With a new college basketball season comes new players, coaches and styles. When can we trust the numbers? How do we use analytics in wagering? Do the architects of these algorithms trust their numbers when placing a wager?
Analytics Q&A with Erik Haslam
I decided to ask Erik Haslam,the architect of Haslametrics, these questions.
MH:How do you compute preseason rankings and why?
EH: I don’t usually put a ton of stock in preseason baselines, but I began using them last season more out of necessity (due to the pandemic) than anything else. We knew last year was going to be wonky with some teams not playing until well after the official start of the season, so the baselines allowed me to maintain realistic ratings and forecasts for all 357 teams throughout the whole year.
Once I had a system in place to create those baselines, I just decided to keep using them — for this season and beyond. The preseason baselines are created from a set of regression equations that account for three years of program prestige, last season’s ratings, returning production, projected production of incoming transfers, recruiting rankings, and head coaching changes.
MH: At what point in the season are the numbers trustworthy?
EH: To a certain degree, they’re fairly trustworthy from the start, though there is more subjectivity and educated guesswork early on in the year. As more actual results roll in, we get a better handle on how each team unit truly performs — it’s more based on real performances instead of model speculation. As a result, the forecasts should become more accurate as the season progresses.
MH:Have you ever vehemently disagreed with what your numbers produced?
EH: I wouldn’t say vehemently, but there are occasions where my eyes tell me something different from the analytics. I may produce a college basketball analytics website, but I still hold the opinion that anyone who goes all-in on the analytics is a fool.
Analytics are like evidence at a crime scene; they don’t necessarily translate to an open-and-shut case. You still need your eyes, ears, and other senses. The analytics can have blind spots, so it’s important to account for those when making projections.
MH: How do you deal with random criticism from anonymous bettors?
EH: To be honest, I don’t receive a ton of criticism. But then again, most experienced bettors understand that no system is perfect. You need to find a system that works for you.
That’s not to say there’s never been any criticism. On Twitter in the year 2021, there are always a few people here and there willing to yell at you for being wrong. You take the good with the bad. Thankfully, there’s more of the former than the latter.
MH: You’re not much of a bettor, but what are some metrics you put some faith in when placing a wager?
EH: You know, that’s a tough question. I don’t know if there’s one specific metric that stands out. I think the Momentum and Consistency metrics can be a big help, especially later in the season.
Honestly, I prefer to follow some “eye test” experts out there, and when their viewpoint closely mirrors those of the analytics, I’m more likely to make a wager on that particular contest.
MH: Do you trust your metrics implicitly or do you find yourself torn when placing a wager?
EH: I’m frequently torn. Very rarely am I extraordinarily confident with any wager I place. I know that humans are unpredictable creatures — especially humans in their late teens and early twenties — so performances can greatly vary from one day to the next.
Therefore, I won’t typically bet that much, both in dollar value and in frequency. But it does certainly make things more interesting if you can be responsible with it.
MH: How much is too much when talking about analytics?
EH: 7. As I like to say, analytics can be like digging a grave. Six inches is not enough, but 600 feet is just plain overkill.
I prefer to determine a reasonable estimate and not worry about chasing my tail trying to chase down every independent variable in order to determine the perfect equation. Perfection itself is unattainable.