BuzzerBeater Forums

You are right, there were assumptions made about my FT frequency ratio.

- That all the FT attempts were made on a two-point FG shot and
- That the player actually missed the FG.
- That all shots, inside or out, have an equal chance of getting fouled. (This of course is wrong, i think it even says in the game manual somewhere, that inside shots and as such, inside players, have a better chance of getting fouled)

In order to incorporate these assumptions i did the following.
I watched about half my games live, and this season, I have seen one player shoot an 'and-one' FT shot.
Likewise on the 3-point shot fouls.
So i decided that these two assumptions, for my team could be ignored. The third assumption I just dont really know how I would incorporate this. Give a slightly higher weighting to 2-pt FG's? A weighting on PF/C ? Not sure. Very difficult to try and incoropate. So lets just leave it at an assumption at this point.

So yes, whilst this is an assumption, I believe that for my team, at this early stage, the frequency of the 'and-one' and the '3pt shot' fouls, would not affect the statistics too much. So I wouldnt call my method 'extremely flawed', but people do need to be aware that if their teams are shooting lots of 3pters with fouls and the and-1 FT's, then this method is not going to be as good a representation as other teams.
But ive just downloaded something off here, which analysis the play-by-play commentary, i havent checked it out yet, it looks like its a good tool, so I might be able to filter my games and provide a representation of how often this happens for my team.

Here is a link graph as requested, sorry I didnt post it last night, I was at home, and I had done the graph on my work computer.

http://imageshack.us/photo/my-images/42/ftvsftaratiograph.png/

My initial correlation coefficient was only 0.337! Which was sucky, so like all good statistical analysis I removed one outlier, and the relation coefficient changed to 0.670

Im really not convinced yet that there is strong positive correlation, I think if there is going to be a correlation, it will be weak at best. If someone wants to send me their list of players stats i can run it for their team. I will need Games played, Average minutes played, FGA-FGM, FTA-FTM, and that players FT rating. I copied mine straight from Buzzer-manager.
It can be annonymous, I dont need players names.

i think the biggest flaw are the "end" game fouls when you watch the pure stats, in this phase of the game the offensiv team is much better in getting their good free throw shooters to the line even without a shot. And it isn't rare that you get 8+ free throws like this in a close game so the number is signifikant and leads to more freethrows of the good shooters.

Also fouls in the bonus, are giving free thows without a shot, even when they are in my eyes independent from free throw skills so not only the 3pt plays have less then 1 shot per two free throws.(and a 2 ppointer + one free throw should be as common in lower and high levels)

Your idea made me curious so I had a look with my roster.

I considered two analysis:

A) FT Skill vs FTs/(FGA+k*FT/2)
where k is a number between 0 and 1 (1 is just the case you made). I considered different choices for k, since, as the GM said, the number of total FTs can be significantly different from the double of shooting fouls (namely, the number of "additional FGA" you must add is smaller).

B) FT Skill vs FTs/(FG Missed + k*FT/2)
A good player should make a lot of FGs, so his calculated ratio would be pretty low even if his FT skill is good. So, I considered only missed shots here (k is the same as before)

Results...not good, I fear.

http://img18.imageshack.us/img18/1340/catturaxs.jpg

Correlation value is 0.37 (0.57 if I delete the "outlier" at the top left corner of my graph, but with a so small sample even a player is significant), while the R-squared value is just 0.14 (0.32 witout the outlier).

Moreover, the player coloured by yellow in the graph only played 20 minutes in my season. If I delete him (after all, he didn't play enough, his data could be caused by random) both the correlation and R-squared stat fall (to 0.26 and 0.07 respectively).

Notice that this result is indipendent from the kind of analysis (A or B) and from the choice of the k value, (there is only a difference of 0.01-0.02, nearly, between the various values of correlation.)

Just to be sure, I added also your sample to mine (I used your graph, maybe the values were not exact but they should work). If I use both the samples together the correlation and R-squared both fall, again (corr=0.31, R^2=0.1)

After all, it's enough to see my graph to think there is no particular correlation. If you don't consider the yellow point, it seems just white noise to me. So i can only conclude there is no correlation between FT skill and the number of FTs.

Which is not strange, after all. If the shooter is not good at FTs, he won't look for them. But the defenders won't mind if they stop him with a foul...

Last edited by Paco el Niño de Piombo at 6/1/2011 5:08:35 AM

Cooooooooool! Nice stats.

Yes i cant believe i totally forgot about bonus free throw shots that happen when the other team has too many fouls. TOooooootally forgot about that. I feel ashamed to call myself a basketballer!

Yer those R-squared values are quite low, although our samples sizes are probably a bit too small for that kind of analysis. Maybe if i had a full year of data, plus satgnent FT levels all year, but at this stage it is looking like its pointing towards no correlation.

Which i agree, i remember when i used to play basketball, and we had aguy who would not miss from the line, we would just set him up to get fouled as often as possible driving to the ring etc. So it was a deliberate tactic, and i know players in the NBA go for that, someone like Dirk nowitzki definately does that. So maybe Buzzerbeater should do this?

i actually think it does, but maybe not all game. Liike the GM says, when you watch the game when there is bonus situation and its close, not only do good FT shooters tend to go to the line more, but also i think to some degree, that the coach sub's some of the better FT players on, dependant upon the situation. So maybe the game engine only takes this into account in this special circumstance?

one reason for the slight correlation might be that the FT skill is higher for older players in general, as are the other skills... so trainees might distort the picture.

Hey guys, good discussion, some interesting ideas coming out here.

To the poster who said 'older players' i can only assume that you actually meant players with higher experience will get more FT's per FGA.
To which i say, probably! Experience is this engima-type skill, which, is probably implemented a lot simpler than what most people would believe! But i do believe that experience has some kind of effect on this ratio.

As for driving, thats an interesting comment. Im not sure. I certainly don't disagree, I quickly analysed it same way i did with FT rating, and essentially its a worse 'direct' relationship between Driving and FTA.

But again, im only looking at my team, whom i mainly play an inside offence, so it could very well be different for a team who plays outside offence. I have a feeling the number of factors are large. Depends on opponents defense, defensive skills of individual opponents, your offense, game situation etc etc. in which case you would need a very large pool of players and data to really analyse this properly, all of which i just dont have. So no biggy.
I think in the end, its probably more related to, as you said, the hidden physicality trait that is within the game that people talk about. And its probably multiplied by a factor of X, by other stats, including Driving, FT, IS, Experience, etc etc.