An annual meeting scheduled for Chris Andrews’ office March 15 was called off. Selection Sunday never happened this year, so Andrews had no need to huddle with his inner circle of oddsmakers to post opening lines for the NCAA tournament.
Andrews, the South Point sportsbook director, holds a similar meeting before opening college football lines in the summer — another tradition that is postponed.
But a meeting of the minds will happen again someday, and Andrews’ guest list always includes Richie Baccellieri and Vinny Magliulo, each with about 30 years of oddsmaking experience.
The drill goes something like this: Andrews presents a game, and each oddsmaker puts forth his number on the matchup. Magliulo might say 6.5, Baccellieri 6 and Andrews 7.5. The three hash it out and decide on the opening line.
“Vinny has his methodology, and I’m sure Richie has a totally different methodology, and that’s good,” Andrews said. “We don’t want to all agree on the number.”
That’s essentially a simple example of how a point spread originates. But how each oddsmaker arrives at his number is a much more complex process that involves his set of power ratings.
It’s important to understand no handicapper or oddsmaker uses the same formula to develop power ratings. Envision the nuances of body language and facial expressions or the uniqueness in fingerprints and snowflakes. Everyone takes at least a slightly different path, and sometimes opposite approaches lead to the same number.
“The way it works for me is not necessarily the way it works for you,” Andrews said. “Everyone has kind of got to do their own thing.”
Every sports bettor should make his or her own power ratings, which are especially valuable when wagering on college basketball and college football and dealing with the hundreds of teams necessary to handicap.
“It’s time-consuming, and you have got to be committed,” Baccellieri said. “It’s not difficult. It’s supposed to be fun, and people are supposed to do their own work.”
The first step in making your own numbers on games is to develop ratings as a guide to gauge the comparative strength of teams.
“It’s a numerical value applied to each team,” Magliulo said. “How do you come to the numerical value? Not every sport is the same.”
Baccellieri, a respected professional bettor and former sportsbook director, has been making college football ratings for almost 30 years. In the beginning, he studied Jeff Sagarin’s ratings (Sagarin.com) published in USA Today and used those numbers to help build his own system.
“I started with the Sagarin ratings,” Baccellieri said. “Find a set that works, make a set that works for you and make your adjustments. Eventually over the course of time, whether it’s one season or two seasons, you have your own ratings. I don’t think there are any secrets.”
Louisiana State finished the football season with a rating of 80, the highest number on Baccellieri’s chart. LSU’s rating to start the season was 69, but what factors went into that number?
“I would look at about 10 categories, and I would give each of the 10 a rating,” he said.
Each position group — such as offensive line, defensive line, linebackers and wide receivers — gets evaluated and assigned a rating on a 1-to-10 scale. The best position groups rate about 7, Baccellieri said, and “the best teams are somewhere in the 70 range.”
Alabama and Clemson started the season above 70. The worst teams, including several from the Mid-American Conference, started with ratings in the high 30s to mid-40s.
“That’s a starting point to get your college ratings,” Baccellieri said. “I have such a strong set of ratings for these teams for the past 20 years, and they don’t change that much, believe it or not. I don’t have to do a lot of adjusting. A few teams overperform and a few underperform, and you make those adjustments as the season goes on. But teams go back to their norms. I have a lot of years worth of data on these teams.”
Similar to most handicappers and oddsmakers, Baccellieri uses his ratings from the previous season as a base for the next season.
But a beginner making them from scratch needs to first determine a ratings scale and then prioritize categories that go into building the rating.
“Usually, I try to shoot for 100 as the high-water mark,” Andrews said. “What’s the most important thing to you as you start breaking it down?”
Clemson and Ohio State opened last season with ratings of 98 on Magliulo’s scale.
“I’m not afraid to get to 100 or go through 100 on a scale,” he said. “The lowest (teams) will be maybe 50 to 55.”
Magliulo starts by rating the top teams and works his way down. He examines factors including last year’s finish, the coaching staff, quarterbacks, experience, the quality of the depth chart and more. It’s common to place added emphasis on coaching and quarterbacks.
“I put a lot of credence into coaching, the values of head coaches and also of the staff, particularly the coordinators,” Magliulo said.
Andrews and Magliulo are old-school in their approaches. The new school is analytics. Math-based handicappers typically develop algorithms to calculate the strength of teams. But few bettors are qualified to work for NASA and have the ability to write computer programs that resemble the launch codes to a space shuttle.
As Magliulo said, “Don’t overthink it.”
It’s possible to make accurate numbers and win without developing a sophisticated math model.
“I’ve been doing it for 40 years, so I know what I’m looking for now,” Andrews said. “What I use is not a math model; it’s a recordkeeping device. A mathematician would laugh and say, ‘That’s not a math model.’ What I do gives you a very basic starting point.”
Andrews analyzes offense, defense, special teams, injuries, statistical differentials and various other factors while forming an opinion and arriving at a team rating or a number on a game.
“Algorithms are not the end-all, be-all, in my opinion,” Magliulo said. “I don’t get into the algorithms. There’s something to be said for feel. Algorithms don’t breathe, and they don’t have eyes.”
It can work to merge old-school evaluation and situational handicapping skills with new-school math.
In creating college basketball power ratings, I take the evaluation approach used by Andrews and Magliulo while also implementing some advanced analytics. But I do not have a math model. Ken Pomeroy (Kenpom.com) has arguably the most reliable set of ratings, so I reference his offensive and defensive efficiency ratings and schedule strength numbers to help form my team ratings.
The top of my scale is 100, with five categories — coaching, point guard value, offense (with an emphasis on 3-point shooting), defense and intangibles (depth, overall talent and schedule strength) — each worth 20 points.
Kansas started the season as my No. 2-rated team at 92 and finished the season No. 1 at 96.5. Dayton required by far the biggest adjustment, starting at 78 in the preseason and moving to 88 in late November before finishing the season No. 3 at 93, only 3.5 points below the Jayhawks. In-season adjustments to that extreme are rare, but preseason ratings involve more guesswork and are the toughest to make. I simply made a mistake with the Flyers’ first rating but was not stubborn and was quick to fix it.
Baccellieri said he bumped up Hofstra’s rating five points during the season.
“That’s a big adjustment, but that’s not overreacting,” he said. “The discipline with making ratings is not to overreact. I think that’s more of a talent than anything else. You can’t overreact. You are moving the rating a lot more early in the season. But the way you don’t overreact is you’ve got to throw out final scores.”
While many bettors overreact to final scores, Baccellieri said, analyzing box scores is the key to getting the right read on teams. Study the game flow, shooting percentages, turnovers, foul trouble and injuries. Read between the lines to find out why a team might have overperformed or underperformed in a certain situation. The next game could be a totally different situation and outcome.
“I use a combination of math and feel to make my numbers,” said Erin Rynning, a professional bettor who focuses on the NBA, NFL and college football. “I’ve made a myriad of tweaks and adjustments along the evolution path. If totals will be strongly considered, a must is having an offensive and defensive rating for each team. Again, one needs to figure out what statistics or numbers are crucial to future performance and weight accordingly.
“In a nutshell, your numbers or power ratings should equal a predictive final score to weight against the betting line. After a game is played, an adjusted game score from the box score should be used in adjusting for a new rating. In using the data from the box score and/or certain analytics, one can convert to an adjusted game score.”
Kenny White, a veteran Las Vegas oddsmaker, uses algorithms to make individual player ratings that he displayed in his first college football power ratings magazine last year.
“I do have a model that I run for projecting final scores,” White said. “The main rating I use is my player power rating system. You’ve got to rate quarterbacks a lot higher than you do position players. A quarterback is probably worth about four times more than a position player.”
White’s player ratings help form offense, defense and special-teams ratings that result in the overall team rating. Before the 2019 season, Alabama was his top-rated team at 129, with quarterback Tua Tagovailoa assigned a player rating of 11. LSU’s rating was 119, with quarterback Joe Burrow getting a rating of 6. By the end of the season, those ratings had basically flip-flopped on the quarterbacks and the teams. The dramatic improvement of Burrow and the Tigers took almost everyone by surprise. It’s an inexact science.
“If you put a number value on each player, if that works for you, that’s great,” said Magliulo, the Gaughan Gaming sportsbook director and VSiN oddsmaker. “To me, the numerical value is what the player means to the betting line.
“You have to adjust the ratings, in the case of football, on a weekly basis. You have to be willing to adjust numbers on a game-by-game basis but not overreact. You can’t just look at scores. Part of that discipline is not overreacting to wins and losses.”
Magliulo advises beginning bettors to start by developing ratings on the NFL, which is easier with only 32 teams, or by focusing on making ratings on one college conference and expand from there.
“As a bettor,” Magliulo said, “you look to see what the oddsmaker posts and how it compares to your number.”
That’s the reason to make power ratings, which are the starting point to everything.