00:08 - Rob Pizzola (Co-host)
Welcome to episode number 47 of Circles Off. A special guest this week for this episode. He's a data scientist. He's the founder of thepowerrankcom, which I highly encourage you to check out, which I highly encourage you to check out. He's the author of how to Win your NCAA Tournament Pool, newer version available on Amazon and on Kindle as well. You can check that out, Dr Ed Fang. Thank you for joining us on Circles. 

00:34 - Ed Feng (Guest)
Off. Thanks so much for having me. It's a real honor and I'm really looking forward to this conversation. 

00:39 - Johnny Capo (Co-host)
Thanks, ed. Yeah, so for everyone listening, today we are going to delve deep into some college basketball stuff right before you know the infamous March Madness tournament NCAA March Madness starting next week. So with Ed today, we're going to go through a bunch of different topics about how to break down different things within this tournament, but also mainly what Ed specializes in here is strategy around filling out your bracket. We know a lot of you guys might have an office bracket pool, a different challenge. You might do something with your buddies. This episode is one you're not going to want to miss because hopefully it's going to make you some expected value in those bracket pools. 

01:15 - Rob Pizzola (Co-host)
All right, Ed, let's get started with your background, as we do with every one of our guests. I have long known that you're a Stanford PhD. I have long known that you're a Stanford PhD, but I was reading the back cover of how to Win your NCAA Tournament Pool and you were a Stanford PhD in chemical engineering, which we've crossed paths many times in the past. I never knew that before, so I'm very curious how you go from a career path towards chemical engineering into predicting the outcome of sporting events. 

01:46 - Ed Feng (Guest)
Yeah. So maybe let me start with how I got into chemical engineering Sounds good. I was a high school senior and I took a college visit to Case Western Reserve University and as a pretentious prick that I was maybe still am I went up to the chemical engineering prof and I said is this the hardest major? And he said yes, and after that I decided I wanted to major in chemical engineering. So I go off to college. I didn't go to case, I went to, I went to Rice and did chemical engineering and enjoyed it, liked it enough and knew that I wanted to go on to grad school. 

02:26
So the summer after my junior year I got an internship out at Stanford. I thought it'd be nice to go check out California, maybe somewhere. I wanted to go to grad school and I got a job in the lab of a guy who does polymer science and the first thing I did in that lab was pour a bunch of chemicals together to try to make a polymer. And it wasn't my cup of tea. I don't really do experiments. I don't like experiments. I guess I like cooking, but that's about the extent of how I like to experiment. So that summer I realized like, okay, I like this chemical engineering thing. I want to go into grad school but I don't do experiments, so let me try to get into the quantitative aspect of it, and that's what I ended up doing. 

03:12
I started studying the quantitative aspect of polymers when I was in grad school, ended up going out to Stanford for grad school, started studying essentially applied math with my advisor, and it turns out that the math behind statistical mechanics is almost a perfect background to get into sports analytics. A lot of the variation that you see in physical systems turns out to be the same in sports. When you look at fluctuations in games, there's some interesting relationships that remind you of things called like the fluctuation dissipation theorem and statistical mechanics. So it ended up being a great background. The other aspect of it that made it a great background was the math that I learned through my thesis. I had to learn about the math behind Markov chains, so these are random processes and um, and I learned about that in my thesis and that turned out to be really useful later because, uh, I went on to a couple of academic stops and, uh, my last stop, man, my last real stop, was at Berkeley, and it was I say it was my last real stop, in the sense that by the time I left there I knew I wasn't going to be an academic. 

04:26
I managed to piss off everybody that I needed to help me get my next job, and you know, I was kind of a young stubborn kid, didn't really want to publish or perish, wanted to write one brilliant paper, and that just doesn't work in the real world of academia, and so I was kind of looking for something else to do. I'd always been a sports fan and you know I was really hoping that I could do something in sports, and I stumbled upon what the original creators of Google were doing and their page rank algorithm, and essentially they were making sense of the internet based on the link structure of the web. So if you think about you know what defines an important website? Well, one definition could be that you're getting inbound links from other important websites, and so I thought that might be an entryway into sports, with the analogy being like websites are like teams, and you know, maybe you're a good team if you beat another good team. And how do you do that? And it turns out the mathematics of markov chains is, is, uh, is how you do that. 

05:31
Um, I went and took page rank, uh, fussed with it so that it would accept margin of victory as an input, and that's the power rank algorithm. It takes the original version, takes margin victory and adjusts for strength to schedule. And you know, margin victory is a point spread, right? I mean, that's essentially what you're looking at at the at the end of every game. Um, uh, it's it. You know, and you want to make a predicted point spread, and you know you, you know for any game you could just take the raw margin of victory of two teams and subtract them and that will give you a predicted point spread. I don't think you should bet with that, but what I do is I take those margin victories in games and I adjust for who you played and that's obviously very important in. 

06:16
I will argue, it's very important in nfl, but there's more parity in the nfl. So you may not accept that argument. But in college basketball it's critical. There's so many different teams, there's so many wide variety of just strengths of teams, strengths of conferences, and college basketball is a really fertile ground to look at that college football as well. I also do some international soccer rankings on my site. The code needs a big update so you don't have to wait four months for the stupid thing to update these days. But international soccer is another sport that you know. There's a wide variation between Brazil at the top and, you know, suriname at the bottom, and so the power rank algorithm really helps you with that, and I guess that's the long way that chemical engineering led into sports. 

07:06 - Rob Pizzola (Co-host)
All right, so let's get into the transition into betting now, which has obviously become a major thing across North, always has been really a major thing in North America, but now, with regulated markets, it's growing more people have eyes on it. What does your day to day look like from a betting perspective? Because obviously you manage the power rank, so some of your time is devoted to that. You're producing content, whether that's written, or your podcast as well, the football analytics show with the power rank and Ed Fang, and then betting on games yourself. What is your day-to-day look like? How is that divided? 

07:37 - Ed Feng (Guest)
Yeah, I mean during football it's kind of a week-to-week thing. I would say, you know, a lot of the early part of the week Sunday, monday is devoted to the pure data science aspect of it updating the numbers, getting that up for members, and then, you know, I would say, early in that week is really when I try to make my plays based on those numbers. So taking those numbers when those numbers suggest value, thinking about whether that makes sense or whether, you know, no model is perfect. Obviously, I definitely trust my NFL model more than any other one that that I run, and so I try to make sense whether that makes sense in terms of what I know about the teams, what Rob Bazzola said last week about teams, what a lot of other people that I respect say, and so I would say, like that mid part of the week is geared towards finding bets, and some of them I bet well, I bet all of them. Some of them end up being content for my site, and then, yeah, and then I but I would say, like you know, more of my time ends up being spent on the site. 

08:44
I had kind of had this vision in August of 2021, that I was going to spend more time betting over the course of football season and you know I'm a little sad to say that that really didn't happen in the way that I envisioned it and I'm. You know, every year is different and I think every year for all of us in the space kind of brings new challenges. It's one of the things that I really love about the betting space and doing my own thing with my own company. But yeah, I would like to figure out a way to do a little bit more substantial betting with it as the future comes. And I don't know exactly what that looks like, but that's kind of a rough estimate of how things go between betting content and the site. 

09:24 - Rob Pizzola (Co-host)
I like how you walked us through your NFL betting over the course of a week. It mirrors a lot of what I do as well. You're a numbers guy, I'm a numbers guy, but I add a layer, a subjective layer of analysis on top of what I do. I'm never betting against my models outputs, but sometimes I'll say you know what, this doesn't seem right to me. Maybe I'm going to toe the line and not bet on this team or so on and so forth. But it also comes with its inherent challenges. 

09:50
You mentioned that you potentially listen to me over the course of the week and consume other content. I'm just very curious. You know kind of how you weigh those things, because from a personal perspective, I think at all times there can come a point where a better lacks confidence. You're on a bad run, you start listening to all these things that are happening around you and maybe someone vehemently disagrees with stuff you have to say and you start to question your thing. So how much of what you do is just strictly trusting the model versus this subjective layer that's on top of it? 

10:22 - Ed Feng (Guest)
Yeah, I think that's an ongoing thing, rob. Let me give you an example from this year. So, so one of the characteristics of my model is there's almost no rushing stat statistics in there. Okay, and when you have an NFL model that has almost zero rushing statistics in there, I mean, and there's a good reason I don't have rushing statistics in there. You know, a lot of my analysis suggests that it's just it's not related to the number of points the team scores. 

10:51
So when you do that, like, you ended up betting against the Eagles and the Colts a lot this year and there was one particular example I think it was Eagles Chargers where I think I forget the exact numbers, but I had the chargers and I felt really good about that bet and, um, and and, and the market moved drastically in the other direction. I, I think it went from maybe charges, a two-point favorite at philly, to a pick or something like that. And you know, looking back, thinking back upon it now, I think that was the right side. Um, I, I think I like, I still like philly in that side. But there were a number of bets over the course of the year where I'm looking at my numbers and you know, I mean Rob, I think Rob and Johnny, I think you can question a model if you're six points off what the market says. But you know, at the end of the season my number was six points off on a lot of these Colts games and especially heading into that season finale at Jacksonville. And you know, I mean I don't, I don't think either one is necessarily right. I think there's aspects of my model that are right. I think there's aspects of the market that overreacted to what, what we saw with the Colts this year and in some sense, I guess I want to say I don't think I trusted my model enough over this past year. 

12:05
Yep, and I think my, as my process evolves, like this off seasons, I'm going to dig in and say like, okay, so let's look at teams that rush really well or stop the run no one really cares about stopping the run, but but has a guy like Jonathan Taylor and let's maybe look back at all the games in which you know my market, you know my model is off the market pretty significantly and see how it did. 

12:26
You know that's going to be a small sample, so I so that's not perfect, but you know, looking back upon what I did. I think I should have trusted the model a little bit more and not talk to myself out of a couple of bets over the course of that season. I think I definitely talked myself out of betting a couple of games against the Eagles and against the Colts and, you know, going back and see how those did and I think that's, you know, that's one way I'm looking to improve the process, but I don't think I think it's always evolving kind of that model versus your subjective analysis. I'm sure it's. I presume it's always evolving for you as well, rob. 

12:59 - Rob Pizzola (Co-host)
It is. I mean, it's pretty regular for me. I think I deal with a lot of the same things you deal with because of the way that we approach sports, which is from a numbers perspective. I've learned over the years that you know, if you have a 10% edge on a game and your model's you know spitting out that type of edge or larger, it's very likely that you're not capturing something that the market is. However, you only have to be directionally correct, right? I mean, even if your edge is overstated, if you still have an edge on that side, unless you're, you're, you're going to, you know, completely bankrupt yourself by following, like a full, strict Kelly criteria and type of um staking pattern. That you just have to be directionally correct and that's the. You know the, the goal more often than not. 

13:43
So I mean, I feel a lot of the exact same things that you do where over the course of the years, it doesn't matter what sport it is that I'm betting on, where you might be constantly betting against a team and the market is constantly moving the other way, and then you kind of have to take a step back and say, hmm, what am I missing here? 

14:01
And you can kind of go crazy when you start evaluating at that level, right To the point where you start to question all the work that you've put in and the back testing you've done and whatever. So it's not a simple game by any stretch of the imagination. I think so much of what comes with it, ed, is the mental aspect of just being willing to say you know what I've put this together, I'm fairly confident it works. You know what I've put this together, I'm fairly confident it works. I've tested it. I know that this doesn't matter, this does matter, and just kind of running with it. And I don't know why it's so hard to do that at points for certain originators, myself included, but I tend to personally struggle with that on like a weekly, monthly basis. 

14:42 - Ed Feng (Guest)
Right. And it brings me to something that Alan Boston told me on my podcast not too long ago. I asked him about closing line value and he didn't think it was that important, simply because he thinks he needs to do something different in order to get an edge. And I'm not going to come on the show and say closing line value is bogus, because I do not believe that. But in a sport like the NFL, where I believe I have dug into the numbers to extent that I would say few other people have, maybe I'm at a point where, like, not every game is going to go my way, but it could still be the right side and I haven't fully processed whether I think that's true. 

15:26
Um, I think the more quantitative analysis you can put behind an assumption like that, like a little bit like like I talked about earlier, looking back at these games involving the eagles and the colts, you know, the more you can say, okay, I don't care as much about closing line value for this team, um, but you know I mean. I mean, for example, like I, you know I'm doing college basketball now and if I don't get closing line value on a game, yeah it's not a good bet, right Right, like my models are very similar to what other people are doing. 

15:58
I'm not doing anything special. I mean I've told you what I'm doing, like I have margin of victory and I just restrict the schedule. I mean, I've told you what I'm doing, like I have margin of victory and I just restrict the schedule and I think my model is good and I try to understand these teams on top of that. But like I haven't really dug into a point where I can say like yeah, I'm ahead of everyone and I have something that that maybe no one else is doing with football. Maybe that's the case. I don't know Again and you can only. It's never a one and zero whether you're right or whether you're not. It's it's. You know how confident you can be in that assumption and and I think there's just requires work to figure out. 

16:35 - Johnny Capo (Co-host)
Ed, I have a question for you actually, because I I'm a slightly different, better in terms of style, of you and Rob here, who you know. You guys are building ground up models originating on the NFL, which I do not do at all. So I also just wanted to kind of dig a little bit deeper into, like, the process of what you do before you actually place a bet. So you're, you've explained kind of, how you actually come up with what you want to bet. But in that example you mentioned okay, I want chargers, I want the chargers today. What then? Into it? Obviously, uh, I mean, and I guess if you, if you want to share where you at, where you're at right now, where you're from, you're obviously in a state with a lot of, uh, a lot of options for sports betting yeah, I'm in the state of michigan. 

17:15 - Ed Feng (Guest)
Uh, we have a lot of online sports books that are available here and um. You know, if I like a game, uh, if my numbers like the game, if my subjective analysis on top of that likes it as well. You know, I look at other things. Obviously, there's a lot of people out there that I trust, like Rob, and then I try to look at where markets have been or how markets have moved on these games, on these teams, in the last couple of games. I think that's an important piece of information that allows you to anticipate what's going to happen in the future. And, you know, if most of these things are pointing in the right direction, then I go ahead and make a bet, and I I don't, you know, I don't make the biggest size bets, so I can get in there pretty early in the week when the, when the lines are softer, and I do that with football, and with football I rarely bet on the weekend, rarely. 

18:08 - Johnny Capo (Co-host)
Yeah, fair enough. I think a lot of people at this point who are originating are finding the edge the same kind of way right Early week or the openers from the overnights on Sunday and then kind of just following the market and then saying, all right, these are my positions, these numbers. I wish I would have bet all these games, these ones out, wish I didn't bet these ones and kind of go, go for it. Does that sound familiar, rob? It does. 

18:30 - Rob Pizzola (Co-host)
Uh, very familiar. I mean, listen, there's there's so many different things with the NFL, um, specifically. But I think, ed, you bring up a lot of, um, good conversation points in that I personally think it's very healthy to I don't want to say question everything, but to have some sort of internal dialogue with yourself of you know. The closing line value argument is a good one, right, I mean, we are conditioned to believe, or you know a lot of the content out there winning bettors in the space are. You know, you need closing line value in order to win. But there's exceptions to those rules as well. Right, we've talked about some of them on the podcast. 

19:07
Before I think over, I think it's very valuable metric over time. But, um, I think it's it's important to just be able to question things in the space and kind of do your own research and and whether or not, um, you think there's value to certain things. So a lot of that certainly hit home with me on a personal level. So a lot of that certainly hit home with me on a personal level. I'm just very curious because I don't model a number of sports, but I still like to bet on them. It's more of a recreational thing for me. Do you find yourself? 

19:36 - Ed Feng (Guest)
doing the same for stuff that you don't model. I try to avoid it. Okay, so I'll give you a good example. So this morning this week I really wanted to bet Illinois in the Big Ten Conference tournament and even though things are kind of busy with brackets and stuff there's a lot of prep that goes in my podcast this week I was like, look, I really want to bet this. 

20:02
So, rob, I didn't bet my hunch and and I was like, well, let me, let me get some numbers on this because I have the code. And then so tuesday, I worked really hard to get that code working and it was just a failure. It's kind of embarrassing for me to say, but like I couldn't do it because of the, like the advanced structure of the bracket, like there's like two sets of playing rounds, and my code was just like no, no, I'm not, I I refuse to accept this. So I started writing a different simulator to do it and then, like that wasn't quite right and that wasn't giving me the same answers as what the other one gets. So I couldn't really trust that. But then I found a workaround. So I was just like forget it, who cares? Who cares about those teams and those, those that first playing game like let's just assume a winner and go on. And so I did that this morning and I calculated a bunch of odds and you know it turned out that there wasn't much value on on Illinois. It was very close to the to the market value. And, spoiler alert, I, I, I don't think Illinois can win the NCAA tournament, but man, they, they can make life hell for a bunch of teams with the talent that they have. So I ended up putting a small bet on Illinois, simply because I think the numbers are underestimating their strength, and one of the reasons is one of their key players has been hurt for the majority of the season and this is a guy that I've been eyeing for the last couple of years, thinking he can be big 10 player of the year by the time he graduates. A guy that I've been eyeing for the last couple years, thinking he can be big 10 player of the year by the time he graduates. Uh, guard named Andre Corbello. So I did put a little bit of a bet there, um, but I tend to lean on the numbers, simply because I think that gives you a a basis to uh to make a bet and because I got that code working. 

21:42
I did the odds for all the other conference tournaments and you know there was a question of. You know I think this Duke team is kind of interesting and it turned out there was a little bit of value of Duke as a pretty big favorite in the ACC tournament. They were minus 125. My number suggested that there was value on that, simply because the rest of the conference is so fricking bad this year, and so I think I think that's a good example of where I kind of let the numbers tell me where there's value, because these margins are really thin Right. There wasn't a ton of that, you know. There was a couple of percentage points, I think I, you know I had to get 57 percent chance to win that tournament versus a break-even probability of 55. So it's not a huge edge, but I think there is one. 

22:31
I kind of had a hunch that there was one, and so I bet that there's also the factor that Duke really embarrassed themselves in that Coach K home finale. They played really bad and they're going to use this tournament as a, as the way to right the ship. There's going to be a ton of motivation in there, whereas maybe there's not 100 motivation every year. So, um, and then you can also use the same like logic for arizona. I thought arizona was going to be a pretty prohibitive favorite because I think they're a really good team. Um, they're about like minus 110, I think this morning. But the number said no. The number said, you know, their probability of winning the Pac-12 was about 50%, and that's just because UCLA is better than anyone in, you know, besides Duke and the ACC. So yeah, I mean I do let the numbers kind of tell me where those small edges are, because these edges should be pretty small at this point. 

23:29 - Rob Pizzola (Co-host)
In terms of the numbers. Anyone out there can find Ed's projections at thepowerrankcom, and I want to talk a little bit about the power rank, because it's a website that I've used for multiple years, ed, since meeting you. I particularly like a lot of the written content that you do. We referenced the how to predict interceptions in the NFL article when we had you on for our Super Bowl props podcast, and what I particularly like about your written content is it's unique. For one, it's typically topics that I don't see elsewhere, but there's also a wide range of them, like covering different sports, a ton of different topics. How do you come up with the topics that you're going to write about, ed? 

24:11 - Ed Feng (Guest)
I mean, more recently it's just been like I want to be a better, better right. So the interception things came about because one off season I was like all right, so there's a lot of randomness and turnovers in football, but like, can we do anything anything a little bit better? And yeah. So actually one thing I'm working on this offseason, which is based on a talk I gave last May, was like I ended up writing the thing about predicting interceptions and I think that in itself is an interesting data thing about, like don't look at a very small set of events that are just picks, right. Look at a bigger set of events that are indicative of, of a quarterback's behavior, um, and. And that bigger set ends up being passed as defended, so, um, that that allows for for better predictions. 

25:03
But the journey to predicting interceptions was not linear at all. I took a couple weeks, maybe a month-long foray into fumbles that, at the end of the day, wasn't particularly fruitful, and it was only a couple weeks later that I was like well, there's so many more picks than fumbles. Focus on the picks. 

25:23 - Johnny Capo (Co-host)
Have you tried it on any other sports, anything else? College basketball potentially. Many more picks than fumbles, like focus on the picks. Have you tried it on any other sports? Like anything else? College basketball, potentially. Or even like anything like the whole concept of hey, don't just look at interceptions, which is one, one stat which is like you know how many interceptions as a quarterback through in the in the entire year? It's not like it's not hundreds, right? So have you tried that same concept with any other sports? Key stats. 

25:51 - Ed Feng (Guest)
I have not tried that concept in other sports. I'm trying to think, I mean, maybe I have, but yeah, no, I don't think so. But uh, you actually reminded me, johnny, thinking about other sports, of just another way in which, um, some content came about this week. Uh, when is this? Uh? When is this podcast going to drop? Uh? 

26:11 - Rob Pizzola (Co-host)
probably thursday, tomorrow, yep okay. 

26:14 - Ed Feng (Guest)
So so I've been working with edward ugras on a lot of the content on the on the football analytics show and he suggested for some stuff. We're doing like like, oh, let's talk to these people that have done some work on predicting upsets in the NCAA tournament, and I was kind of like no, it's like small sample size, right, like I don't you know, like I just don't trust like that work, right, I don't think there's a lot of stuff. Because anytime you say, oh, I'm just going to look at tournament games and look at the characteristics of the games and figure out what upsets are, you're limiting your sample size. And he's like no, no, no, we should really do this. And I don't know. 

26:56
I mean, I guess it's a little pretentious of me, but I was always, I've always been, of the belief that it's really hard to predict upsets beyond what our models tell us, beyond what the markets have, basically because there's a lot of randomness in college basketball and that's also really fascinating in the sense, like I've just been, I'm reading the signal and the noise by Nate Silver, and I've always had this belief that like why the heck can't geophysicists like predict earthquakes, this belief that like why the heck can't geophysicists, like, predict earthquakes? Right, you know? Like not knowing anything about the problem, I'm like, oh, you guys, of course you guys should be able to predict earthquakes. You've been working on this for 200 years. Well, it turns out it's really hard because we don't understand the underlying physics about how rocks move right and and once you understand, they're like, oh, that makes sense. 

27:40
It's kind of like my belief that it's really hard to predict upsets in in college basketball. So, anyways, the reason I asked why this podcast is when this podcast was going to drop, because the episode Friday on the Football Analytics Show, so we're doing a bracket wisdom series, and Well, whatever, I'm just going to tell you what happened because because I enjoy this story so much so I was like all right, let's look at, let's look at how to predict upsets. Right, forget, forget. Looking at one set of tournament games. 

28:09
Like, let's look at all college basketball games in which an underdog is, is, is is a dog by six or more points. Right, that gives you a decent sample size of games. It gives you a sample size of about 400 games from the current season and what I found is like t okay, so you can kind of guess that maybe teams slow the pace down, right, right, you take advantage of the variance. Uh, you know, if, if you, if you have a lot of possessions, you're going to get closer, you're going to closer, you're going to, the favorite is more likely to win, correct? You want to get advantage of the variance? Favorite is more likely to win, correct? You want to get advantage of the variance of fewer possessions. 

28:46 - Johnny Capo (Co-host)
Yeah, same way when you look at in the nfl for anyone listening, if there's a game with a spread of like 55, it's going to be a lot, you know, more variable than a spread of like 39 total total. 

28:56 - Rob Pizzola (Co-host)
But correct total sorry, total yeah, exactly. 

29:00 - Ed Feng (Guest)
So it turns out there's no evidence of that. Like, the pace is actually higher, slightly higher, in games in which an underdog of six or more point ends up winning. So the other thing you can ask is like, well, those underdog teams like jack up threes, right, right again, trying to increase the variance of what's going on. And so you dig into these games in which an underdog of six or more points wins and they actually shoot fewer threes, um, and the favorites shoot more threes. And I think that's because of the game state. I think because to beat a favorite you have to be up late and then you're not shooting threes, right, right, yep, um, I don't know if I don't know if that's completely true, um, but they're not shooting more threes. 

29:44
So I don't want to say that there's not individual coaches out there that are smart enough to say I'm a six point dog, I'm going to slow it down, we're going to shoot more threes. I'm sure they exist, but overall in college basketball that's, that's far from true. And it turns out that the biggest effect is three point shooting and it turns out that the biggest effect is three-point shooting. So dogs of six or more points that win shoot five and a half percent better from three than their season-long average. And three-point shooting is the most difficult, if not impossible, thing to predict. Actually, johnny, you know this right. I mean you guys are betting these three-point props. It's tough. I mean, randomness is your friend. 

30:21 - Johnny Capo (Co-host)
A lot of randomness. There you're, you're 100, right. So then what would you? What would you go from there then? Would it be looking at what teams have the highest range of outcomes within the three-point percentage for the for the the year, I guess, or like? How would you then go about evaluating that to pick which dogs to win? 

30:37 - Ed Feng (Guest)
no, I tell you that it's impossible to pick which dogs to win beyond the point spread just pure luck there's so much luck involved in underdog winnings, which I'm getting at yeah, so okay. 

30:47 - Johnny Capo (Co-host)
But backing up, let's say we, we did and I know it's not firm conclusion, but let's say we did determine that, hey, one of the one of the things that's typically constant when an underdog beats a favorite of, let's say, 10 points, 10 point spread, then we'd say, like this team typically would need to shoot above this percentage from three and would potentially have to have, you know, a good free throw shooting game as well to make sure that the other team's not clawing back in from the missing the front end of the one and one. 

31:15
Let's say, for example, we outlined those two and then you had a way to evaluate which teams were gonna have high variance within the three-point shooting game, whether it be they had a team of high variable shooters or whether they just haven't shot a lot of threes in the season. And maybe, if you haven't shot a lot of threes in the season, come out with that, that mentality like you might be able to get it. If you had a way to then predict the three-point shooting, would then that be an edge that you think would exist in college basketball? 

31:44 - Ed Feng (Guest)
I think, yes, but I, that's the big if right. And like personally, johnny, I'm, I don't feel confident enough that I would find anything interesting enough, like just just the thing that you said how do you predict a team that can maybe, in a game, shoot two or three percentage points better from three point range? A lot of what I've seen in college basketball suggests that that's just simply impossible, and part of that evidence is, you know, on a team level, you can look at what a team has shot from three point range and you see massive regression from early to late season to um. I mean massive regression, like the r squared is like like 1%, like there's literally zero relationship. So you know, johnny, if you can, if you can tell me how to, how to predict the games in which, uh, you know a team's going to shoot better, I'm, I'm all ears because that would not only be good for your props and um, but that would be the key. 

32:37 - Rob Pizzola (Co-host)
I have a theory. I don't know that the NCAA basketball data is good enough to do this, but I would feel that a team using more small ball lineups in recent games would be something that might be a predictor, because obviously the smaller the players on the court, I think the more likely that they are to be better shooters from long range. That's just like a general hypothesis I'm pretty sure is true. So I think maybe something around the player rotations where teams that are shifting towards more small ball lineups. Now, is the NCAA data good enough to? I mean, I don't know, I don't model college basketball anymore, but I mean that's where I would start as just a pure hypothesis of something to look at. It was the first thing that came into my head. Is less big guys on the court more likely you are to be able to shoot threes? 

33:29 - Ed Feng (Guest)
Yeah, absolutely. I like that hypothesis. I think you could do a little bit, because we do have play by play with with. I mean, I don't have a play by play, but someone out there has play by play with college basketball to look at it. But it also gets me thinking. I mean you can definitely do this for the nba, right? Um, whether teams change styles. And I would, now that you've mentioned it, rob, like I mean that's, that's probably the next study that should be done. Right, like, what do? What do teams of dogs of six points in the nba? How do they pull off the win? And our coaches slowing it down, shooting more threes, putting in lineups with guards? I don't know if that's true, but it should be true, right? I mean, the NBA is as analytics-driven as any league out there. 

34:13 - Rob Pizzola (Co-host)
I think also there's a possibility to gain an edge on live markets by doing stuff like that as well, which is probably not taken into account right now, because obviously, even with the deck prism stuff now it's pretty sophisticated models, stuff like that as well, which is probably not taken into account right now because, um, obviously, like, even with the deck prism stuff now it's pretty sophisticated models. But I'm sure those guys would tell you um, david, owen and miller um that you can't just account for everything either and there's ways to kind of think outside the box to try to gain an edge in those which we kind of you know, me and johnny have our own live betting edges in different sports. 

34:42
But yeah, I think, there's tons of applications to this stuff, which is why it's fun, I think. Um, you know, you, you I write down while I'm watching football in the regular season. I have like a notepad where I just tend to write down things that stand out to me and I do a lot of research on them in the off season and it turns out that more often than not, I've potentially found something that is a value that I wasn't accounting for earlier, and I think the more analysis that's done in the space like this, I think, the more edges that can be found now. Granted, they probably are temporary, Everyone will pick up on them over time and then you have to separate in some other way, but I think it's pretty cool stuff. 

35:19 - Johnny Capo (Co-host)
So one other thing I would. I would mention it oh, sorry to cut you off, ed. So if we're looking at the three point stuff, what I was also going to say is like you don't necessarily need to figure out who's going to shoot better from three. Like you don't have to predict who's going to do better, you just have to predict who has the wider range of variance and then you let the variance play out yep on its own. So, for example, even just looking and saying and I it would probably be a small sample size, but looking and saying like these teams typically hover around, they never surpass x amount of standard deviations across around their mean. So their whole season is very, very consistent. Might just be the type of shooters they have. And then you might look at another team which shoots, has shot 10 from three and then has also shot like 48% from three. 

36:02
And if you look at those range of outcomes and just can determine that this team right here is super variable and you can bang them on the money line, that's something that's like potentially a good bet. Obviously, this stuff's probably priced into the market, but we're looking for smaller edges here, smaller conference games, stuff like that. When you look at that, that's one way to do it. The end of the day, if you're betting a team, it's like plus 1600, plus 1700. It doesn't matter if they lose by 67 points or if they lose by two points. You're losing your bet regardless. If you're betting on the money line, so you are just looking for teams with super high variance where they can win that game. And if they can win it obviously within that percentage of the plus 1600, then you're going to get a plus EV bet. 

36:44 - Ed Feng (Guest)
Yeah, no, I like that idea a lot, Rob. One thing I wanted to mention with what you said about you know, can you look at lineups and are they going to shoot a higher fraction of threes? The only caution that I would say is that I've looked back in college basketball and I've looked at the relationship of your three point field goal percentage like attempt rate, so the fraction of field goal attempts that come from three and look to see if, if you have more variability and offensive variance in your offensive output and I found none Interesting there was. There was little relationship there. I believe that there's the same result in the NBA as well. So there's the additional question of if you are going to put that small lineup, are you actually going to get more variance in your results? 

37:33 - Rob Pizzola (Co-host)
Right, it's a really fascinating question. I mean, there's so many ways you can go about it. Like, even as you bring it up, I'm thinking more and more about these types of things and depending on again, I don't know what data is available for the nba uh, for for ncaa basketball versus nba, but I think even isolating uh transition threes versus like threes that are in a half court set might tell you something about um randomness for certain teams, like I don't, I don't know if, um, if there's a, there's something Like it gets the juices flowing for me. I'll tell you that We've said this before, ed. 

38:04 - Johnny Capo (Co-host)
If you're going through and scraping through data and actually like removing variable points and stuff like that, you're going to get a better data set than the rest of the market. Four examples A prime example for you, ed, and for anyone listening, is at the end of a quarter or at the end of a half, on a low shot clock situation, a team chucks up a three from half or from their own thing. That still counts as a three pointer. Right, in theory, you should just remove all those from the data set because they're not actually indicative of the game. You can assign a certain percentage later on if you want to factor into some sort of model, but reality is it's irrelevant. So, because of the fact that that's not a real three that should count into oh, I'm three for 13 and a half. 

38:43
So by going through and just like removing all of these types of things, this is just one example for three pointers. But and then what rob said is like transitioning it out, like how many transition threes? How many were like fast break off, a turnover? How many were, which are, you know, typically less common? How many were actually like drawn up out of a timeout, out of a? You know what I mean, you can actually go through and build a better data set and then have a better angle. So if you're doing any of that, ed, that's money stuff, the real money stuff. 

39:12 - Ed Feng (Guest)
No, that sounds good. I mean, anyone listening to this can go to big data ball and get NBA play by play and start doing this stuff and and following along with what you said, johnny, the Holy Grail is points per possession in the NBA and you can clearly get that from the play-by-play, but I would argue you got to strip out all those foul possessions at the end of the game. That's not normal basketball. This is something that I've thought about doing for a while. I've never gotten around to it, so, um but, but it you're. You should have a higher, higher expected points per possession if you're getting fouled intentionally, right, or you can apply some sort of score effects to or or time effects. 

39:56 - Rob Pizzola (Co-host)
I mean that's more complicated, but yeah, there's ways to go about that. Um, it's a very interesting conversation, ed. We're going to switch into filling out that brackets shortly, an educational type piece. But I noticed you had ken palmeroy on your podcast um this past week. For those who don't know who ken palmeroy is, you probably don't follow follow college basketball at all, but he's basically the college basketball um analytics godfather. I would call him, I mean, um, kenpalmcom. Everybody uses the site in some capacity or another, even if you don't value the projections. There's just so much data available on the site. I'm just curious in your conversation with him, ed, was there any valuable learning that you took away from that convo with him? And just you know anything that you were able to pick his brain about? 

40:43 - Ed Feng (Guest)
Yeah, so many things. I mean, I in particular, kind of enjoyed his breakdown of teams. You know, like I'm at a point where, like, I'm looking for like little things that are not available in box scores, like just insights about players, teams, whatever, and he definitely provided a lot of those. One thing he did say was, straight up, like injuries are overvalued, and we were talking about Baylor in particular and they had an injury, they've had a couple, they have a couple, I would argue, two critical injuries, but it doesn't look like they're playing any any worse. So could be variants, could not be, but he did make the point that. But he did make the point that we tend to put an emphasis on injuries simply because it's a story to tell and our brains like to tell stories, our brains like to explain things, because that's what our brains do, they make stuff up. And that gets me back to the whole thing about how we we tell stories, about randomness, right, which is related to you know how you second guess yourself when you're on a losing streak, even though your model has been winning at whatever rate that you trust, for however many years. 

41:47
I think that same thing comes into play with injuries and it was really similar to what Alan Boston said the week before. He's like some injuries don't matter, the team college basketball teams have programs and they can plug guys in and and do just as well. So I thought that was particularly interesting. Um, but yeah, I mean, you know, I mean you probably check it out for yourself because I mean Ken is definitely worth listening to. He's done a lot of interesting things. Uh, he's got curling rankings, does he really? 

42:17 - Johnny Capo (Co-host)
he's got curling rankings. 

42:19 - Rob Pizzola (Co-host)
We might have to take a look at that we were, we were, we were betting some Olympic curling over here. I didn't find Ken's stuff. Maybe it's not published or public, I don't know if it is or not. 

42:33 - Ed Feng (Guest)
If you go to KenPalmcom there's a little. I don't even know what's the puck thing. 

42:40 - Rob Pizzola (Co-host)
The rock that they throw. It's called a rock. 

42:42 - Ed Feng (Guest)
There's a little rock in the banner, so just click on the rock, okay. 

42:47 - Johnny Capo (Co-host)
Fair enough. Yeah, we got it Like. There's gotta be some detailed analysis. Who's got the hammer for those? For anyone who watches, who doesn't watch, curling that went right over their head. That's a thing, though. Yeah, head, that's a thing though? 

42:59 - Rob Pizzola (Co-host)
yeah, that's a thing, last rock, right, the hammer's last. The last rock is very important very important. 

43:03 - Johnny Capo (Co-host)
Yeah, you should. If you don't, I mean what's it? Is it called a steal? Do you steal? 

43:08 - Rob Pizzola (Co-host)
if the it's a steal or it's a blank end, if if there's no point scored if there's no point scored, you keep the hammer you keep the hammer, so there's no point scored. 

43:15 - Johnny Capo (Co-host)
You keep the hammer if the other team gets a point when you had the hammer big, no, no, that's basically a shorthanded goal. 

43:21 - Rob Pizzola (Co-host)
Yes, it's a steal, it's. It's tough to it's. I mean, people do overcome it, but it's very difficult. You don't want to give up points when you have the hammer. It's usually a very good indicator that you had a very bad we're off the rails already. 

43:33 - Johnny Capo (Co-host)
So, uh, ed, if you don't mind, just uh, where can people find your your podcast? You mentioned, like the ken palm episode, we had alan boston. Just I could give that a little plug right now so everyone could write it down for later. After you're done this episode, you can go watch that one. 

43:45 - Ed Feng (Guest)
Yeah, after Absolutely. It's the Football Analytics Show. You can find it wherever you get your podcasts. Right now it's the Bracket Wisdom Series, which is a daily series of about 10 minutes every day that gets you ready to win your March Madness pool. 

44:01 - Johnny Capo (Co-host)
Okay, we're getting into that right now. 

44:02 - Rob Pizzola (Co-host)
We're going to get into it. So I, for all the listeners and viewers out there, uh, we're getting into an educational piece. Obviously March Madness around the corner, so we want to talk about filling out brackets. Now, some of this stuff that we're going to go over. If you're an experienced better or someone who you know knows a lot of game theory, some of this stuff may not be for you, because we want to appeal to the casual better Someone who might be filling out a bracket for the first time is in an office pool or whatever. But also we're going to go through some more advanced concepts as well. 

44:29
Throughout this and Ed is a perfect guy, Like I said, I have read this book, which is actually a very quick, short read but is filled with information. I would highly suggest it for anyone who wants a copy of it that they can refer to over the course of the years. But let's just start with um. A lot of people get discouraged, so to speak, about filling out a bracket because they don't follow the sport, uh, or they don't have some sort of sophisticated model, um number of variety of reasons. Is it still possible for them to gain an edge over other people in the bracket in their bracket, even though they know nothing about the sport absolutely. 

45:06 - Ed Feng (Guest)
I think if you just use analytics, like if you just go to my site or or whatever, I mean, let's not plug my stuff, right this is your time, ed. 

45:15 - Rob Pizzola (Co-host)
It's your time to shine. You can plug whatever you want yeah. 

45:18 - Ed Feng (Guest)
So I mean, if you just go um just any reputable source of analytics, so whatever, let's just use mine because I actually have some solid numbers on it. You know, like, your odds of beating someone and you just put the higher ranked team in um in every game, you know your odds of beating any one person from um kind of espn's I estimate to be like 75, 80%. 

45:45
So, you there. There's an edge there, right, Like I mean, we're always as better as looking for, you know, kind of square money, and there's no better example of that than than March madness tournaments. Now, if you get into a March Madness pool with, you know, Alan Boston and Mike Craig and and other guys who bet on college basketball for a living, then that's not such a great idea. Even if you're in a small pool and that's one of the things we'll talk about you got to know who's in your pool, right Like you want to get into the dumbest, squarest pool that you possibly can of like 10 people. That's kind of ideal, you know, like those family members that are just like, oh hey, this is fun. And even if you don't know anything like, just go straight with the numbers. Your bracket's going to look really boring. It's going to be a lot of higher seeds in there, but you're going to give yourself a pretty good chance to win. So I do think there's a lot. 

46:44 - Rob Pizzola (Co-host)
it's, it's a plus ev proposition to to get into these pools now when you actually have the bracket itself. A lot of people still print it out and fill it out. Some do it digitally now as um, you know who's still printing these out. 

46:53 - Johnny Capo (Co-host)
I personally still you write them in pen all your teams. 

46:57 - Rob Pizzola (Co-host)
I like to put the bracket up on the wall and after every game I go with a yellow highlighter or a pink highlighter right in like a race of dry erase marker. 

47:06 - Johnny Capo (Co-host)
No. Once I fill it out, it's stuck it's permanent here's what I'm saying. How could you possibly fill it out, though, with a pen? Then when I'm filling in a bracket, the amount of erases, that or the amount of swaps I end up doing, is like probably like upwards of 20 to 30, because I get, I finish it. I'm like no, that's going to be too much chalk. And then you go back, adjust some more dogs. You know, like how do you? 

47:28 - Rob Pizzola (Co-host)
let's just say I waste a lot of paper. Fair enough, let's just say I waste a lot of paper, but uh, ultimately, I think different people have different strategies. I'm curious on yours. Um, I've heard all sorts of things. Some people start with a champion and then they kind of work their way backwards from there. Other people start with the initial rounds and they kind of work their way towards a champion. Do you have one recommendation over another and, if not, what's your own personal preference for doing the bracket? 

47:54 - Ed Feng (Guest)
I think you should always work from champion, just because that choice is the most important in terms of points. It's usually worth 32 points, whereas a round of 64 is only worth one point. So I think, work from champion and that's what I do on my site. Um, that's what I do for for members and people who buy my bracket advice. Like I focus on the eight teams that have a legit chance of winning this tournament and I try to break it down and I try to give them basically every you know like every little thing, that every little edge that you can have and kind of evaluate these teams. 

48:29
You know, for example, like I'm not high on Purdue I think they shot the lights out earlier in the season. They have, they have a great player in Jaden Ivey, but I'm not high on the rest of the team. They don't play any defense and they are first when I look at my adjusted points per possession. But you, there's, there's no planet, there's no way you can convince me they're a better offense than gonzaga. They're just not. Are you looking at the betting odds? 

48:57 - Johnny Capo (Co-host)
are you looking at the betting odds at all when you're look, when you're doing this stage of the brackets, like the, the top eight teams, like, are you just like taking a look at who are the favorites, or is this more like ground up your own analysis? 

49:09 - Ed Feng (Guest)
I only think these teams could win so I certainly consider the betting markets in that. All that's baked into my model um that I use, use for all of this, so that that's all in there. I don I don't necessarily just take NCAA tournament futures and put it in there. I think I mean this is going to sound a little pretentious, but I think the models can actually do better there. Like I don't understand why the book should have such a huge vig on those things. Like, if you have a good college basketball model, you should. You should be able to nail these when probabilities um after selection Sunday pretty easily. But, um, yeah, I mean honestly, johnny, but market data in all sports is kind of one of my tricks that I try not to talk about too much, but it's, it's all in there. Um, using past market data is a very powerful way of projecting forward. 

50:03 - Rob Pizzola (Co-host)
When, when picking a champion, how, how much goes, how much did the other people in the pool play a factor for you? It's like, like I'll give an example. But say you're on the UCLA campus and you're filling out a bracket very unlikely that other people are going to pick Arizona to win. Or you know you can get a lot of UCLA's to win the tournament, like, how much of that factors into, like potentially? Let's say you're and I'm just I'm just throwing out a hypothetical here but you're down on Arizona. You don't really think that they're deserving of a number one seed or whatever just throwing something out there. But you know the vast majority of your pool is going to pick Arizona to to exit early. Would you then consider straight, you know, changing your strategy, going against your own personal beliefs, to throw Arizona in as a national champion, because you would be going contrarian in that in that state. 

50:55 - Ed Feng (Guest)
Yeah, I absolutely would. I mean, maybe you throw up, but sometimes you do that when you're betting on things, right, right. 

51:00 - Johnny Capo (Co-host)
Okay, so you mentioned first. 

51:03 - Ed Feng (Guest)
Yeah, like knowing who's in your pool is a really critical thing. And Rob, that's kind of the perfect example, right, if you are local in Los Angeles and everyone's high on UCLA because they made that magical run last year, um, then yes, you should fade them. Um, you should fade them and pick someone else. Arizona is a good example. Gonzaga is an interesting example. They're going to be the favorite by almost any metric that you respectable metric that you can find. But you know, for that, like it, I would suggest, like you know, arizona, gonzaga would be my top two favorites. 

51:40
But if you're in LA, you know you're a little bit more likely to have someone pick a Gonzaga or an Arizona just because you're on the West coast. So maybe pick the best team you know East of the Mississippi river, uh, which this year might be a Kentucky, uh might be a Baylor. So, um, yeah, but that's absolutely the right thinking. Right, you want to fade what other people are doing? I live here in Ann Arbor. Last year a lot of people took Michigan to win because Michigan was pretty good but they weren't the best team in the country, so there was a lot of value in fading them. 

52:14 - Johnny Capo (Co-host)
Okay, so you start. You got to know who's in your pool. So awesome. It's very similar to like reading a sportsbook's rules or reading your pool's rules before you make a decision, right? So you got to know who's in fade if there's going to be a specific pick. If you're in an nhl playoff pool and you're in montreal, don't take the haps, don't take the canadians. We got it. So that's step number one, most important. Okay, from there you mentioned you like to narrow down the top eight, and then these are the teams that potentially can win the tournament. I'm going to pick one of these as my winners From there. What's next in the process? 

52:46 - Ed Feng (Guest)
Yeah, I mean picking from there. It's pretty easy, right, because you just pick the better team in all the other games. So I actually provide that service both for members of my site, which will be my best numbers and then if you sign up for my free email newsletter, I'll give you that cheat sheet for, based on my public numbers, which are margin of victory adjusted for strength of schedule. I try to make that pretty easy because you know I mean in particular you don't want to spend too much time trying to figure out the 12 over the five. You know it used to be a thing where you know you just write all these articles about, like you know which 12 is going to be at a five, because there was a little small sample size of years where there were a lot of upsets there. Now, if you kind of look at the, if you look at a large enough data set, like 12 seeds are slightly less likely to beat, to win that first game than 11 seeds, or just slightly less, and they're slightly more likely that you know than the 13 seed. Right, it's very linear. There's nothing special about that 12 when you look at a big enough data set. So don't you know? And again, you know the stuff I was telling you about predicting upsets as well. Like that stuff's hard, just just pick the chalk, you're gonna be fine and I it's really similar to you know how. 

54:05
Like you know, rufus always tweets on the Super Bowl Like I'm hoping for the most boring Super Bowl ever. Yeah, like I want, I want to be asleep at halftime, kind of idea. It's kind of the same thing with pools. Like you want a boring tournament. It's not like you're not going to win if, if things go crazy in the first couple of rounds because things do go crazy, right, but the more boring it is, the better it is for kind of your optimal strategy. Both, like kind of the favorites, where you're, where you're in a small pool and you want to pick up, you know the higher rank team, you know a lot of chalk versus, uh, you know the contrarian thing where you are trying to fade other people. In both cases, you want the optimal strategy You're rooting for kind of a boring tournament and um, I think there's a lot of parallels there. 

54:52 - Johnny Capo (Co-host)
So if you're in a bigger pool, it was like, let's say, like 20,000 people even go more than that, a hundred thousand people. Massive pool hosted on some site. Um, how does this change everything, or does? 

55:04 - Ed Feng (Guest)
it. Yeah, I mean, I personally wouldn't join that pool because you know if you're in a a what a hundred thousand person pool I you know my methods can probably get you to win about once every 5,000 years. 

55:17 - Johnny Capo (Co-host)
Yeah, so so let's say it's a free let's say it's a free contest, it's the Warren let's not even say it's a Warren Buffett like perfect bracket thing. Let's just say it's a million dollar. Let's just say a winner wins a trip to the Superbowl, free entry. 

55:28 - Rob Pizzola (Co-host)
some brand sponsors it and whatever we're basically forcing you to join the pool, ed, and we want to know what your strategy would be in that larger pool. 

55:36 - Johnny Capo (Co-host)
Exactly. 

55:37 - Ed Feng (Guest)
Yeah, yeah, I mean, you're certainly not picking all favorites there, right? I mean you're, you're going to go with, you're going to go with upsets, you're going to talk to people who know 358 college basketball teams and and which ones are more likely to win than not. Yeah, I did some stuff on the odds of a of a perfect bracket. It ain't happening. So small. It's so small that I think I I think, if I remember right, the cost to insure it was like two thousand dollars. 

56:07 - Johnny Capo (Co-host)
Yeah, for a billion. Yeah, that probably sounds about right because if I follow it every year, there's like an espn tracker that tracks how many of the espn brackets across all pools are perfect through just the first two days, and it's usually like between like zero and ten. So that would be between the first, just the first two days, right. 

56:27 - Rob Pizzola (Co-host)
Then you just get into saturday, sunday, and then it's just done I believe I could be speaking out of my ass here, but I'm pretty sure that you are more likely to get struck by lightning seven times in your life than you are. To fill out a perfect bracket to give people some perspective what if you're trying to get struck by light? 

56:44
well, then it's a different story. I'm talking about, like not going out with an umbrella and, you know, a steel rod in the air or anything like that. Just like the statistic, statistically speaking, over time you're more likely? 

56:55 - Johnny Capo (Co-host)
what's more likely? You shoot around a golf in the 60s. Stop now, or I fill out a proof bracket. Okay, ed, um, okay, awesome. So basically, to sum that up real quick, what we're saying is like if you are in a bigger pool and I think actually I'll ask you this one, like what constitutes a pool size where you would want to stay with like kind of like the chalk strategy, um, versus actually trying to expand it, like how many is too many? If your pool's 10 people, no doubt you want to make sure you're like chalk pick one of the top eight teams to win the pool. Don't go crazy with like a seven seed winning it, stuff like that. But if you're in a pool, let's say like 100 people or, uh, 150, it's just like an office pool at some some job you work. Does that change, like in terms of the 10-person pool, 150, and then obviously 100,000? Where would you put that 100 to 150 sweet spot? 

57:49 - Ed Feng (Guest)
Yeah, where the favorite strategy has a lower win probability than kind of a contrarian fading strategy. 

57:59
It's usually somewhere between 20 and 50, maybe, man, I guess I've seen some examples where it gets to 100, um, but yeah, somewhere somewhere between there, I mean. So so, basically, you know, all these are estimates based on data from espn and it kind of matters, um, the fraction of people that you're assuming pick the favorite. You know. So you know, if you think back back to I've been using 2019 as an example when Duke was a big favorite, you know my numbers had them about 34% chance to win the NCAA tournament, which is very, very high for the favorite. But let's see, so it was like you know, let's say, like 40% of ESPN brackets picked Duke. The like, the higher. That is like, let's assume like 50 person, let's assume like 50% chose Duke. Just for example, the bigger that gap between the 50% of people and then their true odds of winning at 34%, the smaller that that crossover point becomes right, so, right. So, basically, like, the worse that other people are picking um, the smaller the pool, that that you can potentially start using the contrarian strategy makes sense. 

59:14 - Johnny Capo (Co-host)
Fair enough, okay, um, last question on this, I guess, and then we can start to get into our closing question here. But, um, we wanted to get like you to give out a piece of advice you haven't given out before. So definitely anyone listening, check out ed shows. Um, you know he mentioned before. We'll plug him again at the end. But what's one piece of advice specifically for college basketball, march madness, specifically for march madness? That's like it's not out there. Give us a piece of advice that, or or give us something that's like super overrated and you're like never follow this. Both will be just as valuable, right? 

59:49 - Ed Feng (Guest)
So I mean I mean I already talked about something like don't focus on 12 over fives, like that's like literally the worst waste of time that you can possibly get, and Johnny actually already gave you one right. Like the whole, like the lack of predictability of upsets is something that's going to appear on my podcast the day after this. You know, it's like it's like predicting earthquakes, man, it's, it's hard. So I just, I just wouldn't spend too much time on that. I would spend time with looking at what the market says, looking at what my model says, looking at what any other trusted authority says. It's hard to predict upsets and I wouldn't try. I mean, you can try if you want to have some fun with it. And then I would just, I guess my last piece of advice I would just really be cautious about the stuff you read out there. So, and and you know it doesn't take it doesn't take too much uh, intelligence to figure this out. 

01:00:46
But I was looking for other articles about um, how to predict upsets in the tournament, and there was an entire article on the ncaacom that took the nine times that a 15 had beat a two as their sample set for what constitutes an upset, and obviously nine games is is no, no kind of sample size at all For for upsets. So, yeah, there's a lot of trash out there. I I kind of want to do a podcast episode, just like worst advice on tournament and filling out your poll. I think that would be a lot of fun. It probably wouldn't be that useful for people, but there's a lot of garbage out there. I mean, that's kind of the reason I wrote the book. 

01:01:33
I was looking out there for a lot of advice and I first kind of got wind of the contrarian advice because there was a good article out there on yahoo, maybe like 2008 or something about how you fade other people. But then I started looking more for, like you know, this is what maybe 2012, 2013, before I wrote the book. There's just a lot of garbage out there. So, you know, stick, stick with people that you trust. Like you know, the people at 538 know what they're talking about. Um, I know what I'm talking about. So I mean, there you go. 

01:02:06 - Rob Pizzola (Co-host)
I'd love, I'd love to join you for that podcast of things not to do. That's like kind of my specialty open articles and stuff yeah Like we. 

01:02:13
We kind of have it with tweets that trigger us on this as well, where we kind of turn into educational. But yeah, there are certain things I think, especially um, the ones that trigger me the most are around march madness are the articles of like, always pick a 12 seed to advance in the first round. It's like, okay, now what if all the 12 seeds are 20 point underdogs in their games? Are you going to pick one just to have one advance? Because it's happened in these previous years? Right, and then like the oh, make sure that you have I'm just throwing out random stuff here make sure you have at least 113 seed in the. You know the sweet 16. It's like, why? Like, if they're all 15 point underdogs in the first round, why am I going to pick them to advance? 

01:02:50
I think a lot of people just fall into like the dumb. This has happened in the past. Therefore, it's a indicator of the future. But I mean, we all, like, we know, we know the way that the committee ranks these teams as well, right, lots of times you get some really bogus, crazy stuff out there where it would not be uncommon to see, you know, an 11 seed favored over a six um you know how about? 

01:03:14 - Johnny Capo (Co-host)
how about this one? This isn't even pre-bracket, but after the first round you always see these like news outlets that potentially claim to be be gambling news outlets, or just regular news outlets saying what an upset 10 seed pulls off the huge upset. It's like, bro, there's minus 175 favor. Just looking at the seed numbers that would be a fun one. 

01:03:38 - Ed Feng (Guest)
This is obvious advice for people listening to this show. But if you're filling out a bracket like just just look at who's favored in the markets, right for those first round games. Like look at, I mean you can get sweet 16 odds on on anything. 

01:03:50 - Rob Pizzola (Co-host)
Right that and and and I would even apply that to future rounds. Personally, a lot of people right, um, like you know, if you are going to pick a, an underdog, to make a run in the tournament, as an example, and you want to be contrarian, differentiate yourself. Have a 13 14 seed going far, I'd probably look at the ones that are the smallest underdogs in their games. If I'm going to have them advance multiple rounds, right, I'm not going to take the 20 point underdog in round one to advance multiple rounds. So I I think that plays into it as well. 

01:04:20
And the last thing we haven't touched on yet, which I just think is important to touch on for anyone that's listening, is you have to know the rules of your pool, right? There are different scoring systems, sometimes for March Madness, some that incentivize you to pick an underdog. So if an underdog or a lower rated seed wins their game, you get extra bonus points. In that case, you're definitely going to want to look at the market numbers because, like we said, sometimes lower seeds are actually favored in the game and you're almost always going to want to pick that team, regardless of whether or not you think they're going to lose, because there's just such incentive for you to do so there's also pools where, like I was once in a pool where, like, if the 16 seed wins, then that's's 16 points. 

01:04:59 - Johnny Capo (Co-host)
If the one seed wins only one point, at that point like when I was running through that it was better to just take every single dog right. I mixed in some around the middle seeds. But if that's the case, then you have to look at the max amount of points that one seed can even get Right. It's just like one one, one one. So, as Rob mentioned, always read your rules. We should have mentioned that off the rip. So read your rules and then know who is in the pool with you. 

01:05:22 - Rob Pizzola (Co-host)
Yes, Ed, we will wrap up with our closing question, which we asked every single guest on the podcast, which you've probably heard at some point or another. But if you could go back five years and talk to a previous version of yourself, what piece of advice would you give to your old self version of yourself? 

01:05:40 - Ed Feng (Guest)
what piece of advice would you give to your old self? I would say that you, really you, you need to spend more time trying to help others, and I know that seems kind of trite, but like, um, maybe let me give you an example like of, of of my life, in, in, in, in, in my work, my work, right, like my work, has always kind of been football and um, you know, like after and march madness, and after march madness ends like I don't know. In past years I would just not, uh, you know, maybe do some nba stuff, maybe, like, uh I'm saying this really badly, but but now, the more I think about, the more I've changed over the last couple years is focusing on how. Focusing on, like, how can I help members of my site and people who follow me, right, and that is, knowing more about football and doing better football calculations than anyone else out there, and the you know, you know, the month of of April and May is the time to do that, and, um, I think I get a little bit better at that every year. I hope to continue to get better at that every year. 

01:06:53
But I think, if you can help other people, whether it's, you know, the people that are members of my site or you know your betting partners, like, if you just kind of focus on helping and how you make yourself helpful in the most useful way, like everything else kind of takes care of itself in terms of money, in terms of kind of satisfaction, and you can even apply this to like, obviously, like your life as well. Right, you can get the most benefit out of your marriage If you focus on helping your, your spouse. Uh, you can get the most out of your family and and your life, um, if you kind of focus on helping your kids, even if that means, like you know, the drudgery of spending an hour on a Saturday like filling out some form for, like, the soccer team, and I think that you know, I think a lot of times in life we, we get caught up in the rat race and where we're going and this and that and the other thing I know I certainly do. I think the thing that's kind of helped me the most is just just making it really simple, like waking up and saying, well, how can I most help members of my site and working on my business? How can I most help, like, my wife and and and her you know her job and her career and and stuff like that. 

01:08:05
And I think if you just kind of focus on that, everything else kind of takes care of itself. And, like you know, that doesn't mean like you have to work 52 weeks out of the year on on your business or whatnot, like you should take, you should take some time away. You should take a week to you know, think big and how you want to proceed in the future. That doesn't mean you don't take vacations or whatnot, but you know, when you are focused on work or the meaningful things you know, if you, just if you think about how you can sweat for others a little bit, I think a lot of everything else takes care of itself. Ed. 

01:08:37 - Johnny Capo (Co-host)
I'm not going to lie to you. It was a rough start to the answer for sure. 

01:08:40 - Rob Pizzola (Co-host)
It was really well said, one of the best. 

01:08:42 - Johnny Capo (Co-host)
That was our best answer. I think we've ever we've ever received right. 

01:08:46
Thank you for that advice. I think that's amazing, you know, just making yourself more valuable to others around you. You know, I didn't even, I haven't even. I'm just still. It's still spinning in my mind. Thank you very much, and that was an amazing answer. And thank you for coming onto the podcast and being with us here today. Before we close off, if you just want to plug your stuff one more time, so you've got that podcast coming out on Friday, listeners can find it. Is it the football analytics show, where that will be yeah, yeah, okay, so you go ahead. 

01:09:15 - Ed Feng (Guest)
It's very confusing this time of year because my podcast is the Football Analytics Show, but right now, during March, it's the Bracket Wisdom Series. So it's a daily podcast of about 10 minutes every day. It started Monday, march 7th and it'll go every weekday until the start of the tournament next Thursday. But it really is about football and football analytics because that is the bulk of my business and that's the bulk of of what I'm going to do. And uh, basically as soon as the tournament starts, my next guest is, uh, I don't know who he's going to be, but, but someone to talk about what the heck is going on with Russell Wilson and and nice and Aaron Rogers and and all that stuff. So my podcast is the football analytics show and then my site is thepowerrankcom. 

01:09:58
I think the best way to uh to interact there is to sign up for my uh email newsletter. So it's my sports betting email newsletter and um, uh, you know I try to make it three things valuable, concise and entertaining, and during football season it's mostly about football, um, um, but you know I'm trying to do some golf in the off season, started doing that last year and uh, it was, it was pretty good and uh. So the the newsletter is the best way to go, simply because, like this time of year, like I'm so busy that like I'll send out a newsletter and forget to post it on the site and so, um, you know, if you're on the newsletter, you can you get, get all my best stuff and and you can always just reply. If you have a question, just just reply to the email newsletter and I'll get back to you about whatever, so that that you can get that at thepowerrankcom. 

01:10:52 - Rob Pizzola (Co-host)
If you could go back and talk to a previous version of yourself, would you change the name of your podcast to make it less confusing? 

01:11:04 - Ed Feng (Guest)
um, no, no, probably gets a lot of good seo value. That's what I'll say it's. 

01:11:07 - Rob Pizzola (Co-host)
It is a good name for rank, like traffic. Yeah, it's like you got all the things covered. 

01:11:12 - Johnny Capo (Co-host)
Nobody is searching circles off ever, but people are definitely searching football analytics, so's he's way up on us right now. 

01:11:22 - Ed Feng (Guest)
Yeah, no, actually, rob, I would say no because I chose that at a point in my in the evolution of my business where I knew, like it wasn't like, like, my business is not ranking sports teams, right, my business is to help football bettors with better calculations, and I realized that right when I started the podcast, and that's why I called it the football analytics show. Maybe I should change it to the football analytics and betting show, because that's that's really what it is. But I think that title reminds me, like, like this is what we do, right, and you need to get back to talking about football. And you know, and that's a mistake I've made, I feel like, you know, like I haven't been good about getting episodes up in the off season. Um, I think it's fine to do some episodes that are not necessarily all football in April and May, but, um, you know, people care about it all year round and and, um, the title kind of helps me focus on on on what I should be doing on that podcast perfect, uh. 

01:12:23 - Rob Pizzola (Co-host)
I will plug the book one more time as well. 

01:12:24 - Johnny Capo (Co-host)
How to win your ncaa uh tournament, uh by rob zilley holding up a pockets bible size it's pocket. 

01:12:31 - Rob Pizzola (Co-host)
It would honestly would fit in my back pocket. It's it's a very quick read, but it's actually like there's a lot of detailed things examples in here which I like about it. There are visuals um, it's good. It's actually like there's a lot of detailed things examples in here which I like about it. There are visuals Um, it's. It's really good. Honestly, I'd highly recommend it for anyone who's just like if you want to have it on hand, you can refer to it every single year. I think it's really good stuff. So you can check that out on Amazon or Kindle as well. 

01:13:01 - Ed Feng (Guest)
Yep, um, appreciate it. Go ahead Just just like. If you're if you're on Kindle unlimited, you can get the book at no additional cost. Beautiful, and just download it on your Kindle. And, uh, if you are really fluent in analytics and you trust me that your odds of decrease, uh, your odds of winning a pool decrease exponentially with pool size, you can just skip the first two chapters and go straight to chapter three. Know exactly what to do for your pool. 

01:13:21 - Johnny Capo (Co-host)
Okay, dr Ed Fang, everyone. Thanks everybody for listening. We will see you all next week. Rate review five stars.