As teams finish the regular season and conference tourneys begin, we’ll be treated to an alphabet soup of acronyms as coaches and supporters point to the various analytic models to make their case for a tournament bid. In the third installment of a series of posts taking a look at which models can help as you try to bet basketball’s craziest season, we look at one of the oldest basketball analytics around the Four Factors.
March is here. Teams will finish their regular seasons this week, assuming they haven’t already wrapped that up and are headed to conference tourneys.
As March Madness looms on the horizon, teams are making their cases for an NCAA Tournament bid. That means we’ll be treated to an alphabet soup of acronyms as coaches, supporters, and bracketologists cite the latest analytics to make their points.
In the third installment of a series of posts looking at which models can help as you try to bet basketball’s craziest season, we look at one of the oldest basketball analytics around the Four Factors. What are they, and how can they help provide value for gamblers?
What are the Four Factors?
Dean Oliver was a player and student assistant coach for Cal Tech, where he earned an engineering degree. He later went on to get a PH.D. in statistical applications.
In the 1990s, he came up with the idea of applying the skills he used in his day job, as an engineering consultant, to sports and did research on basketball. In the early 2000s, just as the nation was learning the term Moneyball, Oliver wrote Basketball on Paper, a book that has become the bible for hardwood analytics.
One of the major themes of Oliver’s book was the Four Factors, which he thought were the keys to winning a basketball game.
Oliver put a weight of 40% on shooting, making it the most important of the four factors. It seems pretty straightforward: How many of your shot attempts go in, and how many of your opponent’s shots do you prevent from going in.
The three-point line throws in a wrinkle, however. A team’s effective field goal percentage takes into account the extra point awarded for a three.
Back in the day, a .500 shooting percentage was the benchmark for a team—make half of your shots. But with the advent of the three, that math changed. Making a .333 percentage from three was just as valuable, points-wise, as hitting .500 from two. You can make three of six two-pointers or two-of-six from three, and they produce the same result.
So effective field goal percentage gives an extra 50 percent weight to three-pointers made. It basically gives more value to a team that shoots a large percentage of their shots from three and thus has a lower straight field goal percentage.
It may not seem like a groundbreaking idea, but it has essentially led to the explosion of three-point shooting we’ve seen in the NBA in recent seasons.
In college basketball, the median effective field goal percentage is .500. The best shooting teams top .600.
Turnovers got a 25% weight from Oliver, making it next to most important. A team’s turnover rate measures what percentage of possessions end in a turnover.
The median in college is 18.6%, meaning that nearly one in five possessions ends in a turnover. The best teams are in the 12% range, while the worst are above 25%.
Rebounding gets a 20% weight, and it is the one that is surrounded by the most noise. Raw rebounding numbers can be very misleading because they depend heavily on how well a team and its opponent shoot. A team can have very low rebounding numbers in a game, simply because the shots are going in on one end.
Rebounding battles are not created equal: The team on defense has a much higher percentage of rebounding a missed shot. The defensive team gets about 28% of the rebounds, on average. So, a big rebounding mismatch might just be showing a big mismatch in the two teams’ shooting percentages.
Instead, Oliver looked at how often a team wins the battle of the boards, by comparing one team’s offensive rebounds to the other team’s defensive. For example, UNC beat NC State by a 43-25 rebounding margin in their second matchup this season. State shot just 38% for the game, while UNC shot 52%, so most of the rebound opportunities were when UNC was on defense, and the Tar Heels won the defensive boards by a 33-10 margin (UNC has 33 defensive rebounds, State had 10 offensive). At a 23% offensive rebounding rate (10 rebounds in 43 chances), State was about five percentage points below average.
It was on the other end, however, where the real damage was done. UNC had 13 offensive rebounds, NC State 15 defensive. So UNC had a 46% offensive rebounding rate, 18 points above average.
Comparing teams on offensive and defensive rebounding rates paints a clearer picture of what was going on in the paint than raw numbers of rebounds.
This gets 15% weight from Oliver, and it measures far more than just whether a team is accurate from the free-throw line. The free throw factor looks at how often a team scores from the line, which first involves getting there.
Some teams are better at drawing fouls than others. A team that drives to the basket and scores near the rim is more likely to be fouled than a team that fires away from three.
The statistic used to measure this is free throws made, as a percentage of field goals (not free throws) attempted.
The average is about 30%. In other words, a team makes three free throws for every 10 shots attempts from the field. The better teams are around 40, the worse, around 20. The spread is much higher on defense, ranging from 15 to 50%, depending on how well a team does at defending without fouling.