For wiseguys, a summer of numbers
Chad Millman
ESPN Insider
Through the picture windows in his pad atop Russian Hill, Dr. Bob can see nearly all of San Francisco. To the left he can see Alcatraz, former home to the most misbegotten of gamblers. Then he can swing his head around for a look at the hills of the East Bay and Coit Tower rising from Telegraph Hill and the Bay Bridge, which disappears amidst the industrial cranes that dot the landscape. Finally, he settles on downtown, rising in the distance like a set of Legos. It's the kind of panorama they print on postcards.
And it's a good thing he enjoys the view -- because he'll be staring at it nonstop for the next eight months.
You want to know what makes wiseguys wiseguys? How, like Dr. Bob, they end up profiled in the Wall Street Journal or E:60 or in this column because of the way they move lines the second they post their picks? Why they can afford fancy apartments with breathtaking vistas in expensive cities and trips to Turks and Caicos?
<OFFER>It's because all summer long, while the rest of us are BBQing or golfing and knocking off work a few hours early, they will stare at spreadsheets, tweaking stats and formulas like the busiest disciples of Einstein, looking for the model that offers the slightest advantage. They do not just flip through previews a couple weeks before the season. They do not turn on their computers as Week 1 of the college or pro football season approaches and ask for an answer. I remember talking with Sal Selvaggio of madduxsports a few years ago. It was summer. His wife was about to give birth to a daughter. And he was combing the Internet for Blue Ribbon college basketball previews that were a decade old so he could fill out his database and study how well teams did if their starting point guard returned.
When I called Dr. Bob on Thursday afternoon he was in the midst of what, in my house, we call "The Great Purge." Every situation, every stat, every model he uses was getting the once over and, in a cold and calculating way, he was deciding if it should stay or go. "I am looking at angles and ways to tweak a little bit," said Dr. Bob. "Or sometimes I try to strip them down. And occasionally I just start from scratch."
Imagine if architects rebuilt their buildings every fall. In some instances, that is essentially what Dr. Bob is doing. He subscribes to two different handicapping philosophies to make his picks. One is situational. Using the old Buckeye Database, which I've written about many times before, he punches several parameters into a computer program, which spits out the past results for teams in similar situations. For example, he'll ask how an underdog does on the road after two consecutive wins against the spread (that is called a situation or angle). The database has betting specific info -- past spreads and results -- for college and pro football dating back to 1980 (it also costs $10,000-$15,000). Over the years, he has collected hundreds of these situations, all of which need to be tested and recalibrated during the offseason.
"I did a research project years ago about how predictive these trends are," he says. "If I had a trend that was 100 wins and 40 losses that is 71 percent. It doesn't mean it will win 71 percent of the time the next game. The chances it will actually cover the next time out is at about 54 percent."
It will take Dr. Bob about two weeks working, as he says, "a pretty decent amount of hours everyday," to cull through all of his situations and determine which ones are worthy of keeping.
Then it's onto Step 2 of the purge.
About 11 years ago, Dr. Bob switched from relying solely on angles to incorporating math-based models into his handicapping. Made sense -- the guy studied stats at Cal.
<!-- INLINE QUOTE-BOX MODULE -->
So Bob set about to study every relevant stat in a football game and analyze how that impacts the outcome. It's not just about a team's total passing yards gained or number of turnovers committed. He wants to know how teams that gained five yards per carry fared against teams that allowed five yards per carry. Those stats get poured into a spreadsheet, from which Dr. Bob builds predictive models of how the game will play out. "The situational analysis will tell you when a team is expected to play better or worse than anticipated based on a historical trend," he says. "The math model tells me what is expected in that specific game."
Once the season ends, Bob re-examines all those tiny pieces of data populating his Excel, comparing what his predictive models suggested to what actually happened. And that's when the nitty-gritty examination begins. Because it's only when he goes box-by-box that he can truly tell which stats are working and which ones are making his system sputter. "If I have the predicted box score of every game and then I have actual stats for that game, I can look at every component and see how well I have predicted each component of the box score."
If, for example, he determines he consistently miscalculated the value of turnovers or penalties, those elements get re-examined.
Next comes the mishmashing of data. This is when he surveys all of his situational stuff and all of his mathematical stuff in one spreadsheet. Ultimately, he gets a read on which of those two handicapping philosophies is providing better results. He will go over that for a week, creating a set of rules to follow during the course of the season about when it's smarter to use the situational results or when it's smarter to use the math, and then he ascribes point values to various scenarios.
But wait, that's not all, folks. That was just June and July. He hasn't even gotten to the actual teams yet. Beginning Aug. 1, Dr. Bob will begin an analysis of every football-paying school in the FBS. "I break down every returning starter stat, from what percentage of tackles are back to what percent of receptions are back. I start with Air Force and work all the way to Wyoming."
Sometime this fall, he will look deep inside that spreadsheet, at the numbers pulled from thousands of data points after many hours of labor. And he may -- I stress may -- find a handful of games in which his numbers highlight an advantage over the spread the bookmakers have posted. And, after all that, he'd still be thrilled to consistently win 54 percent of his games. Even still, he's counting on the rest of us who have been chilling all summer to help him along.
"The way the public reacts to short-term variance, which is media-driven most of the time," Dr. Bob says, "enables me to do what I do."
And live the way he lives.
Chad Millman
ESPN Insider
Through the picture windows in his pad atop Russian Hill, Dr. Bob can see nearly all of San Francisco. To the left he can see Alcatraz, former home to the most misbegotten of gamblers. Then he can swing his head around for a look at the hills of the East Bay and Coit Tower rising from Telegraph Hill and the Bay Bridge, which disappears amidst the industrial cranes that dot the landscape. Finally, he settles on downtown, rising in the distance like a set of Legos. It's the kind of panorama they print on postcards.
And it's a good thing he enjoys the view -- because he'll be staring at it nonstop for the next eight months.
You want to know what makes wiseguys wiseguys? How, like Dr. Bob, they end up profiled in the Wall Street Journal or E:60 or in this column because of the way they move lines the second they post their picks? Why they can afford fancy apartments with breathtaking vistas in expensive cities and trips to Turks and Caicos?
<OFFER>It's because all summer long, while the rest of us are BBQing or golfing and knocking off work a few hours early, they will stare at spreadsheets, tweaking stats and formulas like the busiest disciples of Einstein, looking for the model that offers the slightest advantage. They do not just flip through previews a couple weeks before the season. They do not turn on their computers as Week 1 of the college or pro football season approaches and ask for an answer. I remember talking with Sal Selvaggio of madduxsports a few years ago. It was summer. His wife was about to give birth to a daughter. And he was combing the Internet for Blue Ribbon college basketball previews that were a decade old so he could fill out his database and study how well teams did if their starting point guard returned.
When I called Dr. Bob on Thursday afternoon he was in the midst of what, in my house, we call "The Great Purge." Every situation, every stat, every model he uses was getting the once over and, in a cold and calculating way, he was deciding if it should stay or go. "I am looking at angles and ways to tweak a little bit," said Dr. Bob. "Or sometimes I try to strip them down. And occasionally I just start from scratch."
Imagine if architects rebuilt their buildings every fall. In some instances, that is essentially what Dr. Bob is doing. He subscribes to two different handicapping philosophies to make his picks. One is situational. Using the old Buckeye Database, which I've written about many times before, he punches several parameters into a computer program, which spits out the past results for teams in similar situations. For example, he'll ask how an underdog does on the road after two consecutive wins against the spread (that is called a situation or angle). The database has betting specific info -- past spreads and results -- for college and pro football dating back to 1980 (it also costs $10,000-$15,000). Over the years, he has collected hundreds of these situations, all of which need to be tested and recalibrated during the offseason.
"I did a research project years ago about how predictive these trends are," he says. "If I had a trend that was 100 wins and 40 losses that is 71 percent. It doesn't mean it will win 71 percent of the time the next game. The chances it will actually cover the next time out is at about 54 percent."
It will take Dr. Bob about two weeks working, as he says, "a pretty decent amount of hours everyday," to cull through all of his situations and determine which ones are worthy of keeping.
Then it's onto Step 2 of the purge.
About 11 years ago, Dr. Bob switched from relying solely on angles to incorporating math-based models into his handicapping. Made sense -- the guy studied stats at Cal.
<!-- INLINE QUOTE-BOX MODULE -->
“ The problem with trends only are the natural ups and downs teams go through. Also, the oddsmakers may adjust for a trend. For example, 10 years ago in the NFL betting on crappy teams as big dogs was a good bet. Now, a team that was a 12.5-point dog a decade ago is only a 10-point dog. That is why in the NFL situational stuff hasn't worked as well because there are more people aware of these patterns and adjustments are being made.
” <CITE>-- Dr. Bob</CITE>
<!-- END INLINE QUOTE-BOX MODULE -->"The problem with trends only are the natural ups and downs teams go through," he says. " Also, the oddsmakers may adjust for a trend. For example, 10 years ago in the NFL betting on crappy teams as big dogs was a good bet. Now, a team that was a 12.5-point dog a decade ago is only a 10-point dog. That is why in the NFL situational stuff hasn't worked as well because there are more people aware of these patterns and adjustments are being made."” <CITE>-- Dr. Bob</CITE>
So Bob set about to study every relevant stat in a football game and analyze how that impacts the outcome. It's not just about a team's total passing yards gained or number of turnovers committed. He wants to know how teams that gained five yards per carry fared against teams that allowed five yards per carry. Those stats get poured into a spreadsheet, from which Dr. Bob builds predictive models of how the game will play out. "The situational analysis will tell you when a team is expected to play better or worse than anticipated based on a historical trend," he says. "The math model tells me what is expected in that specific game."
Once the season ends, Bob re-examines all those tiny pieces of data populating his Excel, comparing what his predictive models suggested to what actually happened. And that's when the nitty-gritty examination begins. Because it's only when he goes box-by-box that he can truly tell which stats are working and which ones are making his system sputter. "If I have the predicted box score of every game and then I have actual stats for that game, I can look at every component and see how well I have predicted each component of the box score."
If, for example, he determines he consistently miscalculated the value of turnovers or penalties, those elements get re-examined.
Next comes the mishmashing of data. This is when he surveys all of his situational stuff and all of his mathematical stuff in one spreadsheet. Ultimately, he gets a read on which of those two handicapping philosophies is providing better results. He will go over that for a week, creating a set of rules to follow during the course of the season about when it's smarter to use the situational results or when it's smarter to use the math, and then he ascribes point values to various scenarios.
But wait, that's not all, folks. That was just June and July. He hasn't even gotten to the actual teams yet. Beginning Aug. 1, Dr. Bob will begin an analysis of every football-paying school in the FBS. "I break down every returning starter stat, from what percentage of tackles are back to what percent of receptions are back. I start with Air Force and work all the way to Wyoming."
Sometime this fall, he will look deep inside that spreadsheet, at the numbers pulled from thousands of data points after many hours of labor. And he may -- I stress may -- find a handful of games in which his numbers highlight an advantage over the spread the bookmakers have posted. And, after all that, he'd still be thrilled to consistently win 54 percent of his games. Even still, he's counting on the rest of us who have been chilling all summer to help him along.
"The way the public reacts to short-term variance, which is media-driven most of the time," Dr. Bob says, "enables me to do what I do."
And live the way he lives.
