> 00:04:18,200 Speaker 2: you know with what orII scientists can do. Let's put 95 00:04:18,200 --> 00:04:21,320 Speaker 2: something together, and now you can find this The blitzer 96 00:04:21,360 --> 00:04:23,320 Speaker 2: is the same way our quarterbacks trying to I know 97 00:04:23,320 --> 00:04:24,960 Speaker 2: who those threats are, and we'll be highlighted with a 98 00:04:24,960 --> 00:04:27,000 Speaker 2: red orb. A lot of math going on, a lot 99 00:04:27,040 --> 00:04:29,840 Speaker 2: of really cool engineering work that may football fans might 100 00:04:29,839 --> 00:04:32,320 Speaker 2: not care about to make it happen, but you're now 101 00:04:32,360 --> 00:04:34,039 Speaker 2: able to see that. And every year we had more 102 00:04:34,080 --> 00:04:36,719 Speaker 2: AI models that help you watch the game like a 103 00:04:36,800 --> 00:04:38,840 Speaker 2: pro and help you get closer to the game. So 104 00:04:39,200 --> 00:04:41,840 Speaker 2: whether it's the unique camera angle, the AI models, or 105 00:04:41,880 --> 00:04:43,839 Speaker 2: the ATA statistics to help you know more about the 106 00:04:43,839 --> 00:04:45,160 Speaker 2: game that this. 107 00:04:45,160 --> 00:04:47,680 Speaker 1: Is the broadcast for you without getting too deep into 108 00:04:47,720 --> 00:04:51,000 Speaker 1: the algorithms, the math and the AI engine. How does 109 00:04:51,000 --> 00:04:53,640 Speaker 1: that work? How do you guys go about building that 110 00:04:53,760 --> 00:04:57,560 Speaker 1: out where you can predict the blitzer simply based on 111 00:04:57,800 --> 00:05:00,240 Speaker 1: I imagine it's some combination of down and distance in 112 00:05:00,560 --> 00:05:02,280 Speaker 1: where players are positioned on defense. 113 00:05:02,320 --> 00:05:05,520 Speaker 2: Correct, Yeah, it's pretty fun. We started out with a 114 00:05:05,600 --> 00:05:07,880 Speaker 2: rules based model because as a former center, that's what 115 00:05:07,920 --> 00:05:10,359 Speaker 2: I used to do. Okay, we're in trips right, we 116 00:05:10,440 --> 00:05:13,599 Speaker 2: have four coverage defenders over our three wide receivers. Somebody 117 00:05:13,680 --> 00:05:15,680 Speaker 2: might be able to blitz from that area. But then 118 00:05:15,680 --> 00:05:19,240 Speaker 2: when we started working with our machine learning AI scientists teams, 119 00:05:19,240 --> 00:05:20,880 Speaker 2: they are like, we actually might be able to do 120 00:05:20,920 --> 00:05:22,599 Speaker 2: this in a little bit of a more unique way. 121 00:05:22,720 --> 00:05:25,640 Speaker 2: And what happens here is you're able to find blitz. 122 00:05:25,720 --> 00:05:27,760 Speaker 2: Is that your rules base model would never be able 123 00:05:27,800 --> 00:05:30,880 Speaker 2: to pick up. Or you can be more prescriptive. Yeah, 124 00:05:30,920 --> 00:05:33,200 Speaker 2: there might be four over three, but they know the 125 00:05:33,240 --> 00:05:38,359 Speaker 2: down and distance. The model can know what tendencies teams 126 00:05:38,440 --> 00:05:41,560 Speaker 2: may do. I call it thumb print, thumb print and blitzes. 127 00:05:41,800 --> 00:05:44,279 Speaker 2: You know, every week I would have to prepare for Okay, 128 00:05:44,320 --> 00:05:47,599 Speaker 2: this coach who was in Brett Bielama had very unique 129 00:05:47,600 --> 00:05:49,400 Speaker 2: blitzes when he was at Wisconsin. I played them in 130 00:05:49,440 --> 00:05:51,960 Speaker 2: the Rose Bowl, and I then saw them again in 131 00:05:52,000 --> 00:05:53,680 Speaker 2: the in the Super Bowl against the Ram. When he 132 00:05:53,720 --> 00:05:55,280 Speaker 2: was at the Patriots versus the Rams. It's like, oh 133 00:05:55,320 --> 00:05:57,880 Speaker 2: my gosh, but these are blitzer only he runs the 134 00:05:57,880 --> 00:05:59,760 Speaker 2: models able to find a lot of those. Thumb print 135 00:06:00,320 --> 00:06:03,360 Speaker 2: is the crazy stuff that a person who might go 136 00:06:03,400 --> 00:06:06,279 Speaker 2: by rules might not find. And so by being able 137 00:06:06,320 --> 00:06:07,920 Speaker 2: to use AI, we're able to do it in your 138 00:06:07,960 --> 00:06:11,520 Speaker 2: real time, be able to get the models working as 139 00:06:11,600 --> 00:06:13,720 Speaker 2: fast as possible so that Alan Kirk on the main 140 00:06:13,760 --> 00:06:17,280 Speaker 2: broadcast can even read and react to the model showing 141 00:06:17,360 --> 00:06:20,960 Speaker 2: up live. So using AI a lot of us to 142 00:06:21,040 --> 00:06:23,440 Speaker 2: unlock a lot of levers that a rules based model, 143 00:06:23,760 --> 00:06:25,560 Speaker 2: which is if then statements wouldn't be able to. 144 00:06:25,960 --> 00:06:28,680 Speaker 1: I'm sure you've been surprised by some things that the 145 00:06:28,720 --> 00:06:30,919 Speaker 1: A engine in terms of blitzers has been able to 146 00:06:30,960 --> 00:06:33,679 Speaker 1: predict after the game. Are you able to go back 147 00:06:33,760 --> 00:06:36,880 Speaker 1: and is the engine able to tell you why it 148 00:06:36,960 --> 00:06:41,040 Speaker 1: predicted that player in that particular play was going to blitz? 149 00:06:41,080 --> 00:06:44,039 Speaker 1: Can you get the reasoning for it after the facts 150 00:06:44,040 --> 00:06:46,279 Speaker 1: so you can learn from and then maybe apply it 151 00:06:46,320 --> 00:06:47,320 Speaker 1: in different ways. 152 00:06:48,080 --> 00:06:51,160 Speaker 2: I think what I'll go a different direction. We can 153 00:06:51,720 --> 00:06:54,960 Speaker 2: help apply reasoning to the model. So that's why I 154 00:06:54,960 --> 00:06:57,960 Speaker 2: think is the value is we can look at reasoning 155 00:06:58,000 --> 00:06:59,960 Speaker 2: and hey, maybe we missed something, and then we can 156 00:07:00,160 --> 00:07:04,640 Speaker 2: help update the model with annotations to help influence it 157 00:07:04,680 --> 00:07:07,919 Speaker 2: to highlight certain things. So that's typically what we do. 158 00:07:08,000 --> 00:07:10,040 Speaker 2: What's been a great thing this year is now I 159 00:07:10,080 --> 00:07:13,840 Speaker 2: can pull any play and find out what the blitzers 160 00:07:13,840 --> 00:07:17,920 Speaker 2: would be. And so we did not have the AFC 161 00:07:18,000 --> 00:07:20,360 Speaker 2: Championship game and that fourth and five where Josh Allen 162 00:07:20,440 --> 00:07:24,080 Speaker 2: dropped back and they did all the blitzers. Now I 163 00:07:24,120 --> 00:07:26,840 Speaker 2: can see what would we have had on our broadcast, 164 00:07:26,840 --> 00:07:30,200 Speaker 2: and I can see that the real time probabilities for it. 165 00:07:30,240 --> 00:07:32,800 Speaker 2: And so what I like is, you know, there are 166 00:07:32,840 --> 00:07:35,440 Speaker 2: sometimes the model doesn't pick up a blitzer, but that's 167 00:07:35,440 --> 00:07:37,920 Speaker 2: not a bad thing. That means that we've seen something 168 00:07:38,360 --> 00:07:42,160 Speaker 2: completely novel over the past seven years on a blitz 169 00:07:42,520 --> 00:07:44,720 Speaker 2: and so a lot of times it's a player who's 170 00:07:44,720 --> 00:07:48,120 Speaker 2: blitzing from out of position and they shouldn't be blitzing 171 00:07:48,120 --> 00:07:50,360 Speaker 2: that the offense could take advantage of it. But other times, 172 00:07:50,400 --> 00:07:53,840 Speaker 2: if we miss something, that means that we've seen something 173 00:07:53,880 --> 00:07:55,720 Speaker 2: brand new, and that's that's fun for us. 174 00:07:55,880 --> 00:07:58,360 Speaker 1: Puddle up, get in here. If you're lined up here, 175 00:07:58,480 --> 00:08:01,200 Speaker 1: you gotta go over the middle with it to score great. 176 00:08:01,720 --> 00:08:02,720 Speaker 1: How do we make that happen? 177 00:08:03,600 --> 00:08:07,040 Speaker 2: I don't know. Bud Citizen does makes sense of your 178 00:08:07,040 --> 00:08:11,440 Speaker 2: money with citizens Official Bank of Eli Manning and. 179 00:08:11,400 --> 00:08:13,200 Speaker 1: That probably indicates, all right, well, maybe this is a 180 00:08:13,200 --> 00:08:15,720 Speaker 1: pretty innovative defensive coach that we should keep an eye 181 00:08:15,720 --> 00:08:16,240 Speaker 1: on right. 182 00:08:16,640 --> 00:08:21,000 Speaker 2: Exactly, or a guy's freelancing and glance on defense. You 183 00:08:21,040 --> 00:08:23,840 Speaker 2: know that's the now you and Bobby o'kareik Stanford legend. 184 00:08:23,880 --> 00:08:25,480 Speaker 2: He don't have freelance a lot, but there might be 185 00:08:25,520 --> 00:08:27,040 Speaker 2: times where he goes and says, I can make go 186 00:08:27,120 --> 00:08:29,560 Speaker 2: make a play. And so that's that's really fun for 187 00:08:29,680 --> 00:08:30,320 Speaker 2: us to see. 188 00:08:30,520 --> 00:08:32,120 Speaker 1: Yeah, we have we have two Stanford guys for you. 189 00:08:32,200 --> 00:08:33,920 Speaker 1: Now we got Okaica and Paulson a deal with a 190 00:08:33,920 --> 00:08:35,880 Speaker 1: big freege of edition. I'm sure you got very excited 191 00:08:35,880 --> 00:08:39,200 Speaker 1: about that. You're with the predict blitzers. One thing I know, 192 00:08:39,320 --> 00:08:41,800 Speaker 1: and I'm not telling anything you don't already know. You know, 193 00:08:41,880 --> 00:08:44,160 Speaker 1: defenses are really trying to disguise in the secondary. Now 194 00:08:44,160 --> 00:08:47,360 Speaker 1: they're trying to use you know, pre snap alignment shift 195 00:08:47,440 --> 00:08:50,080 Speaker 1: post snap show one coverage to go into another to 196 00:08:50,120 --> 00:08:53,679 Speaker 1: make that quarterback kind of hesitate. How are your AM 197 00:08:53,720 --> 00:08:57,760 Speaker 1: models in terms of predicting what coverage the defenses going 198 00:08:57,800 --> 00:09:00,280 Speaker 1: to show based on the same parameters you put in 199 00:09:00,280 --> 00:09:02,040 Speaker 1: there for trying to identify blitzers. 200 00:09:02,320 --> 00:09:04,800 Speaker 2: Yeah, we had a great one. I got really scared. 201 00:09:04,840 --> 00:09:08,040 Speaker 2: We launched coverage ID when Brian Flores was the defensive 202 00:09:08,080 --> 00:09:11,280 Speaker 2: coordinator in the game, and so when we looked at 203 00:09:11,320 --> 00:09:13,520 Speaker 2: the schedule, like, oh, can we please change the launch 204 00:09:13,600 --> 00:09:17,360 Speaker 2: date for this because I was nervous, right, and we 205 00:09:17,360 --> 00:09:20,640 Speaker 2: were able to have a play where Pukaakula went in motion, 206 00:09:21,360 --> 00:09:23,400 Speaker 2: Harrison Smith followed him in motion. 207 00:09:24,000 --> 00:09:26,680 Speaker 1: That's a typical man man indicator. Yep. 208 00:09:26,880 --> 00:09:29,959 Speaker 2: Our coverage ID model never changed off zone And when 209 00:09:30,000 --> 00:09:31,520 Speaker 2: I went back and looked at the model, it had 210 00:09:31,559 --> 00:09:35,600 Speaker 2: a ninety nine percent zone prediction the entire time, never waivered. 211 00:09:36,160 --> 00:09:38,800 Speaker 2: And so it's fun to see these things then review it, 212 00:09:39,200 --> 00:09:41,199 Speaker 2: and now we have teams that are asking, hey, how 213 00:09:41,200 --> 00:09:43,280 Speaker 2: can I learn There are head coaches out there that 214 00:09:43,320 --> 00:09:46,000 Speaker 2: are asking how can I use coverage ID model to 215 00:09:46,040 --> 00:09:50,360 Speaker 2: make me better at reading coverages? Because it's helping people understand. 216 00:09:50,400 --> 00:09:52,640 Speaker 2: And we saw the Big Data Ball submission. The one 217 00:09:52,679 --> 00:09:55,800 Speaker 2: that won this year was predicting COVERAGEES pre snap and 218 00:09:55,880 --> 00:09:58,679 Speaker 2: it's it's it's now a tool that people can use 219 00:09:58,679 --> 00:10:00,800 Speaker 2: and look and go Okay, what I thought was a 220 00:10:00,880 --> 00:10:03,000 Speaker 2: key indicator may no longer be. 221 00:10:03,679 --> 00:10:06,160 Speaker 1: Now and I got to imagine this is stuff that 222 00:10:06,200 --> 00:10:09,280 Speaker 1: can be certainly useful for game planning for coaches and 223 00:10:09,360 --> 00:10:13,720 Speaker 1: quarterbacks heading into a game. Is this something that teams 224 00:10:13,760 --> 00:10:16,280 Speaker 1: are building out on their own? Are their companies out 225 00:10:16,280 --> 00:10:19,040 Speaker 1: there again, whether it's like a PFF or Assumer Sports 226 00:10:19,120 --> 00:10:21,760 Speaker 1: or the million other analytics companies out there, are they 227 00:10:21,760 --> 00:10:23,880 Speaker 1: building these models that teams that have to invest in. 228 00:10:24,160 --> 00:10:27,079 Speaker 1: How is this type of technology starting to transfer over 229 00:10:27,120 --> 00:10:28,000 Speaker 1: to the coaching level? 230 00:10:28,040 --> 00:10:30,839 Speaker 2: To your point all the above, like you said, there 231 00:10:30,840 --> 00:10:33,520 Speaker 2: are teams, There are people building models for teams. There 232 00:10:33,520 --> 00:10:35,440 Speaker 2: are teams that want to build models themselves because they 233 00:10:35,440 --> 00:10:37,400 Speaker 2: don't want to have to share proprietary data with other 234 00:10:37,559 --> 00:10:41,240 Speaker 2: right other teams? You know, I coaches are very secretive. 235 00:10:41,240 --> 00:10:43,960 Speaker 2: I've had a coach when I was doing the XFL, 236 00:10:44,240 --> 00:10:46,760 Speaker 2: I wanted to put all of our video on the cloud. 237 00:10:47,000 --> 00:10:49,400 Speaker 2: Makes much more sense than buying eight servers for different 238 00:10:49,400 --> 00:10:51,280 Speaker 2: teams and spend all that money. Well, guess what the 239 00:10:51,320 --> 00:10:54,080 Speaker 2: coaches said, No, I want my data in that box 240 00:10:54,080 --> 00:10:55,880 Speaker 2: that I have on a trolley that I bring to 241 00:10:55,920 --> 00:10:57,800 Speaker 2: me to every game. I said, that's fine, it's still 242 00:10:57,840 --> 00:10:58,920 Speaker 2: going in to the cloud because we have to have 243 00:10:58,920 --> 00:11:01,680 Speaker 2: it in the league office. But okay, I'll bet there. 244 00:11:01,760 --> 00:11:05,320 Speaker 2: So there's some secretive stuff that teams go through. But yes, 245 00:11:05,400 --> 00:11:09,079 Speaker 2: all the consumer sports Zala's TeamWorks. Everyone's trying to build 246 00:11:09,120 --> 00:11:12,160 Speaker 2: models uniquely. What I think is really cool about what 247 00:11:12,200 --> 00:11:15,440 Speaker 2: we do is we have a focus on how do 248 00:11:15,440 --> 00:11:18,040 Speaker 2: we make the best TV product, and so we combined 249 00:11:18,080 --> 00:11:23,000 Speaker 2: TV production, football expertise, AI, machine learning expertise, every single 250 00:11:23,000 --> 00:11:25,360 Speaker 2: one of our sciences a PhD. And we're able to 251 00:11:25,400 --> 00:11:28,560 Speaker 2: now combine all that together to create models that now 252 00:11:28,800 --> 00:11:31,760 Speaker 2: are driving a lot of the innovations that teams want 253 00:11:31,800 --> 00:11:34,240 Speaker 2: to see. We can do things that other places can't 254 00:11:34,240 --> 00:11:36,960 Speaker 2: do based off our unique team structure, and so that's 255 00:11:37,040 --> 00:11:39,280 Speaker 2: helping drive a lot of people will go Okay, now 256 00:11:39,320 --> 00:11:41,199 Speaker 2: I want this. I saw this work on TV. Now 257 00:11:41,200 --> 00:11:42,360 Speaker 2: I want to build this for myself. 258 00:11:42,760 --> 00:11:46,280 Speaker 1: Now I got to imagine that this type of technological 259 00:11:46,280 --> 00:11:47,520 Speaker 1: breakthrough the fact that you can do it in a 260 00:11:47,520 --> 00:11:49,679 Speaker 1: real time By the way, I know, the NFL is 261 00:11:49,720 --> 00:11:53,760 Speaker 1: tested out giving people access to video, you know, live 262 00:11:53,880 --> 00:11:55,839 Speaker 1: in game, and the coaches booth on the sideline usually 263 00:11:55,920 --> 00:12:00,080 Speaker 1: just photos. I imagine this would probably slow that push 264 00:12:00,080 --> 00:12:01,720 Speaker 1: a little bit for live video, right, but if you 265 00:12:01,760 --> 00:12:05,520 Speaker 1: start integrating AI into that, that probably gives some real 266 00:12:05,559 --> 00:12:08,360 Speaker 1: advantage game data coaches. Even if you know what the 267 00:12:08,600 --> 00:12:11,720 Speaker 1: play caller can be in the helmet bike to about 268 00:12:11,760 --> 00:12:14,040 Speaker 1: fifteen seconds on the play clock, right, you could probably 269 00:12:14,080 --> 00:12:18,200 Speaker 1: get some real information given to those players. If something 270 00:12:18,240 --> 00:12:20,120 Speaker 1: like this could be made available real time doing a 271 00:12:20,120 --> 00:12:21,160 Speaker 1: game to coaches. 272 00:12:21,600 --> 00:12:25,839 Speaker 2: Yeah, you know, I think I think that there'll always 273 00:12:25,880 --> 00:12:27,640 Speaker 2: be restrictions on what the coaches have to have on 274 00:12:27,679 --> 00:12:29,959 Speaker 2: the replay, even if they go to video on the sideline, 275 00:12:30,440 --> 00:12:32,360 Speaker 2: And the NFL does a great job on like making 276 00:12:32,360 --> 00:12:35,560 Speaker 2: sure there's some integrity in the game for that information. 277 00:12:35,960 --> 00:12:39,720 Speaker 2: And then their fifteen second cutoff switch is really smart 278 00:12:39,720 --> 00:12:41,720 Speaker 2: by them by having it so when there's fifteen seconds 279 00:12:41,800 --> 00:12:43,920 Speaker 2: left on the clock, they can no longer communicate directly 280 00:12:43,960 --> 00:12:46,360 Speaker 2: to the quarterback. And the way that they do their 281 00:12:46,800 --> 00:12:50,960 Speaker 2: boss spotting is there's not hyper organ style. Chip Kelly 282 00:12:51,000 --> 00:12:53,560 Speaker 2: up tempo offenses. Now he's back in the NFL, but 283 00:12:53,600 --> 00:12:56,360 Speaker 2: he's not gonna be running a play every fourteen seconds. 284 00:12:56,360 --> 00:12:59,320 Speaker 2: That's just not what the NFL game is. So I 285 00:12:59,559 --> 00:13:02,440 Speaker 2: doubt that there will be a way to like manipulate 286 00:13:02,480 --> 00:13:04,360 Speaker 2: the models to be able to help you in real time. 287 00:13:04,400 --> 00:13:06,920 Speaker 2: I think the other aspect of it is is you 288 00:13:07,040 --> 00:13:10,880 Speaker 2: have coaches that have the same viewing and ability of 289 00:13:10,920 --> 00:13:13,120 Speaker 2: all the AM models. Now they're not going to be 290 00:13:13,160 --> 00:13:15,640 Speaker 2: as perfect as an AI model, but they still have 291 00:13:15,720 --> 00:13:17,480 Speaker 2: enough experience to say, Okay, now that we know that 292 00:13:17,480 --> 00:13:20,000 Speaker 2: Harrison Smith follows, here's how we're not going to use 293 00:13:20,000 --> 00:13:23,280 Speaker 2: this indicator. We'll find something else for your man's own indicator, 294 00:13:23,720 --> 00:13:26,199 Speaker 2: or we'll always give you an out outlet option. I 295 00:13:26,280 --> 00:13:28,559 Speaker 2: think that's going to be more of a use case. 296 00:13:28,720 --> 00:13:31,120 Speaker 2: It's better for the novice fan at home to help 297 00:13:31,240 --> 00:13:32,520 Speaker 2: learn have that same mindset. 298 00:13:33,320 --> 00:13:35,680 Speaker 1: Are we going to get to player tendencies to a 299 00:13:35,720 --> 00:13:37,559 Speaker 1: point with this too? You know, I talked to the 300 00:13:37,640 --> 00:13:39,880 Speaker 1: former players are like, yeah, I knew whether or not 301 00:13:39,920 --> 00:13:42,320 Speaker 1: this guy was dropping in zone or rushing based on 302 00:13:42,679 --> 00:13:44,960 Speaker 1: and where his knuckles white. How was he leaning? You know, 303 00:13:45,080 --> 00:13:47,000 Speaker 1: was he set up a little bit outside or inside, 304 00:13:47,400 --> 00:13:49,600 Speaker 1: and little clues like that can give a players an 305 00:13:49,640 --> 00:13:52,160 Speaker 1: idea of what his matchup on an individual play is 306 00:13:52,160 --> 00:13:54,680 Speaker 1: going to do. Well, we get to a point where 307 00:13:54,760 --> 00:13:57,679 Speaker 1: these guys maybe don't have to watch the ten hours 308 00:13:57,720 --> 00:14:00,920 Speaker 1: of video on their own at night, and it's gonna 309 00:14:01,280 --> 00:14:03,360 Speaker 1: some you know, AI system is going to be able 310 00:14:03,360 --> 00:14:05,480 Speaker 1: to spit out some of these tendencies on its own. 311 00:14:06,559 --> 00:14:09,400 Speaker 2: I think it'll only help a guided learning experience because 312 00:14:10,360 --> 00:14:13,840 Speaker 2: no matter how much the teams don't have to actually 313 00:14:13,880 --> 00:14:16,880 Speaker 2: clip and make cutups of plays. That's why film was 314 00:14:16,920 --> 00:14:19,840 Speaker 2: called cutups, because they actually cut the film and reordered it. 315 00:14:20,520 --> 00:14:22,440 Speaker 2: They don't have to do that anymore. They still work 316 00:14:22,480 --> 00:14:26,200 Speaker 2: twenty hour days, and so no matter what players are 317 00:14:26,320 --> 00:14:27,880 Speaker 2: going to find a way to try not work, it 318 00:14:28,400 --> 00:14:30,600 Speaker 2: might not actually be about the work. Coaches might still 319 00:14:30,600 --> 00:14:33,560 Speaker 2: sleep in the office even though they might not have to, 320 00:14:34,280 --> 00:14:36,120 Speaker 2: just because we'll always have time in the day to 321 00:14:36,160 --> 00:14:38,480 Speaker 2: find more work and more things. I think what AI 322 00:14:38,520 --> 00:14:41,720 Speaker 2: will help do is help surface that data much better 323 00:14:41,760 --> 00:14:43,960 Speaker 2: to make you have a better learning experience and not 324 00:14:44,000 --> 00:14:46,280 Speaker 2: waste as much time trying to find the right Oh, 325 00:14:46,280 --> 00:14:47,760 Speaker 2: here's the right cutup I should watch? 326 00:14:48,000 --> 00:14:49,560 Speaker 1: Yeah, you want to remove the busy work, right, you 327 00:14:49,600 --> 00:14:51,040 Speaker 1: want to make sure that your time is used as 328 00:14:51,080 --> 00:14:53,040 Speaker 1: efficiently as possible. I'm with you on that, all right. 329 00:14:53,080 --> 00:14:55,680 Speaker 1: We talked about the AI stuff. How about next Gen Stats? 330 00:14:55,760 --> 00:14:58,480 Speaker 1: This is something that is blown up. You know, we 331 00:14:58,560 --> 00:15:01,920 Speaker 1: have access through the club portal. We see how quickly 332 00:15:02,240 --> 00:15:05,240 Speaker 1: the dots populate, They get information in their real time. 333 00:15:05,760 --> 00:15:07,680 Speaker 1: How is that able to help you? And how do 334 00:15:07,720 --> 00:15:10,240 Speaker 1: you use utilize next Gen Stats in the broadcast to 335 00:15:10,280 --> 00:15:13,240 Speaker 1: help just bring more information, data and a clearer picture 336 00:15:13,680 --> 00:15:15,479 Speaker 1: as to what's happening to the viewers. 337 00:15:15,800 --> 00:15:18,800 Speaker 2: Yeah, so they have next Gen Stats, puts an RFID 338 00:15:18,960 --> 00:15:21,040 Speaker 2: tracking chip in each player's shoulder pads. There is a 339 00:15:21,120 --> 00:15:23,280 Speaker 2: chip in the ball, there's chips in the chains, there's 340 00:15:23,320 --> 00:15:25,080 Speaker 2: chips all around the field, so that we know we 341 00:15:25,120 --> 00:15:27,200 Speaker 2: have an entire web of the field. Get that data 342 00:15:27,240 --> 00:15:30,240 Speaker 2: in real time. But the real great part about it 343 00:15:30,280 --> 00:15:32,560 Speaker 2: is the team. They have Bill Smith and Mike ban 344 00:15:32,600 --> 00:15:35,800 Speaker 2: Keegan ab Do and many others. They all are able 345 00:15:35,960 --> 00:15:38,360 Speaker 2: to then take a lot of that data create models 346 00:15:38,360 --> 00:15:39,920 Speaker 2: on top of it to give us insights that we 347 00:15:39,920 --> 00:15:43,280 Speaker 2: wouldn't have otherwise. Right, and that's what powers a lot 348 00:15:43,280 --> 00:15:45,120 Speaker 2: of our show Our show is called Prime Vision with 349 00:15:45,200 --> 00:15:48,200 Speaker 2: Next Gen Stats because next Gen powers so much of 350 00:15:48,240 --> 00:15:51,320 Speaker 2: the features that you see on our show. And I 351 00:15:51,360 --> 00:15:55,120 Speaker 2: think what I love about working with them is there 352 00:15:55,160 --> 00:15:58,920 Speaker 2: are so many different things that they're constantly working on 353 00:15:59,280 --> 00:16:01,360 Speaker 2: to make show the sure the show gets better, whether 354 00:16:01,360 --> 00:16:04,760 Speaker 2: it's getting us that data faster, improving their models, accuracy 355 00:16:05,200 --> 00:16:07,800 Speaker 2: improve giving us new models so then tell us even better, 356 00:16:07,840 --> 00:16:11,240 Speaker 2: more richer stories, and they're constantly I think what gets 357 00:16:11,280 --> 00:16:14,400 Speaker 2: lost in some of the analytics community is that this 358 00:16:14,480 --> 00:16:17,960 Speaker 2: Next Gen Stats team is a football fan of football 359 00:16:18,480 --> 00:16:21,040 Speaker 2: knowledge base first, and then there's all these analytics they 360 00:16:21,040 --> 00:16:22,960 Speaker 2: build on top of it. These guys are big, huge 361 00:16:22,960 --> 00:16:26,680 Speaker 2: football guys and girls, and they are actively trying to 362 00:16:26,680 --> 00:16:28,560 Speaker 2: make sure that we can get a better understand deeper 363 00:16:28,560 --> 00:16:29,000 Speaker 2: into the game. 364 00:16:29,440 --> 00:16:31,080 Speaker 1: If you want to know how to manage two minutes 365 00:16:31,120 --> 00:16:32,880 Speaker 1: of crunch time football, I'm your match. 366 00:16:32,920 --> 00:16:35,200 Speaker 2: But if you're wondering about a long term financial plan, 367 00:16:35,560 --> 00:16:36,800 Speaker 2: you should talk to Citizens. 368 00:16:36,920 --> 00:16:38,920 Speaker 1: Hey, I can also talk long care. I'd like to 369 00:16:39,000 --> 00:16:40,160 Speaker 1: learn about any routine. 370 00:16:40,360 --> 00:16:41,320 Speaker 2: Yes, I knew I. 371 00:16:41,240 --> 00:16:44,720 Speaker 1: Could help make sense of your money with Citizens. If 372 00:16:44,720 --> 00:16:47,400 Speaker 1: you're advising a fan at home, and you would say, 373 00:16:47,440 --> 00:16:51,200 Speaker 1: all right, here are the one, two or three next 374 00:16:51,240 --> 00:16:53,120 Speaker 1: gen stats, things that have come out over the last 375 00:16:53,160 --> 00:16:56,520 Speaker 1: few years that you found particularly useful and even helped 376 00:16:56,520 --> 00:16:58,720 Speaker 1: you understand the game a little bit better. What would 377 00:16:58,760 --> 00:16:59,800 Speaker 1: you tell them to keep an eye on? 378 00:17:00,640 --> 00:17:02,640 Speaker 2: So I would use their EPA stat And fans have 379 00:17:02,680 --> 00:17:05,400 Speaker 2: probably seeing this and it's not the Environmental Protection Agency. 380 00:17:05,480 --> 00:17:10,640 Speaker 2: It's added and I think their stat they actually include 381 00:17:11,480 --> 00:17:14,680 Speaker 2: the time left in the game. EPA was created years 382 00:17:14,720 --> 00:17:18,080 Speaker 2: ago by a guy named Virgil Carter, former Bengals quarterback, 383 00:17:19,480 --> 00:17:23,160 Speaker 2: and it was contextualizing yards based on down distance, locational 384 00:17:23,160 --> 00:17:25,280 Speaker 2: field to how valuable they are to helping you score. 385 00:17:25,359 --> 00:17:27,920 Speaker 2: And that's all it is. And people get crazed about 386 00:17:27,920 --> 00:17:30,560 Speaker 2: this stat o what this stat means nothing. It's actually 387 00:17:30,560 --> 00:17:34,800 Speaker 2: the ultimate football guy stat because football guys, no situational 388 00:17:34,840 --> 00:17:37,040 Speaker 2: football matters the most. It's more important to get ten 389 00:17:37,119 --> 00:17:39,880 Speaker 2: yards on third and eight than this third and seventeen, right, 390 00:17:39,880 --> 00:17:42,600 Speaker 2: that's the kind of the basics of it. And how 391 00:17:42,640 --> 00:17:45,600 Speaker 2: do we convert first downs? Keep the chains moving, So 392 00:17:45,640 --> 00:17:48,000 Speaker 2: that's the number one. And because they have the time 393 00:17:48,040 --> 00:17:49,400 Speaker 2: bounding so they know at the end of the game 394 00:17:49,440 --> 00:17:51,600 Speaker 2: it has different values as what you had at the 395 00:17:51,600 --> 00:17:55,440 Speaker 2: start of the game. Then you have success rate, and 396 00:17:55,480 --> 00:17:57,760 Speaker 2: I think this is the easiest one to know. It's 397 00:17:57,840 --> 00:18:00,720 Speaker 2: just saying did you get positive EPA on a play 398 00:18:00,920 --> 00:18:04,320 Speaker 2: or not? And what success rate helps is knowing, Okay, 399 00:18:04,400 --> 00:18:07,080 Speaker 2: when a player is doing something, how consistently are they 400 00:18:07,119 --> 00:18:09,440 Speaker 2: helping their team. It doesn't give you the magnitude or 401 00:18:09,480 --> 00:18:11,879 Speaker 2: how much they're helping, but you know they're consistent. So 402 00:18:12,000 --> 00:18:14,440 Speaker 2: we had a situation a couple of years ago, Zach 403 00:18:14,480 --> 00:18:19,040 Speaker 2: Wilson was, you know, six of ten for completion and 404 00:18:19,040 --> 00:18:21,119 Speaker 2: they had a sixty percent completion percentage. It's not a 405 00:18:21,119 --> 00:18:24,400 Speaker 2: great completion percentage, but it was even worse success rate. 406 00:18:24,880 --> 00:18:29,320 Speaker 2: Only two of those completions were actually successful. They all 407 00:18:29,320 --> 00:18:31,639 Speaker 2: wore other dink and dung screens. But you so, you 408 00:18:31,680 --> 00:18:34,080 Speaker 2: can't really look at completion percentage with how much short 409 00:18:34,119 --> 00:18:36,800 Speaker 2: passing there is in the NFL as a great value 410 00:18:37,040 --> 00:18:39,520 Speaker 2: or is he getting the most out of those throws 411 00:18:39,520 --> 00:18:43,640 Speaker 2: that he that he's completing. Probably the last one that 412 00:18:43,800 --> 00:18:46,520 Speaker 2: I think fans really need to buy into is the 413 00:18:46,520 --> 00:18:49,280 Speaker 2: fourth down decision making. Every team is not going for 414 00:18:49,320 --> 00:18:52,240 Speaker 2: it on fourth down. Next Gen has a fourth down 415 00:18:52,240 --> 00:18:54,679 Speaker 2: Decision Guide to help teams make the best decisions on 416 00:18:54,680 --> 00:18:56,600 Speaker 2: when to go for it and not. And it's taking 417 00:18:56,640 --> 00:19:00,000 Speaker 2: win probability and having decision trees to help make those 418 00:19:00,320 --> 00:19:02,880 Speaker 2: those best decisions. And what I find value is there's 419 00:19:02,880 --> 00:19:06,119 Speaker 2: a thing called optimal call rate, which how often you 420 00:19:06,160 --> 00:19:09,639 Speaker 2: are giving a two percent probability to go forward or 421 00:19:09,680 --> 00:19:13,120 Speaker 2: more in like you're not trailing by three scores scenario, 422 00:19:14,080 --> 00:19:16,879 Speaker 2: and are you how often is your coach going for 423 00:19:16,920 --> 00:19:19,560 Speaker 2: it when given those chances? And why do we use 424 00:19:19,600 --> 00:19:21,800 Speaker 2: that two percent? Because sixty percent of the time when 425 00:19:21,840 --> 00:19:24,160 Speaker 2: a coach is given a two percent go they go 426 00:19:24,240 --> 00:19:27,560 Speaker 2: for it. Well, that helps us in a fourth and 427 00:19:28,280 --> 00:19:31,600 Speaker 2: one from your own nine when it's saying go for 428 00:19:31,680 --> 00:19:34,560 Speaker 2: it by four percent win probability, people like why are 429 00:19:34,600 --> 00:19:36,359 Speaker 2: you going for it on your own end of the field. 430 00:19:37,320 --> 00:19:39,480 Speaker 2: Those models help us learn more about the game because 431 00:19:39,480 --> 00:19:41,480 Speaker 2: if you punt from that area, there's a safety chance 432 00:19:41,480 --> 00:19:43,679 Speaker 2: because you're punning from inside your own end zone, or 433 00:19:43,720 --> 00:19:45,080 Speaker 2: the ball is going to be given back to their 434 00:19:45,080 --> 00:19:47,480 Speaker 2: opponent on your own forty five, so they're already in 435 00:19:47,520 --> 00:19:50,359 Speaker 2: a great scoring scenario. So there's all these different factors 436 00:19:50,400 --> 00:19:53,280 Speaker 2: that get counted in that that win probability model is 437 00:19:53,359 --> 00:19:55,280 Speaker 2: baking in And so I think the fans can help 438 00:19:55,400 --> 00:19:57,439 Speaker 2: understand why is my team going for it in this 439 00:19:57,480 --> 00:19:59,560 Speaker 2: obscure situation? The model will help you out. 440 00:19:59,560 --> 00:20:01,280 Speaker 1: All right, let me follow up on that real quick, 441 00:20:01,320 --> 00:20:02,480 Speaker 1: and then I have one more question then we can 442 00:20:02,520 --> 00:20:04,199 Speaker 1: wrap up here. Sam, this has been awesome. Thank you. 443 00:20:06,320 --> 00:20:09,960 Speaker 1: What is considered a significant number? You just mentioned two 444 00:20:10,040 --> 00:20:13,560 Speaker 1: percent in terms of the win loss probability shift on 445 00:20:13,680 --> 00:20:16,080 Speaker 1: fourth downs? When is it too close to call? And 446 00:20:16,119 --> 00:20:17,960 Speaker 1: it's like, all right, I could see it going either way. 447 00:20:18,240 --> 00:20:20,439 Speaker 1: Why did you kind of nail two percent? Because I 448 00:20:20,440 --> 00:20:23,320 Speaker 1: think people from the outside end would think two percent, 449 00:20:23,400 --> 00:20:26,359 Speaker 1: that's that's pretty small. If it's a two percent shift, 450 00:20:26,359 --> 00:20:28,720 Speaker 1: that's only going to impact what two out of every 451 00:20:28,800 --> 00:20:31,200 Speaker 1: one hundred games. You know, if if you kind of 452 00:20:31,200 --> 00:20:32,800 Speaker 1: do the math on it, a. 453 00:20:32,840 --> 00:20:36,080 Speaker 2: Two percent we pick out because MGS has that as 454 00:20:36,119 --> 00:20:38,080 Speaker 2: sixty percent of the time they go for it, right, 455 00:20:38,160 --> 00:20:41,000 Speaker 2: So this is a choice that teams would make if 456 00:20:41,040 --> 00:20:44,359 Speaker 2: it was an obvious situation. There's also a lot of 457 00:20:44,400 --> 00:20:47,840 Speaker 2: noise in data, and so you know, we don't have 458 00:20:48,000 --> 00:20:51,719 Speaker 2: every piece of information about every single player on the field, 459 00:20:52,240 --> 00:20:55,760 Speaker 2: or you know, there might be an injury to a 460 00:20:55,800 --> 00:20:59,040 Speaker 2: player earlier in the game on a QB sneak and 461 00:20:59,240 --> 00:21:01,800 Speaker 2: there might not be an information to know. Okay, hey, 462 00:21:02,040 --> 00:21:04,680 Speaker 2: how obvious is the scenario that the court since the 463 00:21:04,760 --> 00:21:07,360 Speaker 2: quarterback's not likely to QB sneak again, and that's such 464 00:21:07,400 --> 00:21:10,480 Speaker 2: a good play on fourth and short. Okay, That two 465 00:21:10,560 --> 00:21:12,840 Speaker 2: percent give us that barrier where we know coaches like 466 00:21:12,880 --> 00:21:16,440 Speaker 2: to go for it. And on top of that, it's 467 00:21:16,600 --> 00:21:19,240 Speaker 2: enough distance away from zero, so sometimes you won't see 468 00:21:19,280 --> 00:21:21,280 Speaker 2: us put it on screen if there's not a two 469 00:21:21,359 --> 00:21:24,199 Speaker 2: percent or more. If it's point five percent, go for it. 470 00:21:24,720 --> 00:21:28,760 Speaker 2: I go, coach, do your thing right two percent, I'm like, okay, 471 00:21:28,800 --> 00:21:32,119 Speaker 2: four percent. At that point, that's over eighty percent of 472 00:21:32,119 --> 00:21:34,159 Speaker 2: the time coaches go for it. That's where I'm like, 473 00:21:34,560 --> 00:21:37,159 Speaker 2: you better have a play ready for this. And so 474 00:21:37,240 --> 00:21:39,760 Speaker 2: a large part of this is when I would advise 475 00:21:39,800 --> 00:21:41,919 Speaker 2: teams before I had this job. I'd advise teams on 476 00:21:42,200 --> 00:21:45,040 Speaker 2: fourth down decision making and things like that. It's I'm 477 00:21:45,080 --> 00:21:47,000 Speaker 2: not telling you to go for it every time. The 478 00:21:47,040 --> 00:21:50,120 Speaker 2: model says, I'm telling you you can make a lot 479 00:21:50,600 --> 00:21:52,959 Speaker 2: or win game, win games a lot more games if 480 00:21:53,000 --> 00:21:56,040 Speaker 2: you have a better fourth down plan. Because in football, 481 00:21:56,040 --> 00:21:59,000 Speaker 2: we have a short yardage plan, a red zone plan, 482 00:21:59,320 --> 00:22:01,639 Speaker 2: a third in the lie long plane. Let's call it 483 00:22:01,680 --> 00:22:03,680 Speaker 2: those third and longs. By going forward on more fourth 484 00:22:03,720 --> 00:22:06,119 Speaker 2: and shorts and mediums. So now you have a fourth 485 00:22:06,160 --> 00:22:09,320 Speaker 2: down plan. So again we don't know the play calls. 486 00:22:09,880 --> 00:22:12,679 Speaker 2: Giants fans remember when the team had never practiced the 487 00:22:12,680 --> 00:22:14,920 Speaker 2: Toush push and then they tried the touch push it 488 00:22:15,000 --> 00:22:16,520 Speaker 2: live in a game for the first time, and it 489 00:22:16,520 --> 00:22:20,080 Speaker 2: didn't work. Practice the plays, spend time in training camp 490 00:22:20,080 --> 00:22:22,160 Speaker 2: and build a better fourth down plan so you're ready 491 00:22:22,160 --> 00:22:24,600 Speaker 2: for those situations where you can win. We saw the 492 00:22:25,240 --> 00:22:27,680 Speaker 2: Commanders have a great run this season. As much to 493 00:22:27,720 --> 00:22:30,560 Speaker 2: your listeners probably gonna hate me talking about this, a 494 00:22:30,640 --> 00:22:32,800 Speaker 2: large part of their ability because they were the best. 495 00:22:32,880 --> 00:22:35,200 Speaker 2: They had some of the most I think top three 496 00:22:35,440 --> 00:22:37,680 Speaker 2: win probability added on fourth downs. 497 00:22:37,800 --> 00:22:41,440 Speaker 1: Weren't they nine fourth down conversions or something ridiculous this year? 498 00:22:41,800 --> 00:22:44,000 Speaker 2: It was crazy this year. But it wasn't just how 499 00:22:44,560 --> 00:22:47,440 Speaker 2: good they were converting, it's how when they picked their 500 00:22:47,480 --> 00:22:51,480 Speaker 2: spots they gained so much win probability off of those spots. 501 00:22:51,560 --> 00:22:54,119 Speaker 2: Zach Ertz former team in Mind, one of my close friends, 502 00:22:54,440 --> 00:22:56,520 Speaker 2: was so clutch for them in the playoffs, adding win 503 00:22:56,600 --> 00:23:01,400 Speaker 2: probability on those fourth downs and so you can extend drives, 504 00:23:01,640 --> 00:23:03,760 Speaker 2: you can keep the ball out of your opponents' hands. 505 00:23:04,119 --> 00:23:07,000 Speaker 2: It's like a great defense, it's a great fourth down offense. 506 00:23:07,480 --> 00:23:09,480 Speaker 1: Let me just circle back to success rate, because usually 507 00:23:09,480 --> 00:23:11,640 Speaker 1: success if you don't get a first down on third down, 508 00:23:11,840 --> 00:23:14,920 Speaker 1: it's not considered successful play. Right but given the high 509 00:23:15,000 --> 00:23:17,560 Speaker 1: level of success right now in fourth and one, if 510 00:23:17,600 --> 00:23:20,320 Speaker 1: you gain nine yards on a third and ten, does 511 00:23:20,359 --> 00:23:23,600 Speaker 1: that then become a successful play because it gives you 512 00:23:23,640 --> 00:23:25,240 Speaker 1: a chance to go for it? How has that kind 513 00:23:25,240 --> 00:23:27,919 Speaker 1: of affected the successful play model, which I know there 514 00:23:27,960 --> 00:23:29,600 Speaker 1: are different ones too, by the way, the traditional one 515 00:23:29,680 --> 00:23:31,720 Speaker 1: is get half the down or distance. That's kind of 516 00:23:31,720 --> 00:23:34,520 Speaker 1: the old one you mentioned, the POSITIVEVPA. How does the 517 00:23:34,960 --> 00:23:37,320 Speaker 1: kind of people being more willing to go for it 518 00:23:37,359 --> 00:23:40,359 Speaker 1: on fourth and short impact that success rate model? 519 00:23:41,119 --> 00:23:42,880 Speaker 2: We lean on the next gen stats model to give 520 00:23:42,920 --> 00:23:44,679 Speaker 2: us whether or not they got POSITIVEVPR or not. And 521 00:23:44,680 --> 00:23:47,320 Speaker 2: because it's baking the time, then they're going to produce 522 00:23:47,320 --> 00:23:49,359 Speaker 2: a different data point, so we can look at it. 523 00:23:49,400 --> 00:23:51,639 Speaker 2: I can pull all the EPA data by sound, distance, 524 00:23:51,680 --> 00:23:54,199 Speaker 2: location on field, and I can average it out and 525 00:23:54,240 --> 00:23:55,679 Speaker 2: I could give you a number if you gave me 526 00:23:55,800 --> 00:23:58,080 Speaker 2: twenty minutes to go and find the exact scenario you 527 00:23:58,560 --> 00:24:00,360 Speaker 2: gave and so oh would this be pause the VP 528 00:24:00,520 --> 00:24:02,880 Speaker 2: or not? But because we're gonna lean on the game 529 00:24:03,040 --> 00:24:06,000 Speaker 2: clock to it, it adds an element to where I 530 00:24:06,080 --> 00:24:09,440 Speaker 2: let that model tell me post this is it now. 531 00:24:09,480 --> 00:24:12,800 Speaker 2: If you're the Eagles and you're gonna have the touch push, yeah, 532 00:24:12,800 --> 00:24:15,080 Speaker 2: I'm gonna say it's positive, VPA. You helped your team out. 533 00:24:15,119 --> 00:24:17,560 Speaker 2: You've got in fourth and one and a half or less. 534 00:24:17,560 --> 00:24:19,840 Speaker 2: That's when they like to use the push. If you're 535 00:24:19,840 --> 00:24:21,760 Speaker 2: a team that doesn't have a QB sneak team, I 536 00:24:21,760 --> 00:24:25,960 Speaker 2: wouldn't say it was successful play right, And so that's why, 537 00:24:26,359 --> 00:24:28,320 Speaker 2: And really it's not necessarily about the push. It's more 538 00:24:28,359 --> 00:24:32,960 Speaker 2: about the sneak. And I studied the CFL extensively, and 539 00:24:33,000 --> 00:24:35,359 Speaker 2: they have a one yard neutral zone instead of our 540 00:24:35,359 --> 00:24:37,840 Speaker 2: football length neutral zone we have in the NFL, and 541 00:24:37,880 --> 00:24:40,720 Speaker 2: it was a guarantee to get even two yards on 542 00:24:40,800 --> 00:24:43,240 Speaker 2: their sneak up there, and yet in the NFL we're 543 00:24:43,240 --> 00:24:45,880 Speaker 2: have four downs, and so we really have value out 544 00:24:45,880 --> 00:24:47,600 Speaker 2: of going for on the sneak even with just if 545 00:24:47,600 --> 00:24:50,040 Speaker 2: one year the football length neutral zone. 546 00:24:49,920 --> 00:24:51,600 Speaker 1: Which is probably why it's important for these teams to 547 00:24:51,640 --> 00:24:54,680 Speaker 1: have individualized models to their teams rather than using a 548 00:24:54,760 --> 00:24:58,199 Speaker 1: universal one. Correct exactly, that makes sense. I find a 549 00:24:58,280 --> 00:25:01,520 Speaker 1: question just pure advertise here, Sam, tell the folks what 550 00:25:01,600 --> 00:25:03,439 Speaker 1: you guys have new coming up on Prime Vision this 551 00:25:03,520 --> 00:25:05,639 Speaker 1: year that people should be looking forward to and just 552 00:25:05,680 --> 00:25:08,280 Speaker 1: why they should tune into the Prime Vision cast on 553 00:25:08,400 --> 00:25:09,240 Speaker 1: Thursday Night Football. 554 00:25:09,600 --> 00:25:11,040 Speaker 2: Yeah. I think Prime Vision you're going to see a 555 00:25:11,040 --> 00:25:13,600 Speaker 2: lot of things that are kind of dreams for what 556 00:25:13,640 --> 00:25:15,760 Speaker 2: we're trying to be able to showcase. We're going to 557 00:25:15,800 --> 00:25:18,520 Speaker 2: be able to get you closer to the game, understand 558 00:25:18,560 --> 00:25:20,200 Speaker 2: some of the more nuanced part of the game, that's 559 00:25:20,240 --> 00:25:22,320 Speaker 2: what we're really excited about, and then some of the 560 00:25:22,359 --> 00:25:24,760 Speaker 2: more information that's helping teams make decisions near the end 561 00:25:24,800 --> 00:25:26,840 Speaker 2: of the game. You're going to get even deeper into 562 00:25:26,880 --> 00:25:29,680 Speaker 2: the game by watching Prime Vision. So every year we're 563 00:25:29,720 --> 00:25:31,800 Speaker 2: working on new features, We're listening to what the audience 564 00:25:31,880 --> 00:25:33,480 Speaker 2: is saying, I want to get closer and deeper to 565 00:25:33,520 --> 00:25:35,200 Speaker 2: the game in this part of it, and then we'll 566 00:25:35,200 --> 00:25:36,760 Speaker 2: work on it. So a lot more to come. 567 00:25:37,240 --> 00:25:38,920 Speaker 1: We thanks aam for joining us right here in the 568 00:25:38,960 --> 00:25:41,760 Speaker 1: Hackensack Comriney Health Podcast studio. Keep getting better and just 569 00:25:41,800 --> 00:25:45,639 Speaker 1: a reminder the only place you can find the Giants 570 00:25:45,960 --> 00:25:49,840 Speaker 1: and Patriots on television. Thursday Night coverage begins at eight 571 00:25:49,840 --> 00:25:53,159 Speaker 1: pm on Amazon Prime. Coverage of that game nationally. Make 572 00:25:53,160 --> 00:25:54,720 Speaker 1: sure you go check that out. They do a great 573 00:25:54,800 --> 00:25:58,320 Speaker 1: job covering the games. Make sure you go and watch 574 00:25:58,400 --> 00:26:01,200 Speaker 1: that game and Amazon Prime, and of course Giants will 575 00:26:01,200 --> 00:26:03,960 Speaker 1: be on Thursday Night against the Eagles on Amazon Prime 576 00:26:04,040 --> 00:26:06,399 Speaker 1: during the regular season as well. We thanks Sam for 577 00:26:06,480 --> 00:26:08,240 Speaker 1: joining us. We'll see you next time on the Johns 578 00:26:08,240 --> 00:26:10,480 Speaker 1: Tottle Podcast. Brought to you by Citizens Official Bank of 579 00:26:10,520 --> 00:26:11,000 Speaker 1: the Giants.