To be perfectly honest, this is likely something that should have been addressed long before now but a recent smattering of questions has smacked me around a bit to dive in to the “what/why/how” of the advanced stats that I run. For me, it started somewhere with my self-diagnosed case of OCD but I’m far from the first to do a lot of this stuff and I’m far from the best. I just try to keep things to, “what makes sense” and then tailor things to Iowa State as best as possible.
If I were to ask you who had the best scoring season in the history of Iowa State basketball what would you tell me? Would you offer Marcus Fizer and cite his 844 points scored in the 1999-00 season? Or maybe you would go with Hercle Ivy and his 1974-75 campaign where he scored just 737 points but holds the season record in points scored per game with 28.3. Maybe if you wanted to try and get really freaky you would say Zaid Abdul-Aziz (Don Smith) and his 1967-68 season was the best because he accounted for an ISU record 34.8% of the team’s scoring in that season despite scoring just 604 points and averaging 24.2 points per game. (Note: Fizer’s 1999-00 season he scored 29.1% of the team’s points while Ivy accounted for 33.0% of ISU’s scoring in the 1974-75 season.)
For that scenario, remove the other infinite variables that can’t be accounted for like depth of each player’s own team, difficulty of the schedule, style of play, and tempo. What we’re working toward is putting statistics in to the proper context to make them more relevant. Comparing a players point totals doesn’t make sense when there is a difference in number of games played. Comparing their points per game average doesn’t make sense when they played a different number of minutes in different styles. Accounting for many variables is still not possible but by bringing things to a per possession level and calculating tempo free stats (read: tempo is accounted for), we can get as close as possible to comparing apples to apples situations.
Tempo is the key one. In today’s world we can easily arrive at the pace of each team’s games with a quick and easy formula that is quite accurate. But it really is just breaking the opportunities down to score one extra level past stats on a per game basis.
This principle is something that first interested me through basketball but I have taken the same theory and applied it to NCAA football stats. Breaking down the best offenses in college football by points per game doesn’t make a lot of sense when one team has the ball for 12 possessions per game and another for 15. What we’re doing by looking at these numbers is getting a more accurate feel for which team’s score, or whatever else, on an efficiency level—which is far more telling.
There are gobs and gobs of advanced statistics we can calculate for basketball teams and especially players but I want to talk about a few that I think are the most worthwhile and why.
Tempo /Possessions per Game
Those terms get thrown around interchangeably but they are the same thing. For season long averages possessions per game is actually shown as possessions / 40 minutes to account for the extra possessions in overtime games. Calculating possessions from box scores is a somewhat old practice but can be easily done in a pretty accurate way by using this formula:
FGA + TO – OREB x (.475 x FTA)
So, what is that all about? If you think through it logically you realize that the formula simply accounts for the ways that a possession ends in a basketball game. A possession ends when you shoot the ball unless you grab your own rebound and a possession ends when you commit a turnover. The trick comes with figuring the number of possessions that end with attempted free throws due to the variety of free throws attempted in any given trip to the line. The .475 is essentially a fixed multiplier that was found to be most accurate based on historical data, it is essentially a coefficient.
That is how a guy like me can come up with the number of possessions per game simply from a box score. One key to remember is that in a real game one team can never have more than one more possession than their opponent, because of that, offensive rebounds are considered a continuation of a possession.
If you understand all of the jibber jabber above you may start realizing areas that are most affected by stats on a per game basis as opposed to a per possession basis. I think one of the best and most straight forward of those is turnovers. Simply comparing turnovers on a per game basis doesn’t make a lot of sense when you can easily figure the percentage of possessions that are used up with turnovers to make for a more realistic comparison. The formula is simple: Total Team Turnovers / Total Team Possessions
Effective Field Goal Percentage
Everyone knows that three point baskets are worth more than two point baskets. If you consider that for long enough you probably realized that a three point shot is worth exactly 50% more than a two point shot. Why is that important? Because looking at field goal percentages without the context of the value of the shots made doesn’t truly evaluate the offense. This is the root of the more known thought of shooting 33% of behind the arc is equal to that of shooting 50% inside the arc.
eFG% is calculated by: ((3FGM*1.5)+(2FGM))/ (Total FGA)
Practically that formula is doing just what I discussed. Since the value of a made three point shot is an extra 50% over a two point shot you multiply the made threes by 1.5 (or 150%) and add the made two point shots before dividing by the total attempted field goals. Make sense?
Rebounding margin may be too engrained in our basketball culture to be able to really get away from its overuse. For a number that is so lacking in context it gets treated almost as importantly as the final score in games. I think that is partly due to the fact that it is so simple. But when rebounding margin is discussed do the total field goal attempts ever get mentioned? Or even just comparing offensive rebounds to offensive rebounds?
By using rebounding percentages separated out by offensive and defensive opportunities you can separate out which teams truly are the better rebounding teams on either offense or defense. Calculating a team’s offensive rebounding percentage is pretty easy a well:
Team Offensive Rebounds / (Team Offensive Rebounds + Opponent Defensive Rebounds)
Obviously the next step is realizing that the opponent’s defensive rebounding percentage can be found by changing the numerator to “opponent defensive rebounds” or by subtracting that offensive rebounding percentage initially found from 100%. Follow?
When you use the percentages to analyze rebounding you eliminate other factors that normally would skew the data. One team simply taking more shots than the other due to turnover discrepancies or one team just shooting a high percentage leaving their opponent lacking in rebounds because there aren’t as many defensive rebounds to grab.
That is the brief rundown of the advanced stats that I run that I find some of the best value in and some reasoning as to why I do it. It is definitely just the tip of the iceberg. If you have more interest you have probably already heard of Ken Pomeroy and perhaps check out his site already but I would also recommend a book by Dean Oliver (no, not the Hawkeye from Mason City) called “Basketball on Paper”.
This covers three (effective field goal percentage, offensive rebounding percentage, and turnover percentage) of what is considered the “Four Factors” that are the major statistical keys in how a basketball game turns out. I only left out free throw rate that is a ratio showing how often a team gets to the free throw line.
If you have comments or questions are big ideas on any of this stuff I would love to hear them. I heard Fred Hoiberg reference effective field goal percentage on a podcast with Andy Katz recently and it really perked up my day. I’m not just a “stat guy” in the sense that I like stats and numbers or even just evidence in general. I am a stat guy in the sense that I like to take numbers and make them as applicable as possible. Hopefully I help you with providing some context to your analysis.