Let's Talk about Stats

Every year I try to push myself to find a new stat or stats to track during the year. During the off-season I may do a random class, listen to webinars, research new trends, connect with coaches, etc to see if there is something I could be doing better. Doing better as far as tracking and presenting the information.

Over the years, I have been pretty good about gathering the info (and I'm still finding new ways to gather) but the presenting and explaining has been a bit more difficult. Knowing and having the information is only half the battle. I'm not on the court playing the game nor am I on the court coaching the players. I have to be able to explain the information to my staff and then to the players. But sometimes the hardest part is how. How do I explain these numbers in a way that they will remember and so it will resinate with them? If it doesn't stick, I'm just talking and everything is going in one ear and out the other. I decided to push myself this year to try some new tactics to help the staff and players buy into some of the advanced numbers I like to use.

Over the next several weeks, I am going to be posting about some of the new initiatives I tried this year and some of the projects I did when looking at advanced stats.

Over the past few years, I have used my own Excel spreadsheet for post-game analysis. This included Individual Player Offense Rating (ORTG) and Defensive Rating (DRTG), Usage % (USG%), Team Offensive Rebounding % (OREB%), Opponent Offensive Rebounding % (OppOREB%), Segments (on the Men's side I would do every 4-Minutes and on the Women's side I do the first two 3-Minute segments of a quarter and then the last 4-minutes of the quarter), when shots were taken in the shot clock, Defensive Shot Charts, a subjective Defensive Points Per Possession per Player, and a few other stats. This year I added Potential Assists, (something I have deemed) our Intangibles Stats (these have a value associated with them and are then weighted by how many possessions a player plays), Team Goals that our Head Coach wants to hit, and I added an Estimated Wins Formulas for us to look at to see how many wins we would have if we had those same stats in a 14 Game Season (more on this later).

This year, my challenge was to make a Pre-Game Stats Report for opponents. Every game I would make between a 25-30 page packet of information for our coaches. I will go into more detail later about this Pre-Game Stats packet. I was able to pull a lot of information from several resources and then add in my own notes to help explain the data.

One thing I want to point after that last paragraph is that Stats are endless. We can track almost anything and everything and there will always be new ways to track and new stats to track.

Before I start to show examples, let me give everyone a very brief overview about the recourses I used to get the information for my Pre-Game and Post-Game Packets.

Pre-Game:

Just Play Sports Solutions - This company is where we do our scouting reports, share film with players, and where I gather a lot of my pre-game data. They are able to provide Team and Player Shot Charts, Lineups, Advanced Team and Player Reports, Rankings in stats, etc.

HerHoopStats.com - I am on this website almost daily just scrolling looking at data. They provide a ton of data dedicated to NCAA Women's Basketball

Synergy - I am able to get an idea of how Teams and Players score and defend. Maybe they play a lot of Zone or they switch between Man and Zone. This gives me a baseline to understanding how they play.

CBBAnalytics.com - Another website I use often. They provide a ton of metrics in a variety of ways. Assist Charts, Player by Minute charts to help get an idea of team rotation, charts to look at how teams rank in a variety of categories (these charts help to see if two stats are related or if one affects the other), On/Off Numbers, etc.

Post-Game:

My own Excel Spreadsheet - This is a spreadsheet I built a few years ago and have continued to alter and add to. After the game, I am type in the box score and it outputs ORTG, DRTG, USG%, OREB%, TOV%, Wins Estimated. Then from how I cut the game in SportsCode, I type in our Intangible Stats and our Potential Assists

HD Intelligence - A newer company that was able to provide us a post-game advanced stats box score with great metrics. This HD Box Score did a great job of breaking down information we can't really get elsewhere.

Pivot Analysis - A company that provides great depth into your team's data. Their site provides a variety ways to breakdown your team. Many filters to see the data in certain segments of games

Again, that is a very brief overview of the resources I use to make these Stats Reports.

Remember, the overall goal is to provide good and useful information that everyone can understand. If there is a stat I don't put a lot of value in or I don't have a good understanding of, I don't include it. I want to be sure I can explain everything I am presenting. Having effective communication is vital. As stated previously, I will go into more detail about my Estimated Wins Formula but to even get the formula required getting a the proper data needed to run a Regression, then looking at the coefficients and intercept, understanding the data used (to be quick was EFG%, TOV%, FTRate, OREB%), looking at the data and understanding if we increased each of those stats by 1 Standard Deviation we would increase our wins by so many. Look, there are a lot of words that I just said that half the time I need to refresh my memory. So if I need a refresher, how can I guarantee my staff will know what I am talking about. For the record, I can't guarantee. This means, I have to find a different way to explain what we are looking at.

One thing that is difficult in College Basketball is Sample Size. This includes how many games you have played. 2-3 games is really hard to get a good grasp of stats or lineups. In my opinion you really need closer to 15-20 games to get a good sample size for data. However, since most college teams only play 30 games, if you wait until game #15 your season is half over. So that means you'll have to start looking at data earlier. For me, around Games 6-8 is when I'll try to start getting an idea of the data. It can and probably will change a good amount as more games are played and the sample size grows.

I will go deeper into the Pre-Game and Post-Game Packets in future posts. Every so often I will take a deep dive into a thought during the season. This results in some random projects to see if there is anything there. A lot of the time there isn't but sometimes you'll find something worth talking to the staff about. And because I don't want to leave this post without some sort of data, here is a look at our Segments through our 1st 7 games. I started tracking segments several years ago because I wanted to see how we were playing as the game progressed (mainly by points and if we won the segment).

After 7 games, our record 5-2 and 2-1 in Conference. Without looking at the stats, I could tell you we were a better 1st half team. I started wondering why. There are a lot of things I could look at but I had to start somewhere. So I wanted to look at our Segments. Below you will see our Segments for all 7 games and then another image of our segments through our first 3 Conference Games. You will see the time frame, Points For (PF), Points Against (PA), Difference between PF-PA, UH Average scored in that segment, Opponent Average scored in that segment, the Difference in Averages, and our Record if we Won or Lost that Segment (ties are not counted). You will also see Segments separated by Wins. (I also have a Segments in Losses but is is now shown here).

This image is our Segments thru Game 7 (please ignore the OT line and the #DIV/0! as we did not play any OT games and there is no data present):

This image is our Segments thru our first 3 Conference Games (please ignore the OT line and the #DIV/0! as we did not play any OT games and there is no data present):

Here is how I saw the data:

1. Looking at the Totals Overall Chart:

   1. Overall we have a pretty good 1st Half record at 22-14. Not amazing but we have a winning segment record in all but one segment. And only one segment are we being out scored in.

   2. But in the 2nd half, we have a losing record at 17-19. And we are being outscored in 3 segments, we are tied in one segment, one segment we are only up by 0.6 Points, and the last segment we are outscoring our opponent by 2.3 points.

   3. Another point that stuck out right away.

      1. Our scoring in the 1st half was 34.7 vs 36.0 in the 2nd half. A 1.3 increase. Always good to see an increase

      2. BUT our opponents were 28.4 in the 1st half vs 39.6 in the 2nd half. A 11.2 increase. THAT is a massive shift.

   4. Not only are we being outscored by 3.6 points but we let them get 11.2 MORE points in the 2nd half. Now what? I'll take a look at the Quarters to see if anything sticks out

   5. Something does and that is the 2nd 3-Minute segment in the 4th. This is the 6:59 mark to 4:00 mark. We are being about scored by an average of 3.4 points. At this point in the season (7 games), I think that is pretty significant.

      1. But here I have to ask is it in all games or maybe our two losses are playing a big part of that. Remember at this point in the season, we are still dealing with a small sample size regardless of what we are looking at

2. So next I want to take a look at our Segments in Wins to see if there is a consistency in this thought

   1. In the 1st half we score 37.8 vs 38.8 in the 2nd - 1.0 increase

   2. We give up 28.6 in the 1st half vs 38.0 in the 2nd half - 9.4 increase & this is consistent with the Overall chart that we are allowing a too big of increase from 1st half to 2nd half.

   3. Now I'll look at the same segment that was an issue (2nd 3-minutes in the 4th)

      1. We are being outscored by 3.0 points in our Wins in this segment - again consistent with the Overall Chart

3. I can take a look at our Segments in Conference Only play and see everything aligns again

   1. We are allowing a 15.6 point increase from 1st half to 2nd half

   2. We are being outscore by 6.7 points in this segment

4. Based off this, I can tell there is clearly something going on in this segment in all of our games

   1. Also, in our two losses, we lost by 22 and 9. The 22 point loss does play a decent part into the Overall Chart. That one game counts for 14% of all of our games through the 7 games.

   2. I do keep this in the back of my mind but I don't find it as important as what I am looking at in the Overall & Wins

After I looked at this information, I decided to take a look at our lineups to see who we have had in the games during this segment. Thanks to Just Play, I can pull our lineups and filter it for just the 4th quarter and then select a time frame. I added a few seconds onto this segment. Below is our 4th Quarter Lineups that were used between the 7:05 mark to 3:55 mark sorted by Minutes Played (MIN):

What I see from our Lineups:

1. We have only used 1 lineup in more than one game

2. The bottom row is the TEAMS stats in this selection

   1. Total of 22:10 minutes

   2. As a team we have a -68.32 Offensive Efficiency (OFF EFF) +/-

      1. This very much falls in line with what we see in the Segments Charts

3. Of the lineups that have played more than 1:00 minute in this segment, only TWO had a postive OFF EFF +/-

From all of that, it confirms for me that we are a better 1st half team than 2nd half. There is a drop off in the 4th Quarter, especially in the 2nd 3-Minutes. I took this information to our Head Coach to discuss. Remember, small sample size but we were able to discuss, with data, what was going on. We looked at each line-up to get a baseline of info. Maybe players were in for too long or maybe focus was wandering or we were shooting bad or a variety of other reasons. We knew something was off in this segment. We didn't come to profound conclusion but we had the data to see something. I ended up presenting this to the team so they could see the numbers. It wasn't a long discussion but was something we wanted to show them so they could think a little bit more about the game. Maybe they won't relax in the 2nd half if we have a big lead. I would consider that useful if that was their take away. I don't believe everything has to have a profound conclusion. As long as there is something to take away and the info is retained, it was worth the deep dive.

To wrap this up, here are our numbers at the end of the season:

1. We scored 33.1 in the 1st half vs 33.4 in the 2nd - 0.3 increase

2. We gave up 29.0 in the 1st half vs 32.7 in the 2nd - 3.7 increase - no where close to the 11.2 we were at thru the 1st 7 games

3. In the 6:59 to 4:00 segment, we finished with a +0.3

   1. Not as high as we want BUT at least we are on the positive side. We were a -3.4 thru the 1st 7 games. Good turn around.

Maybe this little project helped us. I would like to think so. There could have been, and there is, more to look at than what I touched on but remember stats can be endless. I saw what I thought mattered and ran with it. Hopefully you have gained a little bit as to how to use stats and enjoyed this little tidbit into a project I did.

I will be posting soon about some of the Pre-Game and Post-Game stats packets I made during the season. In the meantime if you want to take a look at some of the work I did this year, here is a link:

Thanks for reading!