 This is Covering the Spread. Here are your hosts, Jim Sonnis and Dr. Ed Fang. What is going on everybody? Welcome on in to Covering the Spread. That's right here on the FanDual Podcast Network and NumberFire.com where today we're talking NHL with Rob Pozzola getting his thoughts on how he built his NHL model and his read on the current futures market. My name is Jim Sonnis. I am a senior writer and analyst for NumberFire.com. Joined here as always by Dr. Ed Fang. You can find his work over at thepowerrank.com. Now we're doing two shows today. We're going to talk about March Madness and the Men's Basketball Tournament in the second show with Drew Martin. So we'll get to that then. But Ed, you're vaccinated. What's going on? Yeah, I got my first shot yesterday. I live here in Ann Arbor, Michigan and not too far away in Toledo, Ohio. They are allowing people 40 and up to get their shot. So me and appointment drove down and still feeling pretty good today. So we'll see how long that lasts. Absolutely. If I don't make it to the second podcast today, you'll know why. Right. Exactly. If I don't make it, it's for other reasons. It's been a... I'm trying to recover from the Northwestern women's loss last night. So trying to steadily pick myself back up. So different reasons we may not have the endurance for a full two podcast for today. But I'm glad to hear you're doing well. I think that's been pretty universal, especially after the first shot. So it's important to get our vaccines in there so we can get back to, you know... I was thinking about going to a baseball game and the idea of... Like I hate going to games in person. I don't know if we talked about this before, but... I hate going to games in person. I think it's like a huge burden for me as like a daily fantasy player and like a better because it's so tough on my phone and all that. But I actually kind of want to go to games again. And that was a weird feeling for me to encounter. Yeah. Yeah, for sure. I mean, I think we all kind of want to get back to normal and we all need to live with this economy. And that's the main reason that I went down there as soon as I possibly could. And I obviously know it's not available to everyone yet. Right. But it will be soon. Things are, especially in the United States, things are happening pretty quickly here. So yeah, it was a big moment, you know, and it's kind of a crazy experience. Like, you know, they had a pretty well run. You go into this huge rec center. There's like dozens of tables set up. Right. And then you have you have to wait about 15 minutes so you don't like have a severe allergic reaction to the vaccine. And then they let you go. Yeah, I was thinking about this like two weeks ago. But like if you had told me a year ago that we were still going through this, I would have just been so sad. But if you had told me in December that we'd be to the point where vaccines are opening up to almost everyone. Like I have my appointment a week from today. Like if you had told me that in December, I'd have been blown away and I've been smiling ear to ear. So yeah, things getting better. That is for sure. And they're about to get even better. So we're going to talk to Rob Pazzola. You can find him on Twitter at Rob Pazzola. He is the host of a brand new podcast called the circles off podcast. We'll talk to him about that. We're also going to talk to Rob about building out his NHL model, what goes into it, why he goes to the player level model and then use that to see if there is value in the futures market across the NHL right now. As mentioned that we do have a second podcast coming up later today. If you listen to this on Friday, that podcast already posted breaking down the sweet 16 of the men's college basketball tournament with Drew Martin getting his thoughts on whether he sees value in the individual gains, but also the futures market there. So find that by searching for covering the spread wherever you get your podcasts and while you're there, make sure you leave us a rating and review as well. We'll talk to Rob here in just one second. But first, the sweet 16 matchups are set in Fandall and Amazon original movies are teaming up to give you more ways to win during March Madness. Enter the survive in advance contest presented by Amazon Prime videos without remorse and compete for your chance at the main prize pool. The contest is simple. Just select four of the 16 teams remaining in the tournament and earn points for every win. Those teams earned from the sweet 16 to the national championship game. The best part, the contest is totally free to enter. Game day will be here before you know it. So head to Fandall.com or download the Fandall fantasy app and enter the survive in advance contest presented by Amazon Prime videos without remorse today. And check out without remorse starring Michael B. Jordan on Amazon Prime video April 30th. Eligibility restrictions apply. Let's bring on Rob Pazzola now. Find him on Twitter at Rob Pazzola. Check out the circles off podcast. A link to that is in our show notes over on numberfire.com and get Rob's thoughts on how to build an NHL model in his read on the futures market covering the present. Let's welcome Rob Pazzola back into covering the spread to talk a little hockey Rob. It's been a benefit since we had you on the podcast. How you doing today? I'm doing well. I mean, I'm surviving the COVID landscape right now, but the weather is getting better. More sports around the corner with baseball starting. So good time to be a sports fan, but things are going well. Yeah, the better weather is nice. More things we can do outside, especially you in Toronto, me and Syracuse like it's been. It was pretty rough there for the past six months, but things finally feeling less constrictive, which is good. And you also get to ride a second high because your pet tortoise, Tortellini, what a betting competition last week. So we're talking to you today about NFL model or NHL modeling, how you considered incorporating Tortellini into the NHL model going forward. I have not, but I mean, based on some of the runs I've had so far this year, it might not be a bad idea. I mean, I know he's just coin flipping and whatever, but it was pretty amazing to watch him come out on top over the course of one week. So that was, that was pretty fun to do. I enjoyed it quite a bit and it kind of blew up. So yeah, I mean, really, really good stuff. Does Tortellini have a Patreon yet? He does not. He has a Twitter account now at at Torts the Tort. We had to shorten or abbreviate it, but not nothing else yet. Although, I mean, if if I ever get into a really bad rut with the sports betting, I'm sure I can monetize off of Tortellini in some way now. Right. You've got the Crypto punks. You've got Tortellini. You have options as far as monetization. So you said you had a rough run. It's no sweat. You know, you're all set with the Crypto punks and Tortellini. So there's there's really nothing to worry about here, right? It's it's I'm lucky that the the rough streaks have not been prolonged this year because there's always going to be rough streaks regardless of whatever you're betting on. You're going to have some some periods of bad variants, but things have always turned around and that has continued this year. So fingers crossed that that continues going forwards and I didn't just jinx it. Yeah, excellent. So, you know, Rob, we've had you on to talk about your NFL model, but not NHL. What are you doing for the NHL? Are you on a team level or do you look at it from the player level? It's player level. I think several years ago, I just came to the realization that when you're doing things at a team level and then trying to account for injuries and other factors on top of that, I just kind of thought it was a flawed approach in general. And I started to really notice it across several sports at different times where any team was dealing with like a cluster syndrome cluster injury situation where it's not as just it's not as simple as just subtracting the values of each one of those players. There's compounding effects of having multiple players out at one position or multiple players out for a team and it took some major losses to come to that conclusion. But since then, I've been aspiring to model everything I can at a player level. Most of the major sports, it's fine because at least you have good data available and you're able to do so. Some of the smaller markets, it's a little bit more difficult. I've ventured off into stuff like rugby as an example, trying to model other sports and you just really can't get to the player level for stuff like that unless you're willing to pay extremely large amounts for data. But I've kind of taken that approach across the board now where I think it's just really hard to beat markets at a team level at this point. Well, the cluster thing is important too, especially now with all with COVID contact tracing and stuff like that. Do you feel like that is giving you an advantage kind of being able to better account for that within your numbers? I do. I think I'm very good at accounting for that in general, but just the general unknowns of COVID this year have caused a different dynamic embedding. So normally where I would be very willing to fire on something at nine o'clock in the morning, maybe I'm less inclined to do so now because a team is either coming off a COVID break or they've had one COVID case within their team, which could potentially lead to more. So I think, yes, it gives me an advantage in terms of I think I'm very well equipped to model the game itself when I know everything that's going to happen in that game. But we've seen all sorts of late scratches this year. There are times where teams are warming up and they'll scratch someone in warmups. And that creates a problem just in general because I'm pretty reliant on having lineups available for hockey in order to project the games and when you lose that ability, you kind of either have to take a shot earlier on in the day and just deal with the fact that sometimes it's going to work against you, which it has a lot of times this year, or you just wait till you have all the information, but then someone else might have beat you to the market at that point. Right. So you may have to deal with getting a less efficient line, but also that can be offset because if you get a line before COVID news is announced, it's hideously inefficient. So it's kind of a tradeoff in that sense. Right? It is. Like I still very much lean towards betting early just in general. My general philosophy is that, you know, half the time the news is going to work in your favor and half the time it's going to work against you. And I do believe that it balances out over the long run. I do feel that I've been unlucky this year on new situations where more often than not, I've gotten hurt by the late COVID scratch or some surprise news because you have to remember beat reporters weren't with the teams at some points this year. There's less beat reporters with the teams in general. So news is just harder to come by. Sometimes you'll go an entire day without a beat reporter for a team, even tweeting out a lineup or a projected lineup or something like that. But I'm still of the philosophy. I'm better off betting early, but it can certainly be frustrating. Obviously, when you're putting a large amount of money down on a game and then something happens where it's just not what you projected. Yeah. Hey, Rob, I wanted to ask you a little bit more about the player modeling. Is it like if you had to compare your player versus team, because I presume you're still running your NHL team model. If you had to compare them, like is it that the player model is just better at finding inefficiencies compared to, you know, predicting the outcome of every game? Yeah. So I mean, with everything I do, I'm tracking error metrics when I run a model, right? So every model that I produce, I have some way of being able to back test that against previous seasons. It makes it a little bit more difficult this year because I've had to make some major tweaks this year based off of the COVID landscape, I guess you would say for NHL, where the back testing might not matter as much. But for me, ultimately what it came down to is I'm back testing my error metrics, whether that's log loss or whatever you want to use the complete suite and the player level was outperforming pinnacle closing lines, not only my team level model, but outperforming the closing lines as well. So at that point, I was just like, I might as well go this route. There makes no sense to continue down on the team level. I still do run the team level stuff and honestly, I don't know where I stand with it because I'm not making plays off it or not. I think probably I could beat like an overnight market with the with the team level stuff, but day of when when limits come off of the games, I think it would be an extreme challenge. So that's specific to your NHL player and team model? Depends on the sport like MLB. I've always done at a player level because I don't think that there's really a good way to market it to model it at a team level. And then football would be the same thing, but football. There are inherent challenges with backtesting games as well. So it's tough at the end of the day, though, I do have a lot more confidence in being able to to do things at a player level now and say this is what I think the number is on a game. And in some cases, there are some pretty drastic differences between doing it at a player level and team level. And I dig into those as well to see what's causing this difference. And if logically it makes sense to me, then then I just roll with it. That's interesting. And I think that NHL specifically is interesting because as someone who doesn't, you know, who hasn't bet it heavily, I'm curious about like the impact of like line changes and because that's player meshing and for a player level model that can definitely impact things. Do you have to adjust when teams make adjustments to their line? How can you model for that? I guess if you're explained to someone like me, who is very much a newbie to this, there's only there's a level of sophistication you could get to. And then there's a point where you're like this is just too much or it's not worth the effort. So for me, like lines are important, but first and foremost, it's like a modeling the NBA where I would project what I think a player minutes are going to be for a player. I do the exact same thing for NHL. Now, part of that is based off of the line combinations that includes whether they're on the power play or penalty killing and then some of that is coaching tendencies as well because some coaches will tend to let's say lean on their best players a little bit more, especially when they're trailing. Some player coaches will tend to use for lines a little bit more. So I'm really trying to get as accurate a minute projection as I can on each player and then everything else just falls into place from there. Okay, perfect. So let's go back to you when you were initially building out this NHL model. What were the biggest difficulties for you in building that one versus building one for MLB for NFL or any other sports you may have built one for? I don't think I honestly encountered any major roadblocks. I think I had like, I always draw out my logic on paper just in general of like, this is how I think I should do this and I took computer programming when I was in college and I wasn't a very good programmer, but that's just kind of the way I think it's from a very logical point of view. So I kind of map it out that way and then I go through it and I never really encountered any roadblocks in the actual modeling process. It's it's over the course of the years where like integrity of data has was a huge issue for me last year because the entire data set for the NHL in the first month of last season was completely wrong in terms of where they were tracking shot locations on the ice and for someone who is running some sort of expected goals for players, I'm very reliant on those shot locations being accurate. And then you'll start to learn things about the tracking is different rank by rank, depending on who's actually tracking that data. It's it's no different in baseball with some of the stat cast data where there's I think it's Detroit's ballpark or something that is very a very hot gun. Yes. Right. So and I learned that over the course of the years as well. So it's you have this data set and you assume that everything there is just I mean 100% you're 100% confident in the data, but sometimes you're there's reasons not to be and and that's kind of something that's really they've really honed in on over the course of the years is whenever I have outliers in my model. If I'm constantly betting on I don't know and be over in the flyers games regularly and the market is not reacting. I really start to dig into those and the data that's leading to that for one. And then if not if it's not the data that's causing it, what logic is causing this to happen. So I never really had yeah, any any real roadblocks building the model, but along the way definitely stuff has come up and like this season as an example is very different because and it's posed some very interesting problems for me because with the way that the NHL structured this year, you have four divisions that are only playing games within their division. They don't play the the other teams in the league. So you almost have to treat the league like it's four different leagues. Yeah. So, you know, if I use an example and I learned this the hard way very early on in the year this year, but there's been an average of roughly 5.85 goals a game in the NHL this season. Three of the four divisions are actually below that. Only one is above that, which is the North Division. So if I'm treating the entire league is having the same scoring environment. I get to what I got to in the third or fourth week of the season where I'm constantly playing the North Division under and I'm constantly playing the central, which I think is the lowest scoring division over regularly. And I'm like, hmm, what's going on here? And then I dig through the logic as an example. And it's like, ah, this is the mistake I've made, but I would have never had to deal with that before. So this year, there's certainly been probably two or three different examples of stuff that I've learned along the way, expensive lessons I would call them. I'm kind of hoping that at some point down the road that we do have another season like this. So at least I'm kind of ahead of the curve and dealing with it. But even if we don't, at least it's an exercise in in, you know, modeling properly or just learn teaching yourself something new. Yeah, no, that's fascinating, Rob, because because you're doing this player modeling, you figured out all these tricks with average goals in the division. But that should actually probably help you in the playoffs when these teams start actually playing each other, right? Because someone who does like a team level thing. There's there's no data there to mesh any of those four divisions together, right? So you're probably going to come out ahead then. Yes, yes and no. So by the time we get to the playoffs, we're likely going to have a very different environment than we have now because we started the season with zero fans in the stadium. And now we have some fans and by playoff time it's possible that entire stadiums are filled and then maybe you're not comparing apples to apples. If you're starting to use the regular season data and it's it's tricky. I mean, there are times where I'm I'm literally I would call it throwing darts at a wall, but it's it's educated guesses, right? And either I can take this stance that I'm just not going to bet this because I'm not confident enough or I can take the stance that no one else is going to be confident enough in this. Maybe I can take an educated best that guess that's better than than others. So we'll see what comes down the road for playoffs. Typically, I've used NHL playoffs is kind of like a break period for me because the regular season is a pretty big grind, but if I feel like I have a pretty strong edge there or I can account for something better than than others then certainly I'll I'll I'll go down that route. Excellent. Well, let's talk here about the actual data that you're looking at here and obviously it's player level. So that does change the conversation a bit. If someone is looking to get into NHL modeling, they've not done in the past. What data should they be looking at? And how long does it take for that data to actually stabilize and become sticky? Yeah, there's a there's a couple of sites that I would recommend for data first and foremost. There's one called Devolving Hockey and the other one's called natural stat trick which will have the full data suite that you actually need. Like you can build your own data set obviously by using NHL play by plays and doing that but for the average person that's going to be very overwhelming. It's going to be time-consuming. There are inherent advantages to doing that because you're not reliant on on someone else. We've seen data sites in the past that get bought up by teams before and then they just go offline overnight and then you're kind of I mean you're kind of screwed in that sense. So there are advantages but there's a bunch of them out there now that you can use. I think first and foremost when you're when you're looking we've kind of seen like this evolution towards expected goals in hockey. It's not a perfect metric but it's the best that we have right now in terms of an indicator of what a team should have scored based off of their their shots from from all over the ice and the expected goals is going to differ depending on what site you look at but they're relatively going to be in the same range because each person can build out expected goals and they're in the same way but that's certainly some you know point where you want to start because there's so much variance in hockey right it's it's such a low scoring sports or a low event sport that you see runs in the sport that are unlike what you would see in anything else really so expected goals always brings you right back down to reality now there's other metrics that can tell the same story scoring chances high danger scoring chances Corsi which is a possession based metric basically the amount of shots that you take and now we're starting to get actual real possession based metrics which is like offensive zone time just in general how often you spend in the other team zone and things like that but there's really a ton of different things in general is that it might seem very overwhelming because there's so much data you can filter so many different things but they really have like really good glossories as to what these metrics are and they explain them pretty well so somebody could go through them like that in terms of how long it takes to stabilize it really depends on what you're looking at like it's roughly going to take you 15 to 20 games into an NHL season before you can really truly say that this team is likely going to perform at this level going forwards but there's other things to account for whether they've had players out during that period what their goaltending play has been like because goaltending is highly random over the course of a season so roughly 15 to 20 games is a pretty good you know event size to to fall back on but that that can really differ depending on what you're looking at and it sounds like a lot of this is data that's come into the collective knowledge semi the way that you model things what's the process like for you when new data points pop up investigating that data and deciding if it's something you want to incorporate within your own model yeah so at first first thing I do is is I'll find that data so I used to run a model that was strictly based off of Corsi actually even before that strictly based off shots on goal when I this is 10 years ago shots on goal which I mean I can laugh about it now but at that time it was good enough to be the market and then I moved on to Corsi and then I kind of moved into expected goals so what I'll typically do is I'll I personally my my process is to apply the changes that I think are necessary and then back test it and I I check my error metrics obviously and then I I run a test against a full season of closing lines to see how I would have fared against the pinnacle closing line or just a sharp books closing line in general and and if it didn't I'll pull it it's not that I won't explore it again down the road but there's been times where I've I've made some changes and it just didn't test well and I was like okay you know what that's it is what it is I thought this would test better and it didn't and that's kind of just the goal for me is continuously improve error metrics going forwards excellent so Rob it's I mean it sounds like with the with the type of data you're talking about you're getting more on offensive players is there Oh defense this is this is where I think there's an advantage to be had for sure in hockey right now because most of the data available is is definitely geared towards offense I mean you can obviously flip every metric right if if you have an expected goals for the the opposing team as an expected goals against for the and so on and so forth so you can flip anything but there's really not a lot of great metrics in terms of measuring defense I think you see the same thing across baseball where depending on which metric you're going to use you can get a highly variable results either a guy can be an elite defender or an average defender and you're like well now the none of that really makes sense we're kind of at that point with hockey right now where there are defensive metrics out there but there's no consistency across the board and how you would rate a defender I think if if there's an area that's really going to improve over the course the next two or three years especially with the NHL player data that's going to become available for next season as well the official licensed player data there that they're tracking and speeds on ice and and things of that nature it's probably going to come in the form of defensive improvement excellent yeah because the you know expected goals against is again like a team metric right which is exactly what you're trying to get away from right and and trying to figure out exactly how an individual player impacts their teams expected goals against or I mean these are all challenges that you face in general how much how much was to do with the defensive pairing that they were in was his other defenseman dragging him down or weighing him up and the same comes with forwards right how much was to do with him playing with two good forwards on the line or too bad forwards and it's tough I mean it's time consuming but at the end of the day I I want to try to get as close to a true probability as possible so and the unfortunate part is separating player from context is something that translates to literally every sport you could possibly bet on so difficulties across the board but at least a good foundation here people are trying to get into NHL especially with the playoffs just around the corner let's talk about the future's market here Rob because it's a good time to do so I think with the playoffs kind of taking shape here for sure looking at the Stanley Cup Odds over at Fandell Sports but we got the avalanche and the lightning both six to one got the Maple Leafs and the Knights at plus 750 so there's kind of this group of four at the top do you see any value in those four teams or is there value elsewhere in the market or is it one you're just kind of staying away from right now so I do see value in a couple of these teams actually and I think ordinarily when people think about hockey they think about a sport with a ton of randomness and that's that's true there is a ton of randomness they're not wrong about that historically it probably worth taking a flyer on a few long shots here and there over the course of the season but I do think that this year especially there's a pretty sizable discrepancy between Tampa Bay Colorado and the rest of the league especially when you consider that Tampa Bay is likely going to get Nikita Kutcherov back for the playoffs who is literally their best forward that they've been playing without this entire season on route to an amazing record so you look at at six to one which is what Tampa Bay and Colorado are right now that's an implied probability of just over 14% for both of those teams I have both of those teams roughly 20% or higher right now so this is highly unusual that I would be recommending two teams at six to one to win the Stanley Cup but I think either of those are a good bet right now if I had to pick one of the two would be the Avalanche I know Colorado's trailing Vegas in the division right now but they suffered through a rash of injuries so far this season including the season because you have way more back to backs if you have to go to a worse backup goalie you're more likely to lose that game and I think Colorado's really suffered from that but they're arguably the most stacked team I've seen in the NHL in a long time definitely in the last decade or so they can roll four lines they have some elite defenseman Phillip Grubauer who's their starting goalie has developed into an upper echelon goalie so I think Colorado's definitely the team to beat this year six to one seems like seems like excellent Fandall sportsbook also has some division odds up and all four division seem very far from decided are there any numbers that you see as an tages here I do I'm looking right now so Boston is plus 330 to win the East which I think is a fine bet that's roughly about 23 percent implied I think they're closer to 30 percent true probability there if you're looking at the standings you're you know it's going to look like Boston's pretty far off the division lead but they have a bunch of games in hand that they have to make up because they had a covid situation last week and missed a bunch of games and that's going to lead to some rough rest situations for them later on in the year which I'm a little bit scared about but they arguably have the best gold tending tandem in the entire league with Tuca Rask and Yaroslav Halak so you know you're going to get good gold tending on any given night even and if you have to play you know seven games in 11 days or whatever so I think that you know while the condensed schedule is going to hurt some of the aging veterans on the team they're going to get good gold tending everywhere plus 330 you're more likely to lose the bet than to win it but I'm just looking for value and I I still think Boston can go on a late season run here this is where the player level modeling comes in too because you can kind of suss out the teams may have benefited from the truncated schedule then we go to the playoffs and it's a different schedule you can separate that out and it's something that we've seen a lot in the NBA with rotations and stuff like that rotations timing during the playoffs is made models kind of or some models certainly be pretty far off so I think that that's going to be advantageous for you in the playoffs the player level model specifically there I think so I mean I think people who generally struggle in the playoffs they're kind of using the same exact model as a regular season and just thinking that the it's just going to apply to the playoffs but the reality is in every single sport the playoffs change the NFL which is plays very much like the regular season in general but certainly the NBA and the NHL the games are much lower scoring because there's less of a referee impact that's just proven over the the course of the years they let the teams play a little bit more in general so there's a lot of things that you know that don't necessarily apply and I mean I think that's just an important point to stress in general when if you are betting post season in any sport don't assume that that the same things from the regular season the NFL is running back usage as far as snap rates but that's not going to make a whole heck of a lot of a difference in terms of trying to model things out so the final market we have here as far as futures over an NHL Fandall Sportsbook is betting on teams to make or miss the playoffs and the juice on these is a little little steep at times but are you seeing any value in that market right now as far as teams make you missing the playoffs I I am I mean I have no problem laying laying juice or vague if if I feel that there's value there and I'm definitely say that LA to miss the playoffs is minus 220 that has some serious value right now like that should be priced closer to minus 400 something like that I mean they're five points back of a playoff spot they're five points back of St. Louis for the final spot in the West but the blues are one of the teams that have been decimated by injuries this year they're slowly starting to get back to full strength they got Vladimir Terasenko back last week they're going to get the defenseman back this week as well it's a team that's really bordering on be getting back to full strength in the course of the next few days or so and frankly when they're at full strength they're just so much better than LA and not only does LA have to catch St. Louis but they have to pass Arizona who's in between them as well who has a chance so yeah like looking at this minus 220 I think is roughly 70 percent implied for them to miss the playoffs it's much higher than that it's we're talking 85 90 percent here so I think that's far and away the biggest standout I guess in the same vein you could you could take a look at St. Louis minus 175 to make the playoffs on the other side of things but I'm not as enthused on that because they could get a push from Arizona who is that that team in between so I do think that the Kings is the safer of the the two bets there but I mean you could potentially if you're willing to double dip there's some sort of correlation there as well where I think you can you can kind of get both that correlated markets absolutely take advantage of that whenever you get the opportunity to do so because we don't always get that chance for sure now Rob before I let you go I did want to give you a chance talk about your new podcast it's a circles off podcast and pretty new so for those who have not listened to the circles off podcast yet what are you doing over there yeah so I think there's kind of like a gap in the market right now for kind of on the pod but just in general like most the podcast you hear are general picks podcast so a lot of daily stuff we've kind of taken a weekly approach where we're discussing topics that seem to come up a lot whether that's through DMs on Twitter people reaching out to us in general and just trying to like really educate the population on on ways that they could take advantage rather than focus on strictly here and go out and bet these types of things so it's lead to some entering interesting discussions in general but the feedback has been good so far and I'm I'm enjoying it quite a bit okay perfect you can find the link to the circles off podcast in the show notes up on numberfire dot com or just search for wherever you get your podcast that is Rob Pazzola makes you check him out on Twitter at Rob Pazzola Rob we appreciate the time and kind of diving deep into the way you build out your NHL model really appreciate that good conversation good luck to you for the rest of the regular season the playoffs and everything else you're betting to hopefully we'll talk to you again here soon thanks guys thanks for having me covering the future big thank you once again to Rob Pazzola for swinging by and breaking down everything that he discussed and had I love getting to hear about what goes into a model and I think that we talk about this when we had Rob on to talk about the NFL the value of a player level model but I think that the more it's the way you should be doing things I the more you hear about the more you hear the value of it I think that you've been tinkering with that too like toggling players on off and stuff like that to an extent as well is that correct and not really kind of falling behind on the player level stuff but yeah no I mean it was interesting to hear Rob talk about his air metrics because that that's kind of the gold gold standard right right you want to check how they do in every single game and you know I'm a little surprised that the player level does that well he's clearly doing something right and the way he's meshing things and I kind of wanted to dig a little deeper into how much you know whether he's thinking about players as individuals and just putting together or how much you know like how much teamwork kind of plays a role because you can actually look at how well to and three player groups work together I've done a little bit of that for the NBA but yeah there's a lot of interesting things and I think just you know when you say you back tested and the error metrics are better than clinical like that that gets my ears to perk up and get my interest and for you when you're testing your model what error metrics do you use to test the efficacy of it yeah I mean I'm always looking at root mean squared air so you know you take you know in the NFL in any football game you look at the actual game result and then you subtract what your prediction said and you square that and from my academic time I mean there's a lot of different ways to do it but when I was looking in the academic world just like your squared air has a ton of good mathematical properties and there there was a lot of work there that said like this is a good thing to do so so it's pretty simple it's relatively easy to explain and that's what I'm using to quantify all my models okay perfect so you check out what Rob is doing over at the circles off podcast checking out on Twitter at Rob Pizzolo let's dive into covering the spread or covering the future for today and again we have two shows coming up today here on covering the spread we're talking sweet 16 later on the Drew Martin but you wanted to start things off here by talking sweet 16 for your covering the future as well talking about one of the darlings of the opening weekend here and Oral Roberts what are you seeing for their match against Arkansas on Saturday yeah I mean honestly not a lot Oral Roberts has been made a great run so the second 15 seed to make the sweet 16 when they actually made it I was like yeah they're the first right and then someone quickly reminded me about Florida Gulf Coast I kind of forgot that they don't city man yeah don't see so first of all yes so clearly don't city the internet has told me that don't city is the proper name for that Florida Gulf Coast team but I thought we had anointed them as Lobb City because they were throwing so many lobs that year especially in those first two games and the but Florida Gulf Coast 2013 so the second team 15 seen to make the sweet 16 and they have that little guard Max it's app it I'm I'm going to say this wrong but Asmus and then and then they have Kevin O'Banner so these two players have have been really good leading the teams past Ohio State and Florida but you know let's get a little bit into some of the metrics like why exactly they won and the reason is that the three-point defense has been incredible so Florida and Ohio State shot 7 for 45 7 for 45 from 3 that's 16 percent and one thing I talk about my bracket wisdom series is just how unsustainable three-point defense is it's it's kind of remarkable so you know anytime you have three-point defense as low as 16 percent you're definitely going to see a lot of regression to the college basketball average it's been it's been quite a defensive run this is unsustainable in the next podcast I'll talk about how this is unsustainable for another team as well the exact same metric but you got to remember I mean this all Roberts team finished fourth in the regular season in the summit league and they only made the tournament because they beat two teams ahead of them South Dakota State and North Dakota State by a total of five points in the conference tournaments and you know they've only one their two tournament games I think both by three points so they're winning close games again another thing that's kind of unsustainable especially when you have this kind of three-point defense and then Arkansas is a really solid team it's an athletic SEC team they're pretty good on offense they're actually they thrive more on defense their 11th when I look at my adjusted points per possession the offense isn't as strong but a respectable 29th so definitely kind of worthy of of that three seed you know and they have a 66 wing and a lottery pick in ESPN so you know the numbers I trust the most like Arkansas by more than 12 points so I definitely like Arkansas minus 11 and a half also I also like the moneyline here too because I have it at about 91 percent win probability so yeah I don't see much much coming from oral Roberts and I like Arkansas in this game if you ever need we heard Rob talking about implied odds too and I know that free people who are new to betting may not know what different numbers imply if you go to oddsfire.com you can go to numberfire.com slash odd slash converter you can plug in the American odds it'll spit out the percentage odds and the fractional odds the moneyline implied odds in Arkansas right now 86.1 percent and you said 92 or 91 percent on them models yeah so 5 percentage points of edge there you're going to take that pretty much every time because you don't find that number very much so if you need help with that converting those numbers numberfire.com slash odds slash converter to get that and I'm actually going to talk about some of that actually no or a different thing but either way it is helpful if you are new to this trying to convert those numbers into different probabilities and again more college basketball talk coming up later today with Drew Martin and I will be sobbing over my Big Ten assumption lineups not going to be fun there for sure my cover in the future for today this first podcast talking about baseball because we are one week away from opening day and I want to talk about a guy we discussed earlier a couple weeks ago with Ariel Ariel Epstein I was trying to get her to bite on betting Jean-Carlo Stanton to win the AL MVP and Ariel said no and I fully understand that because he gets hurt a lot she's a Yankees fan doesn't want to jinx it I get it but there is another route to get exposure to Stanton that offers a little bit more flexibility and that's betting in a 20 to 1 to lead the league in home runs those odds are tied for 11th highest in baseball right now they've shortened to bits so make sure you do some price shopping see if you can get a better number than or a number better than 20 to 1 out there but I think there is a chance this number shortens even a bit more before opening day because Stanton he was hurt a lot last year which is why this number is so long but when he was healthy he was just hitting lasers his 18% barrel rate would have ranked 3rd in baseball had he qualified one of the guys ahead of him was Miguel Sano who strikes out to a point where can't really expect him to lead the league in home runs because just doesn't put enough balls in play barrel rate is based on number of balls in play so Stanton has has a leg up there his barrel rate through 18 battered balls this spring is 22.2% it hasn't led to a huge isolated slugging percentage but the battered ball quality matters more here ISO is actually a number you can look at during the spring power numbers and strike out numbers are numbers that can stabilize over to 50 to 60 plate appearances so you can look at those we're not going to see that with Stanton right now the barrel rate very good for him he's likely to be the DH the Yankees and most of the regular season that'll keep him healthy hopefully not playing in the field and they said they're open to having to play outfield the games where they lose the DH playing in a National League Park as of right now there is no universal DH so do you want to make sure that guy will be out there when they're facing the National or playing the National League Park does sound like Stanton will be doing that during the regular season healthy hitting the ball well in a hitter friendly park probably going to play pretty much every day I do still think there's value in him at MVP to an MVP at 35 to one but that's an award that involves voting which is not super fun always it includes performance on defense Stanton won't be moving the needle there by any means so I think the better market in which to get exposure to Stanton is to lead the league in home runs there's no voting there it's all about power and we know that Stanton has that so I'm Stanton at 20 to one to lead the league in home runs for this year and we'll see how things break down and if you had time to look at baseball yet given we've been in the throws of March for so long no no I'll be looking at I'll be getting some of my stuff together for next week hopefully so okay perfect we'll look forward to that we'll have a show next week to talk about the final for the masters just around the corner so a lot of good stuff coming up here in the not too sure you are subscribed to cover in the spread wherever you get your podcast once again if you have not yet listened to our podcast with Drew Martin you'll listen to this on Friday go check that out on this exact same feed and while you're there leave us a rating and review as well big thank you to Rob Pazzola find them on Twitter at Rob Pazzola check out the circles off podcast that is a link to that on the number fire podcast note also you can find that by searching for circles off wherever you get this week over at the power rank yeah talking about sweet 16 on my email newsletter you can get more of my sweet 16 predictions so check that out at the power rank dot com you can find Ed on Twitter as well at the power rank I am at Jim Sonnis J I M S A N N E S you can also follow the Fandual podcast network at Fandual podcast big thank you to our video producer Calvin Theobald for running the video side of things here today thank you cow as always and thank you to everyone for tuning in good luck to you if you decide to build out in an hl model or dive into any of the futures bets that Rob mentioned as well good luck and I hope it goes well we'll talk to you once again later today to preview the sweet 16 this has been covering the spread right here on the Fandual podcast network