1. Sharps will disagree with Sharps, Squares will disagree with Squares, Analytics will disagree with Analytics
As you get into the gambling market, you will hear statements such as “the sharps say”, “the squares say”, and “the analytics say” as if there is some sort of secret club where these groups come to a consensus on matters. That is certainly not the case.
The reality is that there will be different opinions within each of these groups. A “sharp” bettor theoretically has a better understanding of gambling markets and the teams involved, and is therefore making better bets, but one sharp’s method could disagree with another’s.
I don’t think I’m a total square when I dabble in betting on other sports, but I’m certainly much “sharper” betting on the NFL.
– The Professor
Likewise, the conclusions drawn from an analytic model depend on the “training set” the model learned from, in addition to a variety of other factors, so a model that says “teams convert this 3rd down situation X percentage of the time” is really saying that “based on the data provided to the model, teams are likely to convert this 3rd down situation X percentage of the time.” If your training set is all NFL snaps from 1970 to present, you will get different numbers than if your training set is from 2000 to present, but both would be “what the analytics say.”
As for “squares”, we’re talking about casual gamblers who are placing bets without doing a lot of research. Squares are more inclined to bet on “their team”, and as every team has a fanbase, it’s easy to see where squares would disagree. It’s also important to realize this is a sliding scale; for instance, because I have a good understanding of odds, I don’t think I’m a total square when I dabble in betting on other sports, but I’m certainly much “sharper” betting on the NFL.
2. If someone says you need to be successful on a certain percentage of your bets to be profitable, they are referring to specific odds
One downside to living in the United States is that our measurement systems don’t make intuitive sense. The prime example is the metric system versus whatever it is that we’ve come up with, but in betting this shows up in the way we write odds.
Abroad, you will see betting odds that are either in percentages, or in fractions that translate intuitively into percentages. Some online sportsbooks give you the option to switch how the odds are presented, but if that’s not the case, there are two equations you will need to use to translate “plus” or “minus” odds into percentages. I’ve included these in the show notes, along with a formula you can drop into Google Sheets or Excel to get your break even percentage for a particular bet, but as you grow familiar with odds, you’ll learn some “landmark” numbers that allow you to do quick math in your head.
The most basic lesson is that +100 odds, which indicate you need to lay $100 to win $100, have a break-even percentage of 50%. Any plus odds above 100, say +125 or +200, have a break-even less than 50%, these are underdogs, and any minus odds, such as -110 or -200, have a break-even greater than 50% and are favorites.
Once you translate the odds to percentages, the important question will be whether you believe the event in question will happen more or less than the break even percentage. If you believe it will happen more often, it’s a bet you want to take.
3. There is no perfect way to grade player performance
I’m going to pick on Pro Football Focus here, because I think they do a very good job defining their grading criteria and applying those grades uniformly. However, I’ll use a writeup of a Packers Chiefs game from 2015 that saw Aaron Rodgers throw for 333 passing yards, 5 touchdowns, and no interceptions, but earned an average PFF grade of -0.8, to flesh out a point on a gap in their method.
As PFF writer Ben Stockwell noted in his writeup, “The context surrounding his (Rodgers) grade is crucial. The greatness of Rodgers’ performance last night was in the intangibles. Recognizing the blitz, drawing the defense offsides, catching the Chiefs in bad situations, and exploiting those scenarios with simple passes to open receivers. But you cannot — and we do not try to — quantify intangibles, or what comes pre-snap. Our system grades what can be graded — the execution of the play post-snap — and in that regard Rodgers did not stand out in the same way that his statistics did.”
you absolutely can quantify pre-snap intangibles
– The Professor
As an aside, you absolutely can quantify pre-snap intangibles. I could assign a -1, 0, or 1 based on whether I thought a quarterback did a negative, neutral, or positive job getting his team into the right spot. Folks could (and would) disagree with whatever method I use, but folks disagree with what PFF does grade, so the objection that you “can’t” quantify intangibles runs out of steam quickly.
While PFF has an excellent grading system, they ignore what I would argue is the most significant aspect of quarterback play. That doesn’t make either of us right or wrong, it just highlights the fact that grading football performance is an entirely subjective exercise. By the way, if anyone tells you they’re looking at something objectively, all they mean is that they’re attempting to account for their subjective bias. It isn’t possible for a human being to look at something objectively, because we all see life from our unique perspective, and as human beings design analytic models, the models aren’t objective either.
I’ll circle back to a famous quote from a British statistician, George Box, who said that “All models are wrong, but some are useful.” I believe this applies to grading systems as well as models. I’m not suggesting you shouldn’t use a grading system to help inform your bets, but I do think it’s important to understand that no system is perfect if you want to apply the information in a skillful manner.
Show Notes:
“Plus Odds”: 100/(100+odds)
“Minus Odds”: (odds * -1)/((odds* -1)+100))
=IF(I4>0,(100/(100+I4)),((I4*-1)/((I4*-1)+100)))
( I4 = odds for a particular cell, but this could be any column (letter) and row (number) designation.
Steven Clinton, also known to BeerLife Sports fans as “The Professor”, is an expert quantitative modeler and former college football researcher at his alma mater of Northwestern, where he broke down film for the Wildcats.
Steven Clinton, better known as "The Professor", is a former D-1 Quality Control Assistant (Northwestern, Toledo) who holds a B.A. in Economics and M.S. in Predictive Analytics from Northwestern University. He maintains an end-to-end NFL game projection model and is a film junkie who breaks down the tape of every NFL game.
You must log in to post a comment.