WEBVTT 00:00:00.000 --> 00:00:06.360 How often do you rely on a gut feeling when you're 00:00:06.360 --> 00:00:08.240 trying to predict how something is going to turn 00:00:08.240 --> 00:00:10.599 out? Oh, constantly. We all do it. Right. I mean, 00:00:10.619 --> 00:00:12.060 we do it all the time. You're watching a game 00:00:12.060 --> 00:00:14.439 and you just confidently declare which team is 00:00:14.439 --> 00:00:16.820 going to win the championship based purely on, 00:00:16.879 --> 00:00:18.920 well, on the fact that they just look like winners. 00:00:19.179 --> 00:00:22.679 Yeah. Or they have that that mysterious momentum. 00:00:22.920 --> 00:00:26.719 Exactly. Momentum. And for decades, that highly 00:00:26.719 --> 00:00:29.719 subjective, intuition heavy approach paired with 00:00:29.719 --> 00:00:32.840 some. frankly, very rigid traditional ranking 00:00:32.840 --> 00:00:35.299 systems. That was exactly how we predicted outcomes 00:00:35.299 --> 00:00:37.899 in sports. It was the only way, really. But today 00:00:37.899 --> 00:00:40.939 we are going to look at how a purely quantitative 00:00:40.939 --> 00:00:44.299 math -based approach completely dismantled that 00:00:44.299 --> 00:00:47.119 mindset. specifically in the world of men's college 00:00:47.119 --> 00:00:49.399 basketball. Which is a massive paradigm shift. 00:00:49.560 --> 00:00:51.320 Honestly, when you look at the evolution of how 00:00:51.320 --> 00:00:53.579 we process information, this isn't just a sports 00:00:53.579 --> 00:00:56.140 story. No, not at all. It's a prime example of 00:00:56.140 --> 00:00:59.140 data synthesis and critical thinking in an era 00:00:59.140 --> 00:01:01.759 where we are constantly drowning in information 00:01:01.759 --> 00:01:05.099 overload. Well said. So the source material we're 00:01:05.099 --> 00:01:08.099 working from today is a Wikipedia article detailing 00:01:08.099 --> 00:01:11.040 the Pomeroy College basketball ratings. Better 00:01:11.040 --> 00:01:13.459 known to most people as Ken Palm. Ken Palm. Exactly. 00:01:13.760 --> 00:01:16.599 And our mission for this deep dive is to unpack 00:01:16.599 --> 00:01:20.760 how a totally free online rating system revolutionized 00:01:20.760 --> 00:01:23.500 predictive modeling, how it disrupted the establishment 00:01:23.500 --> 00:01:26.540 and managed to capture the attention of the highest 00:01:26.540 --> 00:01:29.140 levels of mainstream media. We're essentially 00:01:29.140 --> 00:01:32.079 looking at the jump from niche internet math 00:01:32.079 --> 00:01:35.819 to a foundational analytical tool. Right. We're 00:01:35.819 --> 00:01:37.900 exploring how someone looked at an established 00:01:37.900 --> 00:01:40.579 accepted system, realized it had massive blind 00:01:40.579 --> 00:01:43.780 spots, and then used objective math to cut through 00:01:43.780 --> 00:01:46.060 all that noise to find a more accurate reality. 00:01:46.319 --> 00:01:48.959 Okay, let's unpack this. Let's start with the 00:01:48.959 --> 00:01:51.560 origins and the math behind the madness. Sure. 00:01:51.640 --> 00:01:53.640 So we're looking at predictive ratings for American 00:01:53.640 --> 00:01:55.920 men's college basketball. And these are published 00:01:55.920 --> 00:01:58.400 by Ken Pomeroy. Hence the name Ken Palm. Right. 00:01:58.540 --> 00:02:00.659 But the most striking thing about this, especially 00:02:00.659 --> 00:02:02.739 if we go back to its early days, is the presentation. 00:02:03.040 --> 00:02:06.500 Oh, it's so bare bones. It really is. He publishes 00:02:06.500 --> 00:02:09.939 these completely free of charge online at KenPalm 00:02:09.939 --> 00:02:13.400 .com. And if you visit the site, the logo is 00:02:13.400 --> 00:02:15.919 just. brilliantly understated. It's literally 00:02:15.919 --> 00:02:18.780 just the text of the URL. Exactly. It's literally 00:02:18.780 --> 00:02:21.400 just the logo of the website itself. No flashy 00:02:21.400 --> 00:02:24.060 graphics. No aggressive branding. None of that. 00:02:24.099 --> 00:02:26.439 It's just a wall of data. And he first started 00:02:26.439 --> 00:02:29.419 publishing these ratings way back in 2003. Man, 00:02:29.479 --> 00:02:32.879 2003. That is, I mean, that's the dark ages for 00:02:32.879 --> 00:02:34.960 mainstream sports analytics. Totally different 00:02:34.960 --> 00:02:37.759 landscape. But he wasn't just throwing random 00:02:37.759 --> 00:02:40.460 numbers on a wall to see what stuck. There was 00:02:40.460 --> 00:02:43.419 a very specific mathematical engine powering 00:02:43.439 --> 00:02:46.680 the entire system. Yes. Our source notes that 00:02:46.680 --> 00:02:48.599 his sports rating system is based on something 00:02:48.599 --> 00:02:52.159 called the Pythagorean expectation. Though Pomeroy 00:02:52.159 --> 00:02:53.979 makes some of his own specific adjustments to 00:02:53.979 --> 00:02:55.580 it. Right. He had to tweak it for basketball. 00:02:55.819 --> 00:02:57.719 Yeah, because if you follow sports analytics, 00:02:57.919 --> 00:02:59.759 you probably know that the Pythagorean expectation 00:02:59.759 --> 00:03:02.120 originally comes from baseball. Bill James, right? 00:03:02.199 --> 00:03:05.020 Yep. It essentially takes runs scored and runs 00:03:05.020 --> 00:03:07.460 allowed, runs them through an exponent, and spits 00:03:07.460 --> 00:03:10.889 out an expected winning percentage. So Pomeroy 00:03:10.889 --> 00:03:13.629 adapted this for college basketball. Right. He 00:03:13.629 --> 00:03:16.849 transitioned it to points scored versus points 00:03:16.849 --> 00:03:20.030 allowed. And the fundamental logic here is that 00:03:20.030 --> 00:03:23.689 a team's win -loss record doesn't always tell 00:03:23.689 --> 00:03:26.349 the truth about how good they actually are. That's 00:03:26.349 --> 00:03:28.650 the core of it right there. Because a win -loss 00:03:28.650 --> 00:03:32.699 record is heavily influenced by luck. The sequence 00:03:32.699 --> 00:03:35.599 of games. Or winning several close one possession 00:03:35.599 --> 00:03:38.340 games that could have easily gone the other way. 00:03:38.500 --> 00:03:40.860 Exactly. A bad call by a ref, a lucky bounce. 00:03:41.139 --> 00:03:44.460 The Pythagorean expectation strips all that away. 00:03:44.740 --> 00:03:46.960 It calculates what your winning percentage should 00:03:46.960 --> 00:03:49.020 be based on your fundamental underlying efficiency. 00:03:49.340 --> 00:03:51.060 It's looking under the hood. It really is. It 00:03:51.060 --> 00:03:53.379 asks, based on the core mechanics of how you 00:03:53.379 --> 00:03:55.699 are outscoring your opponents or being outscored, 00:03:55.819 --> 00:03:58.139 what is your true baseline level of performance? 00:03:58.300 --> 00:04:00.770 Right. And then Pomeroy added his own adjustments. 00:04:00.949 --> 00:04:03.270 He factored in things like the pace of play to 00:04:03.270 --> 00:04:05.189 make it highly specific to the college basketball 00:04:05.189 --> 00:04:07.569 environment. So it basically separates the signal 00:04:07.569 --> 00:04:09.729 from the noise. Precisely. Like you might have 00:04:09.729 --> 00:04:13.449 a team that is 10 -0, but they won every single 00:04:13.449 --> 00:04:15.969 game by a single point against totally mediocre 00:04:15.969 --> 00:04:18.329 competition. They look great on paper. Right, 00:04:18.389 --> 00:04:20.709 but they're creeping by. Then you have a team 00:04:20.709 --> 00:04:24.290 that is 7 -3, but their three losses were on 00:04:24.290 --> 00:04:27.189 crazy last -second shots, and their seven wins 00:04:27.189 --> 00:04:29.850 were absolute 30 -point blowouts. Right. And 00:04:29.850 --> 00:04:32.370 traditional metrics might rank that 10 -0 team 00:04:32.370 --> 00:04:34.730 much higher. But the Pythagorean expectation 00:04:34.730 --> 00:04:37.509 looks at the point differentials and identifies 00:04:37.509 --> 00:04:41.430 the 7 -3 team as the inherently stronger squad 00:04:41.430 --> 00:04:43.470 moving forward. What's fascinating here is that 00:04:43.470 --> 00:04:46.560 this mathematical approach wasn't just an isolated 00:04:46.560 --> 00:04:49.839 experiment. How so? Well, there's a broader lineage 00:04:49.839 --> 00:04:52.300 of advanced analytics converging at this exact 00:04:52.300 --> 00:04:55.180 time. Our source points out that variations of 00:04:55.180 --> 00:04:57.680 this Pythagorean expectation are also used in 00:04:57.680 --> 00:05:00.800 basketball by noted statisticians like Dean Oliver 00:05:00.800 --> 00:05:03.920 and John Hollinger. Oh, right. Dean Oliver essentially 00:05:03.920 --> 00:05:06.839 wrote the book on basketball on paper, and John 00:05:06.839 --> 00:05:09.079 Hollinger created the player efficiency rating 00:05:09.079 --> 00:05:12.649 for the NBA. the per exactly so when you see 00:05:12.649 --> 00:05:16.810 a pattern of analytical minds independently utilizing 00:05:16.810 --> 00:05:20.730 similar mathematical models to evaluate a game 00:05:20.730 --> 00:05:24.449 it validates the methodology it proves it's not 00:05:24.449 --> 00:05:27.529 just a fluke exactly it demonstrates that beneath 00:05:27.529 --> 00:05:30.920 the chaos of a live basketball game the bad referee 00:05:30.920 --> 00:05:34.240 calls, the lucky bounces off the rim, there is 00:05:34.240 --> 00:05:37.139 a measurable mathematical pulse that dictates 00:05:37.139 --> 00:05:39.600 long -term success. Which brings us to the actual 00:05:39.600 --> 00:05:41.720 results of all this math, because, you know, 00:05:41.800 --> 00:05:44.420 having a theory is one thing, but proving it 00:05:44.420 --> 00:05:46.420 works is another. The proof is in the pudding. 00:05:46.540 --> 00:05:49.100 Here's where it gets really interesting. We have 00:05:49.100 --> 00:05:51.379 a source from a 2011 article in the New York 00:05:51.379 --> 00:05:54.180 Times. Specifically, it was written by Nate Silger 00:05:54.180 --> 00:05:56.779 on his FiveThirtyEight blog. Oh, Nate Silger, 00:05:56.800 --> 00:05:59.000 yeah. A heavy hitter in the analytics world. 00:05:59.160 --> 00:06:01.540 Definitely. And in this article, Silver noted 00:06:01.540 --> 00:06:04.199 that the Pomeroy College basketball ratings boasted 00:06:04.199 --> 00:06:08.079 a massive 73 % success rate. 73%. In college 00:06:08.079 --> 00:06:10.379 basketball, that is a highly significant figure. 00:06:10.600 --> 00:06:12.439 Let me play devil's advocate for a second, though. 00:06:12.540 --> 00:06:15.339 Go for it. 73 % means the model is getting it 00:06:15.339 --> 00:06:18.839 wrong 27 % of the time. True. It's missing more 00:06:18.839 --> 00:06:21.720 than one in four games. If I took my car to a 00:06:21.720 --> 00:06:24.459 mechanic and he fixed it 73 % of the time, I'd 00:06:24.459 --> 00:06:27.879 find a new mechanic. Yeah, fair point. So why 00:06:27.879 --> 00:06:30.279 was the sports analytics community looking at 00:06:30.279 --> 00:06:33.459 a 73 % success rate and treating it like a complete 00:06:33.459 --> 00:06:36.279 revolution? Because you have to consider the 00:06:36.279 --> 00:06:38.040 baseline of the environment you're operating 00:06:38.040 --> 00:06:41.680 in. Yeah. College basketball is inherently chaotic. 00:06:42.160 --> 00:06:44.500 Pure chaos sometimes. You're dealing with 18 00:06:44.500 --> 00:06:47.560 to 22 -year -old kids playing in highly emotional 00:06:47.560 --> 00:06:50.160 environments where variance is just massive. 00:06:50.319 --> 00:06:52.519 100 % success rate is mathematically impossible. 00:06:52.959 --> 00:06:55.360 Right. In predictive modeling, whether you're 00:06:55.360 --> 00:06:57.839 looking at sabermetrics in baseball or the ELO 00:06:57.839 --> 00:07:01.379 family of ratings like GLICO or DWZ used in chess, 00:07:01.620 --> 00:07:04.360 you are measuring yourself against the prevailing 00:07:04.360 --> 00:07:07.350 standard. not against absolute perfection. Ah, 00:07:07.589 --> 00:07:09.829 okay. And the prevailing standard at the time 00:07:09.829 --> 00:07:13.709 was the RPI. The infamous RPI. Yes. The source 00:07:13.709 --> 00:07:17.009 explicitly notes that Pomeroy's 73 % success 00:07:17.009 --> 00:07:19.529 rate was 2 % better than the traditional rating 00:07:19.529 --> 00:07:22.089 percentage index, or the RPI. Which was huge. 00:07:22.209 --> 00:07:25.069 Right, because the RPI was the entrenched official 00:07:25.069 --> 00:07:27.750 standard used by the NCAA Tournament Selection 00:07:27.750 --> 00:07:31.730 Committee for decades. It was the absolute goliath 00:07:31.730 --> 00:07:34.170 of college basketball metrics. It was gospel. 00:07:34.759 --> 00:07:37.459 And this raises an important question. To understand 00:07:37.459 --> 00:07:40.079 why a 2 % edge is an earthquake in predictive 00:07:40.079 --> 00:07:42.959 modeling, we have to look at how the RPI actually 00:07:42.959 --> 00:07:46.879 functioned and why Pomeroy's system exploited 00:07:46.879 --> 00:07:49.540 its flaws. Let's break that down. So the RPI 00:07:49.540 --> 00:07:51.860 was entirely based on winning percentage, your 00:07:51.860 --> 00:07:53.980 opponent's winning percentage, and your opponent's 00:07:53.980 --> 00:07:57.040 opponent's winning percentage. That is it. It 00:07:57.040 --> 00:07:58.899 only looked at who won and who lost. Exactly. 00:07:58.980 --> 00:08:01.000 It completely ignored the margin of victory. 00:08:01.180 --> 00:08:03.579 Which is a massive blind spot. A fatal flaw. 00:08:04.189 --> 00:08:06.689 Think about it. If Team A beats a top -tier opponent 00:08:06.689 --> 00:08:09.610 by 40 points, and Team B beats that same opponent 00:08:09.610 --> 00:08:11.949 by one point in overtime after a terrible referee 00:08:11.949 --> 00:08:14.329 call... The RPI treated both of those outcomes 00:08:14.329 --> 00:08:17.310 as exactly identical. Yes. It registered as a 00:08:17.310 --> 00:08:19.230 win against a strong opponent for both teams. 00:08:19.509 --> 00:08:22.769 No distinction. But Ken Palm's adjusted Pythagorean 00:08:22.769 --> 00:08:25.790 math factored in the efficiency and margin of 00:08:25.790 --> 00:08:29.529 that game. It recognized that Team A was fundamentally 00:08:29.529 --> 00:08:32.230 dominant, while Team B was really fortunate. 00:08:32.750 --> 00:08:35.230 So by capturing that nuance, Pomeroy squeezed 00:08:35.230 --> 00:08:38.690 out an extra 2 % of predictive accuracy over 00:08:38.690 --> 00:08:41.470 the institutional standard? Precisely. And in 00:08:41.470 --> 00:08:43.710 the world of win probability and sports rating 00:08:43.710 --> 00:08:47.129 systems, 2 % is everything. You know, a 2 % edge 00:08:47.129 --> 00:08:50.149 over the RPI might sound like a rounding error 00:08:50.149 --> 00:08:52.970 to someone completely outside of analytics. But 00:08:52.970 --> 00:08:55.250 I want you to think about this in the context 00:08:55.250 --> 00:08:58.269 of your own life or profession. It applies everywhere. 00:08:58.529 --> 00:09:00.629 It really does. I know from personal experience, 00:09:00.789 --> 00:09:03.129 I used to lose my office bracket pool every single 00:09:03.129 --> 00:09:05.230 year because I relied on those traditional metrics. 00:09:05.549 --> 00:09:07.549 You and millions of others. I would pick a team 00:09:07.549 --> 00:09:09.830 with a shiny 30 -2 record because they look like 00:09:09.830 --> 00:09:12.370 winners on it. completely ignoring that they 00:09:12.370 --> 00:09:14.929 were barely squeaking by inferior teams. The 00:09:14.929 --> 00:09:16.970 classic trap. And then I'd lose to the one guy 00:09:16.970 --> 00:09:19.370 in the office who just blindly followed efficiency 00:09:19.370 --> 00:09:23.110 data. That 2 % edge in predictive modeling changes 00:09:23.110 --> 00:09:26.309 the whole game. In finance, a trading algorithm 00:09:26.309 --> 00:09:29.009 that is 2 % more accurate than the market standard 00:09:29.009 --> 00:09:31.950 generates billions of dollars. In sports betting, 00:09:32.129 --> 00:09:35.070 a 2 % edge is the difference between going bankrupt 00:09:35.070 --> 00:09:38.429 and owning the casino. It compounds over time. 00:09:38.789 --> 00:09:40.409 It really does. But, you know, when you have 00:09:40.409 --> 00:09:43.990 a grassroots math -based system outperforming 00:09:43.990 --> 00:09:46.190 the traditional official index, how does the 00:09:46.190 --> 00:09:48.889 establishment react? Usually not well. Right. 00:09:49.309 --> 00:09:52.470 The RPI was safe. It was institutional. The Pomeroy 00:09:52.470 --> 00:09:54.950 ratings represented a disruptive innovation that 00:09:54.950 --> 00:09:57.110 fundamentally challenged the established criteria 00:09:57.110 --> 00:10:00.590 for evaluating team performance. And historically, 00:10:00.950 --> 00:10:03.490 large institutions do not like being told their 00:10:03.490 --> 00:10:05.909 legacy systems are flawed by a guy running a 00:10:05.909 --> 00:10:07.870 free website. They usually fight it tooth and 00:10:07.870 --> 00:10:10.110 nail. They do. But what makes this story unique 00:10:10.110 --> 00:10:12.909 is a transition from niche math to a mainstream 00:10:12.909 --> 00:10:15.750 media phenomenon. The establishment ultimately 00:10:15.750 --> 00:10:18.029 surrendered to the data. They had no choice. 00:10:18.429 --> 00:10:21.169 By the 2010s, this predictive model completely 00:10:21.169 --> 00:10:23.769 broke out of the math nerd bubble and staged 00:10:23.769 --> 00:10:26.509 a full media takeover. The adoption rate among 00:10:26.509 --> 00:10:29.490 traditional journalists was staggering. Our source 00:10:29.490 --> 00:10:32.990 lists the specific diverse outlets where Pomeroy 00:10:32.990 --> 00:10:35.350 is routinely mentioned or interviewed. And the 00:10:35.350 --> 00:10:37.889 list shows just how widespread his influence 00:10:37.889 --> 00:10:40.110 really became. It wasn't just blogs anymore. 00:10:40.269 --> 00:10:43.850 No. We are talking about ESPN's College Basketball 00:10:43.850 --> 00:10:46.269 Nation blog. We are talking about SB Nation, 00:10:46.669 --> 00:10:49.789 Basketball Prospectus. The big online players. 00:10:49.990 --> 00:10:52.850 Right. But also local and regional papers like 00:10:52.850 --> 00:10:55.669 the Topeka Capital Journal. Mediate covered him. 00:10:55.809 --> 00:10:58.250 And the one that really signals a major shift. 00:10:58.250 --> 00:11:01.129 shift. The Wall Street Journal's daily fix column. 00:11:01.350 --> 00:11:04.230 That is the one. The jump from sports blogs to 00:11:04.230 --> 00:11:06.090 the Wall Street Journal is the critical moment 00:11:06.090 --> 00:11:07.730 in this story. It's a totally different audience. 00:11:08.029 --> 00:11:11.269 Exactly. When Carl Bialik at the WSJ writes columns 00:11:11.269 --> 00:11:13.350 detailing Ken Pomeroy's mathematical approach 00:11:13.350 --> 00:11:15.690 to a tournament bracket, it proves the respect 00:11:15.690 --> 00:11:18.289 for this data had transcended sports. It moved 00:11:18.289 --> 00:11:20.970 into the realm of respected statistical analysis. 00:11:21.389 --> 00:11:24.570 Right. The financial sector understands the value 00:11:24.570 --> 00:11:27.090 of a 2 % predictive edge better than anyone. 00:11:27.500 --> 00:11:30.379 So they immediately recognized the rigor of his 00:11:30.379 --> 00:11:32.960 methodology. And he didn't just remain an external 00:11:32.960 --> 00:11:34.899 source that journalists occasionally quoted. 00:11:35.000 --> 00:11:38.320 He gained true insider status. He became part 00:11:38.320 --> 00:11:40.460 of the machine. The source points out he became 00:11:40.460 --> 00:11:43.440 a contributing writer for ESPN's insider feature. 00:11:43.929 --> 00:11:46.429 His rating system became a standard citation 00:11:46.429 --> 00:11:49.090 in major legacy newspapers like the New York 00:11:49.090 --> 00:11:51.350 Daily News. It's an incredible trajectory. He 00:11:51.350 --> 00:11:54.029 went from a guy tweaking baseball formulas for 00:11:54.029 --> 00:11:57.370 college hoops in 2003 to the definitive authority 00:11:57.370 --> 00:12:00.450 that the entire sports media apparatus relied 00:12:00.450 --> 00:12:02.929 upon. We really need to contextualize why the 00:12:02.929 --> 00:12:05.049 media shifted so aggressively toward his model, 00:12:05.110 --> 00:12:07.889 too. Because for decades, sports media was built 00:12:07.889 --> 00:12:10.929 almost entirely on narrative. Oh, pure storyline. 00:12:11.570 --> 00:12:13.570 Broadcasters and journalists rely. it on the 00:12:13.570 --> 00:12:15.789 eye test. They'd talk about which team wanted 00:12:15.789 --> 00:12:18.250 it more, who had the hot hand, or which coach 00:12:18.250 --> 00:12:20.710 just knew how to win in March. Lots of cliches. 00:12:20.929 --> 00:12:23.389 But as advanced analytics became more accessible 00:12:23.389 --> 00:12:26.309 to the public, those narrative -driven journalists 00:12:26.309 --> 00:12:28.710 started getting exposed. Because readers could 00:12:28.710 --> 00:12:30.950 just pull up a spreadsheet and prove the columnists 00:12:30.950 --> 00:12:34.080 wrong. Exactly. When a legacy columnist claims 00:12:34.080 --> 00:12:36.740 a team is a powerhouse because of their winning 00:12:36.740 --> 00:12:40.019 culture, but a guy on a blog points out their 00:12:40.019 --> 00:12:43.000 adjusted offensive efficiency is ranked 150th 00:12:43.000 --> 00:12:45.279 in the country, and they are mathematically due 00:12:45.279 --> 00:12:47.799 for a regression. The audience notices who ends 00:12:47.799 --> 00:12:51.779 up being right. They do. So to maintain their 00:12:51.779 --> 00:12:54.320 credibility, the mainstream media had to adapt. 00:12:55.100 --> 00:12:58.259 When the New York Daily News or ESPN cites Ken 00:12:58.259 --> 00:13:00.840 Palm, they are signaling to their audience that 00:13:00.840 --> 00:13:03.740 they are prioritizing objective data over pure 00:13:03.740 --> 00:13:06.740 subjective opinion. It elevated the entire discourse 00:13:06.740 --> 00:13:09.059 of college basketball analysis. It proved that 00:13:09.059 --> 00:13:11.840 audiences actually crave deeper empirical insights 00:13:11.840 --> 00:13:14.559 rather than just relying on the same old barroom 00:13:14.559 --> 00:13:16.759 debate talking points. So what does this all 00:13:16.759 --> 00:13:19.340 mean? We started by looking at a completely free 00:13:19.340 --> 00:13:22.159 website launched back in 2003. Very humble beginnings. 00:13:22.519 --> 00:13:25.279 We saw how a single individual took a foundational 00:13:25.279 --> 00:13:28.120 mathematical concept. The Pythagorean expectation 00:13:28.120 --> 00:13:30.730 tweaked it. and applied it to college basketball 00:13:30.730 --> 00:13:34.169 to identify a team's true underlying efficiency 00:13:34.169 --> 00:13:37.190 rather than just looking at their basic win -loss 00:13:37.190 --> 00:13:39.720 record. Stripping away the luck. Right. We discussed 00:13:39.720 --> 00:13:42.399 how this rigorous, data -driven approach achieved 00:13:42.399 --> 00:13:46.340 a 73 % success rate, outperforming the NCAA's 00:13:46.340 --> 00:13:49.779 deeply entrenched RPI by a crucial 2%. That vital 00:13:49.779 --> 00:13:52.559 2%. And finally, we tracked how that undeniable 00:13:52.559 --> 00:13:55.919 accuracy forced a media revolution, turning his 00:13:55.919 --> 00:13:58.299 independent ratings into a mandatory analytical 00:13:58.299 --> 00:14:01.600 tool for everyone, from dedicated sports blogs 00:14:01.600 --> 00:14:04.100 to the Wall Street Journal. If we connect this 00:14:04.100 --> 00:14:06.600 to the bigger picture, the core lesson is really 00:14:06.600 --> 00:14:09.379 about the power of challenging. the default metric. 00:14:09.539 --> 00:14:12.700 Yeah. The RPI was the default for decades simply 00:14:12.700 --> 00:14:14.580 because it was the way things had always been 00:14:14.580 --> 00:14:17.379 done. But it was inherently flawed. Deeply flawed. 00:14:17.620 --> 00:14:20.179 Pomeroy proved that progress doesn't always require 00:14:20.179 --> 00:14:22.559 inventing something entirely from scratch. Yeah. 00:14:22.879 --> 00:14:25.120 Sometimes it just requires applying a better, 00:14:25.159 --> 00:14:27.379 more analytical perspective to the data we already 00:14:27.379 --> 00:14:29.860 have. Just looking at it differently. Exactly. 00:14:29.860 --> 00:14:31.759 It shows that no matter how entrenched the system 00:14:31.759 --> 00:14:34.419 is, objective accuracy will eventually win out 00:14:34.419 --> 00:14:36.580 over tradition. And that leaves us with a final 00:14:36.580 --> 00:14:39.320 thought for you to ponder on your own. We've 00:14:39.320 --> 00:14:42.379 seen what happens when one person applies a tweaked 00:14:42.379 --> 00:14:44.960 mathematical concept to a traditional index and 00:14:44.960 --> 00:14:47.799 completely reshapes mainstream analysis by finding 00:14:47.799 --> 00:14:51.240 just a 2 % edge. A small but mighty edge. Look 00:14:51.240 --> 00:14:53.639 at your own industry or your own daily workflows. 00:14:54.080 --> 00:14:56.559 What established metrics are you currently using 00:14:56.559 --> 00:14:59.080 that are just blindly accepted as the standard? 00:14:59.340 --> 00:15:01.919 What is your profession's version of the RPI? 00:15:02.000 --> 00:15:04.240 And what data -driven adjustment could you make 00:15:04.240 --> 00:15:06.340 to uncover the real truth hidden beneath the 00:15:06.340 --> 00:15:08.629 surface? Thank you for joining us as we... explored 00:15:08.629 --> 00:15:10.990 this shift from intuition to analytics. Keep 00:15:10.990 --> 00:15:12.929 questioning the numbers, look for the margins, 00:15:13.049 --> 00:15:15.210 and we will see you next time for another deep 00:15:15.210 --> 00:15:16.370 dive. Take care.