[Tool] NBA Stats Tracker

[Leftos]

I understand your English just fine. The addtional columns you requested shouldn't be too hard, I'll get them in my ToDo list and do them some time soon.

As for the copy-paste issue, I have no idea what could be going wrong, really.

[WBT99]

First of all, thanks for the Tool. Much appreciated.

Anyways, about using your tool for custom leagues or other leagues aside from the NBA, I would just like to ask about the formulas you used for the Metric Stats. It's because I noticed some inconsistencies and discrepancies.

For example:
I created a new database for a custom league (4 Quarters, 10 Minutes per Quarter)
To make it easier, I only put in 2 teams (Team 1 and Team 2) for the said database. After I put in the Box Score for one game, I checked the Team Metric Stats in the League Overview. Several things came to me as odd.

1.) The ORTG of Team 1 is not equivalent to the DRTG of Team 2, and vice-versa when, theoretically, they should be equal.
2.) The number of possessions had a little discrepancy with my own calculations. Did you use HoopData's or BR's formula? Also, shouldn't both teams have the same number of possessions ala BR's Poss formula which averages each teams possessions?
3.) The Pace for Team 1 and Team 2 was not equivalent to their respective Poss value when, again theoretically, they should be equal since they only played one game against one another.

Working backwards and using formulas from BR and HD, I figured that the calculations for Team Metric Stats, specially Pace, are focused on 48 minute games like those in the NBA. I know that this tool is called the NBA Stats Tracker but I believe that it's also very useful for non-NBA leagues -- leagues that don't necessarily play 48 minute games. Also, I think that Team 1's ORTG not being equal to Team 2's DRTG because is caused by Team 1 and Team 2 having different Poss and Pace values, am I correct?

A suggestion would be to change the formulas for these Metric Stats to be more generic and custom-league-friendly. If you'd be so kind to tell me what formulas were used, I could help you as well.

The matter regarding Possessions being equal or not is a matter of preference, I think. Some believe that it is better to average the number of possessions of two playing teams since teams usually have the same number of possessions per game (see: Basketball Reference; A Starting Point for Analyzing Basketball Statistics) while some don't. It seems that the formula used for Poss in the Tool is more inclined to the latter, possibly solving the number of Possessions for each team in a game and not averaging them to make the number of Possessions for each team in a game equal. This then results in the difference in ORTG and DRTG as well as the other Team Metric Stats.

With regards to Pace: Instead of using HD's formula for Pace - Pace = ((FGA - OREB + TO + (0.44*FTA))/82)*(19860/Minutes), you can change 82 to a variable holding the value of games played and 19680 to total minutes played without OT (games played * minutes per game).
Or simply use something akin to BR's formula like: Minutes per game * (Team Possessions + Opponents Possessions)/(2*(Total Team Minutes Played/5))
I've done something like this for custom leagues using spreadsheets but I think your tool is a much viable option than creating new spreadsheets per season/league so I'm hoping that you can add this option as well.

Again, thanks and I do hope I don't sound ungrateful and pushy. Haha.

[Leftos]

WBT99, first of all, I want to start by saying thank you. Thank you for taking interest in NBA Stats Tracker, and for being so detailed in your criticisms and your suggestions.

I'm using BR's Possessions formula from here to be exact. I just checked the formula in the code for NST, and it matches up with the one I've linked you to, unless there's some weird operator precedence in C# (the programming language I'm using) which I'm not aware of, but I've even replicated the parentheses in the formula, just to be sure. Problem with your suggestion regarding getting the average, is that for performance reasons, the Team Metrics aren't calculated on a per-game basis, but on the total team stats. So there's no way for the tool (currently) to make it so both teams that play against each other have the same Poss for the same game. Unless we can edit the formula that's based on Totals to something better, I don't know what else to do, and I really don't want to make it calculate metrics on a per-game basis. I may do it in the end, if there's no better way to be accurate, but I'd rather not.

You're right as far as Pace goes. I dropped the ball there. I used the BR formula without changing the first 48 (which even the BR Glossary mentions should be adjusted, duh) to the Minutes Per Game. Do you think it should be the team's minute per game average (taking into account how many times each team has gone to overtime), or should it be a fixed value by asking the user regarding the league's rules on full game length? Say, if a team has played 48 minutes and 53 the next time, should MPG in the Pace formula be 50.5 or 48?

On ORTG and DRTG, I've used the simplistic formula of (Team Points / Possessions) * 100, to calculate the points each team does per 100 possessions. It doesn't take the calculated Pace per se into account, but you understand how it's tied to it indirectly. So yes, if both teams had equal Possessions, the one's ORTG would be the other one's DRTG. But that requires that we modify the Possessions formula, or that it's calculated based on the per-game average.

What you're saying regarding spreadsheets is exactly why I decided to get NST to be something more than a tool to fix NBA 2K12's stats bug; I want to help people like you. Coaches, fans, everyone that wants to keep track of their favorite league and do their own analysis on it. So be as ungrateful and pushy as you want, if you're going to go into this much detail every time. I really want to work with you and discuss your suggestions on what you believe is inaccurate. You seem to have more experience than I do in the analysis part, and I have the programming knowledge to put what we discuss to code, so that you can enjoy it afterwards. I want to make it friendly and compatible with any league besides the NBA (well, besides those that allow draws, NST doesn't handle that currently), so suggestions like yours are quite important.

With that said, waiting for your answers then! If you want us to have a more direct way of discussing this so that I could try some things out as we talk, send me a PM to exchange IM info.

P.S. I've mostly used BR for reference on Metric Stats formulas. Do you think there's things I'm missing by not checking out Hoop Data, like some metric I haven't implemented yet?

[WBT99]

Thanks for the response.

I understand now why the Poss values as well as the ORTG and DRTG vary. You use Team Totals for the season rather than compute it per-game, which is indeed better performance-wise since you only need to save one set of values (Team Totals) rather than create a whole new set for every game and add those again for the Totals. I understand that and, as I said, when it comes to Possessions, there's nothing wrong with two teams playing against each other not having the same number of Possessions. If you do decide, in the future, to save and compute Team Metric Stats per-game, I believe it will be a useful addition to the tool as it could be used to view and compare the different Stats and Metric Stats of a Team as a whole on a per-game basis (like how well the Team performed during a particular game) and probably tell you how well a team has improved or worsened over the course season. Still, even if you don't include or implement what I said above, what you've done here with the Tool is already an amazing thing and highly commendable.

With regards to the Pace: From what I'm familiar with, the minutes per game in the formula has a fixed, constant value. It is equivalent to the number of minutes of a normal, non-OT game or, as you've mentioned, dependent on the league's rules on full game length (48 minutes for the NBA, 40 minutes for the WNBA, etc.).

In terms of the Metrics available in the Tool, I believe that you've covered all the important stuff -- ORTG, DRTG, Pace, Poss, USG, PER -- that most users and analysts need. Of course, these are also the Metric Stats in NBA 2k12. Still, if you want to add more Metrics, there are still plenty of not-so-popular Team and Player Metrics -- 3PointRate, Scoring Possessions, Non-Scoring Possessions, InFloor%, WinShare, etc. -- both at BR and HD. These metrics aren't used as often or as known as the more popular ones like those already implemented in NST but they also have their merits at times. But if they'll just clutter up the NST, it may be best to stick to the more popular and known Metrics.

[Leftos]

I could later on add more metrics, wouldn't be a problem. The interface of NST is highly adjustable, and I've created so it fits on nearly any window size you want.

As for the Poss values and what you've described, doesn't limiting the timeframe fix that? I mean, when you switch from Season stats to time-frame stats in League Overview, it calculates the Totals based on any box scores that exist within that timeframe. So wouldn't that solve your issue on per-game or per-week or per-month or per-[insert period here] stats and metric stats? Timeframed stats I feel is one of the most useful features of NST. Still, it doesn't average the Possessions for each game, but it allows you to judge the progression of a team on any basis you want.

Thanks for all you kind words. I'll have to add some code that asks the user for the minutes per game, so they're the ones used in the calculation of the Pace metric.

Oh, and by the way, when NST asks for the minutes in the box score, you're supposed to enter the actual minutes of the game, not the usual (minutes * 5) most sites use as a total of the minutes played of all the players.

[WBT99]

Yup. I figured that NST asks for the actual number of minutes of the game while trying the Tool. I also agree that the Time-frame feature is great and already allows the user to view a team's progression and even allow you to view a team's stats for each game the season (given of course that a team does not play more than one game in one day, are there leagues who do that?). The functionality that a per-game calculation of statistics for teams could add is that it would allow the user to view a team's stats in multiple games simultaneously. Again, you're the developer and the decision will be up to you.

Thanks again and I'll be waiting for the next release.

Regards and Cheers.

[Leftos]

I could add Metric Stats like that in the Box Scores tab in Team Overview, just like I've added Game Score in Player Overview's Box Scores tab. Something like that?

[Leftos]

v0.13.1 released!

- Addition: A game length different than 48 minutes can be set via the Tools menu
- Fix: Team Pace Metric in now calculated on a Game Length basis, instead of assuming games are always 48 minutes long
- Change: All screens showing best performers or performances now present the 5 most significant stats, instead of 4
- Change: Multiple instances of NBA Stats Tracker are now allowed
- Change: Miscellaneous menu was renamed to Tools menu
- Improvement: Various minor interface improvements

[WBT99]

I think so. Just something to allow users to simultaneously view the stats of a team in the games that the team has played.

[Leftos]

Can you give me a mock-up or a description of what you'd like to see in which tab? Wouldn't some additional columns in Team Overview's Box Scores tab fix that?

[WBT99]

In the Insert Boxscore screen, for a custom league with custom players, is it normal for the stats of a player to change after I've inputted those stats for the said player and went to input new stats for another player on the same team? This happened quite frequently while inputting stats for a game. For example:

Say for Player 1 of Team 1, I'll input: 12PTS, 5REB, ..., 4FGM, 8FGA, ..., 2ORB, 3TO. Once I reach the end of the line and hit TAB, or enter, or even change to another player, some of the values I inputted for Player 1 suddenly change. Oftentimes, these are the values for REBS and FGM. The value for REBS either increase or decrease (I think it has something to do with the values in ORB), while the FGM values become zero. The odd part is that even after the FGM values became zero, the FG% values remained the same - for the example above, the FG% of Player 1 will sitll say 0.50 even if the FGM value became 0.

I had to recheck the values of the stats and reenter them as necessary every time I went to enter new stats for a new player (the changes only occur on the stats of the player I last edited the stats of). I also found out that even if the values I see in the cells change, when I double click on them the highlighted value that is shown is the original value I inputted.

However, in the case of the REB and ORB, if I change the ORB value, the REB value changes as well. Though this does not always happen.

Will this be a real issue or could it just be an isolated case?

Thanks in advance.

[Leftos]

Are you using the Update with Box Score, or the Live Box Score screen?

[Leftos]

v0.13.1.1 released!

- Fix: Fixed bug since v0.12.2 that wouldn't allow user to insert new rows in Player Stats tabs in the Update with Box Score window

[WBT99]

I was also experiencing the bug that didn't allow the addition of rows in the Player Stats tab in the Update with Box Score window. A workaround I came up with was to use the Live Box Score Screen to choose the Teams then use the Copy to NST Box Score. That was what I was using when I noticed the glitch I reported.

Since you've mentioned that you've already fixed the bug, I'll try using the Update with Box Score directly and see if the glitches still occur.

[Leftos]

When you have to find a workaround like that, you should report it to me. Some things are just meant to work, and the fact that new player rows couldn't be added to the box scores wasn't on purpose.

In any case, you're saying the bug happens when you copy a live box score to a normal one and then try to edit the information?

[WBT99]

Yup, looks like it only happens when you try to edit a Live Box Score that's been copied to a normal one.

I tried the update and used Insert Box Score directly - and, voila, the glitch I encountered hasn't occurred yet.

[WBT99]

Oops. I think I may have encountered a problem. I was able to create rows but now, after saving the Box Score, the Player Statistics are not saved.

Several things that I noticed where that:
1. The Player Statistics inputted cannot be viewed in Player Stats Tab in Team Overview.
2. When I open the Box Score, the stats i inputted for the Players disappear.
3. After inputting the stats in the Player Stats Tab in Update with Box Score, the Metric Stats were not automatically calculated.
4. Also, unlike when using the Live Box Score-Copy to NST Box Score method I used before where the Minutes is entered automatically by the tool, when using Update with Box Score, I have to manually enter the minutes played by each team (This could be the source of error or I could be completely wrong).

Again, thanks and sorry for the BIG bother.

[Leftos]

There were various bugs in the lists that held the Player Stats information both in the Box Score window and Live Box Score. I've corrected them and put up v0.13.1.2. See if they're fixed, and if they're not, reply again with the problems you're having, and please include steps to reproduce for each issue.

Don't apologize, I apologize for not being able debug the tool more deeply so that problems like these don't come up. It's still in development, unfortunately. And I appreciate your help.

EDIT: About #4, the Update with Box Score does not insert the minutes automatically on purpose. If it did, you may forget to edit that and correct the minutes. It's a minor annoyance, but it's on purpose.
EDIT 2: About #3, unless both Team and Player Stats are entered correctly, Metric Stats aren't calculated, and usually an error comes up to tell you why.

[Leftos]

v0.13.2 released! Thanks to Lagoa and WBT99 for the ideas

- Addition: Added FGM/G, FGA/G, 3PM/G, 3PA/G, FTM/G and FTA/G averages to Player Stats in Team Overview, League Overview and Advanced Player Search
- Addition: Added TS%, 3PR, PythW, PythL metrics to Team Metric Stats in League Overview
- Addition: A season length different than 82 games can be set via the Tools menu (used in Pythagorean metric calculations)
- Fix: Box Score Window's player metric stats can now be copied to clipboard correctly
- Addition: Team Overview's Box Scores tab now features more information regarding the team's performance in each game

[WBT99]

I'm just glad I could help in the development process. I'm not complaining or anything at all. I'll check out 0.13.2 and tell you if I find any more problems with it.

[WBT99]

Looks like the glitch I mentioned got resolved in the update. Great work!

However, I found another problem (I know, I can't seem to run out of them. I'm sorry.) that has something to do with the consistency and realism of the inputted Team Box Scores.

It seems the the NST allows unrealistic stats to be inputted. I realize that it's the responsibility of the user to ensure the accuracy of the stats being inputted but sometimes mistakes happen and its up to the NST to check for those mistakes and handle them. I've seen several implementations of this in the NST like the message telling you that you can't have more 3PA than FGA or 3PM than FGM or OREB than REB. But I think that other error-checking mechanisms are still needed specially for consistency and realism of the Team Stats.

To show you what I mean, here's a sample of what I inputted in the Insert Box Score screen:

Team 1: 20 REB, 20 AST, 20 STL, 10 BLK, 10 TO, 15 FGM, 15 FGA, 15 3PM, 15 3PA, 15 FTM, 15 FTA, 10 OREB, 10 FLS, 40 MIN
Team 2: 20 REB, 15 AST, 20 STL, 10 BLK, 10 TO, 10 FGM, 10 FGA, 10 3PM, 10 3PA, 10 FTM, 10 FTA, 10 OREB, 10 FLS, 40 MIN

If you'll notice, the Box Score above is inconsistent and unrealistic for a couple of reasons:
1.) Teams have more AST the FGM.
2.) Teams have more BLK and/or DREB (REB-OREB) than the number of FGA missed by their opponents.
3.) Teams have more ORB than missed FGA.
4.) Teams have more STL than the Opponent's TOS.
5.) There are several other inconsistencies that go against the basic logic of a basketball game - like Team 1's 15FTA on (15FGA and 15FGM) mean that on each of the 15FGM of Team 1, they were fouled. This is inconsistent with Team 2 only committing 10 FLS.

The problem is that NST saves and accepts the Box Score above without reporting an error or warning the user. Can this be remedied? As I said above, inputting the data and checking for consistency is also the user's responsibility and I would understand if you don't feel the need to include this or add more methods to handle this on NST since it could demand a lot of work. I just wanted to point it out because it could be important.

[Leftos]

Those are some nice checks that NST could do. Numbers 1, 2, 3 and 4 are easy to do. Number 5 is more subjective, and can't be really checked for. I mean, you can't judge "realistic" FTAs based on FGAs, because a team could have been fouled all game and got even more FTAs than FGAs. Highly improbable? Yes. But it's not a check that NST can do and just "fail" if it isn't as it's supposed to be.

A check that could replace #5, is FTAs being at most OppFOUL * 3.

[WBT99]

I figured as much. The first four only entails comparing the values entered by the user while number 5 is indeed tricky since it deals with the logic and dynamics of the game.

Anyways, I'll look for other inconsistencies to report. One thing that I'm planning to check is the consistency between the inputted Team Stats and Player Stats. I know there's an option to Calculate the Team Stats directly from the Player Stats and inputting only select Player Stats but does NST have the ability to check if, say, the Accumulated Player Stats is greater than the inputted Team Stats. That can also be added if not already there.

[Leftos]

Should be really easy to do as well. You're not going to let me have any summer holidays, are you?

Thanks for all the suggestions, keep working with NST, see what you can come up with, and I'll work on everything we discuss.

[WBT99]

Maybe. Hahaha.

In my defense, I didn't know it was summer holidays there in Greece. It's just mid-semester from where I'm from and classes are suspended this week so I have a lot of free time on my hands.

Thanks also for taking the time to read my suggestions.