Availability Bias
I went to a wedding on a fall day, a few years back, and as I often do at such events, I monitored the football scores closely on my phone. Unlike the situation I find myself in now, the system I had created was having another good start and another good day, but of particular interest to me was that an underdog I had taken, plus a whole lot of points, was actually threatening to win the game outright. For the life of me, I can’t recall or figure out which game this was, but it was definitely a big name program from a power conference against another FBS team, but of little renown and definitely not from another Power 5 school. It was a small, outdoor wedding with more charm than finance, the kind where friends actually have a shot of spending good times with the bride and groom, and what I remember vividly was the moment that my friend, the groom, while taking a minute to look at the scores I already knew everything about, snickered about this particular game, or most importantly the power team that was struggling against their measly opponent. I’m sure I said something about my having this right, in fact I remember that I did, but who cares, right? He didn’t seem to.
Where my head went next though is what’s important here. What if my friend now assumed that the big name program was actually far worse than they really were because of his perception of them from this one game? After all, he was obviously not looking at the opponent with any objective quality other than the conference they played in, and perhaps a lack of press coverage prior to the game was contributing to his judgment. For most people, it hardly matters, but for any person putting up money on games, this could be an unidentified crisis moment in a day where future predictions are made without the proper acknowledgement that the underdog team was actually pretty good in the first place. This is called availability bias, where conclusions are drawn from the first memory that pops to mind. And in sports gambling, it can be a killer.
I think that availability bias is most common when a person, like most of us, follows one team, or maybe one conference, a whole lot more than they follow everybody else. It doesn’t necessarily mean a person is totally ignorant of the sports world in general, but whenever an emotional investment is made, the likelihood of one game being more prominently remembered than others is always there. You see this all the time, for example, when one fills out a bracket for the basketball tournament. We played them! They have a shooter! They can beat Duke. The solution to expelling availability bias is fairly simple in concept, but a lot of work in practice. Regardless of what metric you want to use, the data within that metric has to apply to all of the teams you wish to draw conclusions from, meaning you gotta follow all of them. And then, and this part is important, when one team beats another by more or less than you expected, you have to give equal credit to each team for that outcome. So if your numbers, by whatever scale you’re using, are responding to one team beating another by 40 when your numbers suggested 10, a 30 point difference, give the winning team credit for 15 of those points and penalize the losing team for the other 15. I’m telling you all of this as a week by week narrative and will get more into metric and scale as it goes on, but remember this as I proceed, do not let your subjective interpretations affect your attempt at objective reasoning. Availability bias is a real thing and we all have it, so do don’t go meddling with your numbers just because you were surprised when they were right or wrong.
And I say all of this now, even as I’ve seen the system botch the first few weeks of the season. The system was badly off on BYU/Navy, then Navy/Tulane, and then Tulane/Southern Miss. We should be able to identify the culprit here, Navy in game 1 threw the system of kilter. Blame what you like, but in 2020 I have an untestable idea of what has happened here, Navy just wasn’t ready to play BYU and there were likely scenarios such as this across the board. So I’ve lost the early season bounce I ordinarily get and now I’m playing from behind rather than having a cushion to play with. That’s a bummer. But my instinct, as well as my experience with the system both suggest that as these teams continue to accrue data, as well as settling into being the teams they are to be in 2020, the system will adjust and still has great potential to bring in a winning season, even when picking Navy, BYU, Tulane and Southern Mississippi games. Hopefully, it starts this weekend. Here are the system’s picks:
Friday
BYU -23.5 at home vs Louisiana Tech
Saturday
Florida -18.5 at home vs South Carolina
Texas -13 at home vs TCU
Alabama -17 at home vs Texas A&M
Virginia Tech -10.5 on the road vs Duke
Air Force +7 at home vs Navy
Georgia Southern -19.5 on the road vs Louisiana Monroe
Auburn +6.5 on the road vs Georgia
North Texas -1.5 at home vs Southern Mississippi
The system would also take Oklahoma against Iowa State if the spread dropped to -5.5, but once again it seems unlikely. Hopefully we get to see all 9 games play this time around; have a great weekend!
