The lifespan of an edge is fragile, like holding an Under 8.5 ticket in a 6-2 game that Kyle Finnegan is trying to close out.
Some of them can be measured in just months. Some, maybe mere days. But sometimes you get hold of one that can carry you for years before the world catches on. That’s the golden ticket, and Captain Jack Andrews used it to stroll through the Wonka factory when he picked up on the effect of barometric pressure on MLB games.
In the early 2000s when he was starting out, Andrews saw that books would hang a prop on total home runs in a game. The line was always 2 or 2.5, whether that game was being played in Coors Field or Oakland Coliseum.
For anyone paying attention to complex and subtle handicapping variables like “Is the wind blowing out at Wrigley today?” it was a huge lift to their bottom line.
Just that simple observation led Andrews to check weather reports before ballgames. At the time, weather wasn’t necessarily baked into the lines. Weather-based betting angles were ripe to exploit.
The Rabbit Hole
Wind begat temperature and temperature begat humidity, until, in an effort to really get under Al Roker’s hood, Andrews picked up on something a friend had mentioned about barometric pressure.
He picked up a copy of The Physics of Baseball, written by Yale physicist Robert K. Adair at the behest of commissioner Bart Giamatti.
The book touched on how weather, altitude and barometric pressure affected the flight of the ball.
“I was unaware of what barometric pressure even represented,” Andrews said. “I started to dig into this and around that time I discovered Weather Underground. You could get granular historical data, down to a weather station, which was sometimes right at the ballpark.”
Around 2009, Andrews put together a crude web scraper (so crude it was built in Excel). He gathered data on temperature, humidity, pressure, and the game lines. It showed humidity didn’t correlate much to lines, but pressure did.
Mercury Falling, Altitude Rising
Charlie Hayes was a 23-year-old September call-up for the Giants in 1988. In his first five years in the bigs he bounced around three different teams. No one’s idea of a future Hall of Famer, the Rockies still liked him enough to take him with the third overall pick in the 1992 expansion draft.
In his first year at Mile High Stadium (Coors Field opened in 1995), Hayes led the National League in doubles with 43, and launched a career-high 25 home runs, good for 12th in the NL. (Just behind Eddie Murray and tied with Hard Hittin’ Mark Whiten, whose 12 RBI, four-in-one-game tater barrage came that year against the Reds.)
Hayes wasn’t the only newly minted Rockie to turn on the power. Dante Bichette set a new career high with 21, and 32-year-old Andres Galarraga belted more than 20 for the first time since 1989.
While dinger totals at Mile High might not have reached the homer heights of pre-humidor Coors (149 of the Rockies’ 228 bombs in 1996 came at home) it was clear early on that altitude was going to be a factor in baseball.
The reason is that air is more dense at sea level, and less dense the higher you climb. The ball travels farther through thinner air.
Barometric pressure is directly proportional to air density. The more pressure there is, the more dense air is. The less pressure, the thinner the air, and the more it would play like high altitude games. The trick was to figure out how to normalize the effect across pressures and across parks.
Under (Differing) Pressure
Globally, the average barometric pressure is 29.92 inches of mercury. But what’s average in Seattle isn’t necessarily what’s average in Arizona. Andrews needed a way to compare what was a relatively low pressure for parks near sea level versus ones higher up, and that all normalized across all of MLB.
The answer to the problem was more Charles Lindbergh than Charlie Hough.
“In the end, just stumbling around various Google searches, I came across this air density formula that pilots use when they want to right-size their altimeter before takeoff so that they know what it should be in, say, Denver compared to Chicago,” Andrews said. “In looking at this, I’m like, ‘Wow, this is exactly what I need because really what I’m playing when I’m playing barometric pressure is I’m playing air density.’”
After landing on the formula pilots use to adjust for air density, Andrews discovered that it would tell him the equivalent altitude when taking into account temperature, barometric pressure, and the starting altitude of a given stadium. From The Physics of Baseball he knew that every thousand feet of elevation, the ball traveled, on average, seven feet further in flight.
Now his model could tell him on a low-pressure day at Kauffman Stadium, its normal 850 feet of elevation might play closer to a game at 2,000 feet.
Andrews got so into the weeds on the betting angle he stationed a friend inside Tropicana Field in an effort to quantify dome effects. The friend took air density readings from his iPhone’s aerometer to compare it to the weather report.
“You think that’s a cheat code,” he said. “Imagine knowing that a game is being played in Denver altitude-wise and nobody else knows that. The trouble is other people had independently found that information out.”
Shifting Weather Patterns
It didn’t take long for Andrews to learn that other sharps were hitting that betting angle, too. Some of them were very big players, including one member of Billy Walters’ Computer Group.
By 2012 after he had refined the model through around 2014, Andrews was hitting around 60 percent with baseball totals. Exploiting an angle, though, can be the thing that helps hasten its demise.
“There was a lot of learning in terms of how do I apply it?” he said. “I was working with some people to get money down elsewhere for me on these. That might’ve eroded some of my edge because when you do that with somebody, chances are they’re copying it. Maybe selling it to somebody else or they’re betting it elsewhere for themselves. You’re kind of using a multiplier effect on your own betting without getting the benefit of increased profits.”
By 2016, there were signs that the market was absorbing the information.
Back then, baseball lines would start moving around 9 a.m. Eastern time, and they’d jostle again at 11 a.m. when the West Coast chimed in. He could run his model in the morning and have plenty of time to get down.
But before long, that 9 a.m. window was moving earlier. Games were starting to move overnight, even when limits were lower.
“It’s almost like a war of attrition when too many people find out about an angle,” Andrews said. “It’s who’s going to bet it earliest and then that moves the line and the markets replicate around the world. It just keeps getting earlier and earlier, and then you get people betting into openers and right-sizing the market almost immediately. Then the angle’s gone.”
Death of an Angle
From 2016-19, Andrews kept pushing on the angle, but he was breakeven or worse playing barometric pressure.
He tried to apply the concept to other sports, but there wasn’t a strong enough correlation in football or basketball to make it worthwhile.
Andrews is content with the run he made using barometric pressure (other than not bailing out in 2016.)
“This was as much as I could handle as my skill level,” he said. “Maybe if I was working with somebody like Rufus (Peabody), we could have taken it to another level. But this was about the peak of my capacity.”