Basketball Stats for Dummies
I am relatively new to the advanced stats in college basketball. I'm trying to digest as much information as possible but trying to keep up all of the many sources feel like drinking from a fire hose. Are there some key books, blogs, Twitter follows, etc that would give me a really good baseline before I dive into some of the more advanced stuff?
Thanks everyone.
February 11th, 2020 at 3:31 PM ^
Lesson 1. When you lose a basketball game badly, you did not play good defense, even if you only gave up ".93 points per possession" during the game.
Lesson 2. Breaking down shooting percentages to what a player shot "on twos" is really dumb. There are probably 500 different kinds and attempts of two point shots and a player is not good or bad at shooting "twos." That is irritating.
Lesson 3. If you are relying on an advanced metric that includes "expected performance" in its calculation for the first 2/3 of the season, you need to find a new metric.
February 11th, 2020 at 3:54 PM ^
Ignore Ijohnb's first 2 points. Those points indicate he does not understand what those metrics are measuring.
1) You can get blown out of the gym playing solid defense if you can't score. It happens.
2) Players absolutely can be bad or good at shooting two's. More importantly, players/teams may be less or more likely to attempt low-percentage shots if you deny the high percentage shots and dare them to take long twos. If a player shoots a high volume of two's at 35%, but hits 38% of their threes, that absolutely changes how you should defend them. If that guy happens to be on your team, he shouldn't be allowed inside the arc unless he has a dunk or uncontested layup.
February 11th, 2020 at 4:07 PM ^
Carry on.
Advanced metrics are irritating. Most often discussion regarding a game now sounds like two people observing a science experiment. It seems they are mostly utilized to convince fans of losing teams why they did not play as bad as their eyeballs indicated.
February 11th, 2020 at 4:21 PM ^
So you prefer the good old days when basketball analysis was mostly total bullshit?
February 11th, 2020 at 4:25 PM ^
Yes that was better.
People who list off 15 metrics when they are discussing a basketball don't know more, they just sound smarter.
February 11th, 2020 at 4:36 PM ^
I think your first lesson above demonstrates perfectly that people who shun advanced metrics aren't necessarily very good at basketball analysis.
February 11th, 2020 at 4:48 PM ^
People who dismiss the value of additional knowledge don't just sound stupider, they actually know less as well.
Advanced stats are popular because they tell you more about the future than standard stats do.
February 11th, 2020 at 4:59 PM ^
Slow done pleaze i dont read to good.
February 11th, 2020 at 5:38 PM ^
You are so correct. This game is so profoundly nuanced. Stats have value but they are just as profoundly overrated. Stats are simply a recording of symptoms-they tell you absolutely nothing about the root causes of how and why they come into being.
February 11th, 2020 at 10:58 PM ^
Username doesnt check out
February 11th, 2020 at 11:24 PM ^
This is terrible, terrible information. Some stats are basically records, true. Like points. Some stats are very predictive, especially tempo-free stats. Or style stats, such as the percentage of your shots that are twos vs. threes, or possession length. Good offensive or defensive teams will have - SURPRISE - good offensive or defensive efficiency stats.
February 11th, 2020 at 8:01 PM ^
Except nearly every coaching staff utilizes advanced metrics to some degree nowadays.
February 11th, 2020 at 8:51 PM ^
Lol, you sound like an old baseball scout arguing against advanced metrics--like those dudes in "Moneyball" insisting some kid's swing looks "good." Your first two positions are ludicrous, and frankly, not supported by data--which is more reliable than the "eye test." Advanced stats have literally changed the game, and have guided some of the notable shifts in the NBA and NCAA (i.e. less mid-range jumpers, more 3's, more emphasis on guarding 3's, etc)
February 11th, 2020 at 8:32 PM ^
I think all he wanted was to be directed to the kenpom.com site. Lots of fun data there, some of it even understandable to us who didn’t take math after that long division monkeyshine.
February 11th, 2020 at 3:33 PM ^
I'm sure Wolverine Devotee has a few binders of stats he could loan you.
February 11th, 2020 at 3:37 PM ^
Get a subscription to UMHoops - it is inexpensive and the content, especially with the stats and film break down, is amazing. Between Dylan's coverage there and what you get from Ace, Brian, and Matt here that's a good start. Those guys like a variety of advanced stat websites like KenPom, Synergy, HoopLens, etc.
February 11th, 2020 at 3:39 PM ^
Super For Dummies Basketball Stats:
1. Shoot only 3 point attempts and dunks
2. Prevent other team from shooting 3 point attemps and dunks. Make them shoot 2 pointers out of the lane.
....that's kind of the baseline theory behind most modern college basketball game theory I think.
February 11th, 2020 at 4:36 PM ^
I was watching one of our games earlier this season and it felt like we only shot threes and attempted layups. No pull up jumpers, or even dunks. Hell, I don't think Simpson even had his sky hook going. Is he doing that this year? The game was exciting, but it didn't seem as exciting as back in the day. Now get off my lawn.
February 11th, 2020 at 4:57 PM ^
That's a super simple baseline. I think the driver behind both those points is really expected points per attempt, which explains why its generally better to avoid long two pointers and take a three pointer instead - a decent shooter will likely hit both at approximately the same rate (probability), but the three is worth more, resulting in a higher expected value.
I'm no coach, but my understanding is that's how a coach with good understanding of analytics (like Juwan) would approach constructing a gameplan. In the first MSU game, UM primarily stuck with its drop coverage on ball screens, which ended up being relatively unsuccessful because Winston was very efficient (high expected point value per possession) on the relatively uncontested twos UM gave him. In contrast, in the second MSU game, UM pressured Winston to give up the ball out of ball screens, often time to worse shooters. UM was content to give those shooters 3s (and did a good job recovering to at least contest them) because the expected point value of a Tillman 3 pointer was less than that of an uncontested mid-range 2 from Winston.
February 11th, 2020 at 4:58 PM ^
To be clear, I didn't disagree with your baseline either - just tried to expand on it a little bit.
February 11th, 2020 at 6:24 PM ^
- determine the expected points from any possible attempt
- on offense try to take the shots with the highest values above
- on defense try to induce the shots with the lowest values above
#1 being generally 3s and dunks, tho brooks is like 20% in the lane and some people are 20% from 3. The scouting report should tell you these things so you know who to let shoot 3's, who cant go left etc. etc.. If you take better shots than your opponent you will usually win.
February 11th, 2020 at 4:29 PM ^
Shot exit speed, direction and trajectory are the critical stats I keep an eye on.
February 11th, 2020 at 4:31 PM ^
You might need a subscription to do this, but if you:
- go to kenpom.com
- pick something from the Stats menu
- click on the yellow question mark icon
you'll get an explanation of the stats that KenPom uses. KenPom is a must as a baseline level of knowledge for advanced stats.
February 11th, 2020 at 5:56 PM ^
The overarching principle is that rate stats are more useful than raw stats. Per possession, per opportunity (for example, use offensive rebounding % instead of raw rebound margin), take into account that 3 is greater than 2. You can get into crazier stuff, but efficiency and the "four factors" get you a long way there as far as team stats go.
February 11th, 2020 at 11:31 PM ^
Yeah, Dean Oliver invented the four factors. It is pretty amazing to me how much what was originally a very nerdy concept and formula has penetrated popular understanding.
February 11th, 2020 at 8:21 PM ^
Basketball on Paper is the bible for basketball analytics departments. It's by Dean Oliver. I read it while studying Stat in a Dept. of Math postgraduate.
The book actually got him hired by the Denver Nuggets. But this is the one that lays the foundation for a lot of what is reported today as 'advanced stats' and specifically how they are calculated. By the end of it you would know how to actually calculate them in an Excel file, for example.
(It is very math heavy.)
Aside from that, a deceptively simple stat that is always meaningful is offensive rebounds. But that is just an IMO.
February 11th, 2020 at 10:12 PM ^
fwiw, I think you mean rebound % ? Too much noise in raw #'s of Oreb (i.e. lots of shots, a poor shooting night, ugh).
February 11th, 2020 at 11:25 PM ^
Well, yes and no. My point is that offensive rebound is one of those stats where you can take one look at it and very easily, for example, ballpark OReb% to within an accurate range of what is happening.
At the end of the day, if you are an assistant coach on the bench and you look for 3 secs during a timeout and see you have 10 OReb and your opponents have 5, that is something you can feel good about... whereas defensive rebounds can be wildly misleading for the reasons you specified.
I'm just saying that's a simple stat that extrapolates very well in context.
February 11th, 2020 at 8:25 PM ^
To piggyback off of my last comment (about Basketball on Paper, by Oliver), what it does that is primarily most beneficial is it teaches you how to determine from a simple box score how many possessions each team had during the game. What that allows you to do - is narrow down things like per possession analysis and more specifically and perhaps useful, tempo analysis.
You know? And it looks into things like what constitutes a possession - is an offensive rebound a new possession or a continuation of the last one? Is a five second violation on an inbound pass a possession and how do we interpret that from the box score?
But it does so less superficially, meaning rather than ask these questions for the purpose of debate it asks these questions for the sole purpose of mathematically determining the most accurate answer for the purpose of further analysis.
February 11th, 2020 at 10:59 PM ^
Serious question:
I'm following my local D-2 team, Minnesota-Duluth, quite closely this year. Does anyone know of a spreadsheet that allows me to plug in counting stats for a player that will produce some useful advanced stats? Thinking specifically about stuff like O-Rating.
February 11th, 2020 at 11:30 PM ^
I did it once in Excel. The book I mentioned above (Basketball on Paper) teaches you how to do exactly that; BUT, it is a very complicated Excel formula. If you read that book and understood Excel functions on an intermediate basis, (re: no coding experience necessary) you would know how to do it by the end of it, but again, one typo throws the whole thing off and formulas get crazy long.
I was on adderall at the time and I would never have done so was I not. Point blank.
But, in terms of a short answer like a website or something for D2? I honestly don't.
But, it is entirely possible to do so with nothing more than box scores from their games.
February 12th, 2020 at 12:35 AM ^
The formulas for Ortg are here: https://www.basketball-reference.com/about/ratings.html
Pretty complicated because the effect of assists and orebs gets a bit wacky. Also need to use possessions while the player was on the floor which isn't available from box scores are normal data sets. That goes for any per possession metric too.
Easy to calculate eFG which is pretty just (2pt makes*2 + 3pt makes*3) / FGA*2 but that's about all you can do from counting stats (you can also do TS% by adding a FT component).
February 12th, 2020 at 5:08 AM ^
You are awesome.
Good day. I also a big fan of basketball, I follow college championship and of course NBA, it is very interesting hobby. As I know there are many different sites, online services and applications where you can check necessary information like scores and odds, predictions and betting tips. Just search in Google.