BuzzerBeater Forums

PG scoring rating on the away team is 98.6. It means on average his PGs should score 99 points on 100 shots. The PGs in that game combined for...6 points on 17 shots. Now since you are so fond of averages and you don't have to be a stat genius, or any kind of genius to understand this, that rating means that if they play this exact same game again th PG could score 47 points on 17 shots. Simple napkin math is your friend here.

That's not how statistic works, another important fact is that the past does not influence the future. If you flip a coin and it's tails, doesn't mean the next time it has to be heads or even have a higher chance because it needs to average out. You still have 50% chance to get heads, regardsless of what previous results are. So no, the PG's don't have to score 47 on 17 shots for the scoring rating to be true. Besides that, I don't you can adjust PP100 that way because of actual limitations on Points per shot. The ratings just say that if those PG's would play a statistically relevant number of games, say 1000, on average over all those games they'd score 98.6 PP100.

What would you want those ratings to display then? The actual PP100 of that game? What use is that, you can just calculate that yourself and it doesn't give you more information. The information you can gather from this is that on average the PG's should score 98.6 PP100, and because they only scored 6p on 17 you can conclude they had a bad offensive game, an outlying result. Since the PP100 is a decent number, you don't need to change anything. Now if the PP100 was much lower, you can conclude the performance of the PG's is bad because appearantly they are bad offensively, so you could try and change that.

If you don't think it's logical that a player that scores 98.6 PP100 on average has 1 game in which he scores 6p on 17, that's just how it is, good players can have bad games. Lebron James normally has a great PP100 but he has games in which he shoots badly too. Now if that player has 10 games in a row in which he shoots as bad as 6p on 17, then you can ask yourself what's wrong, because that is a much smaller chance than 1 guy having a bad shooting night.

Look I don't want to sound like a smartypants, so sorry for that, I genuinely want to explain basic statistics to you if you like. Just tell me what you don't understand and I'll try to elaborate.



Last edited by Jeründerbar at 2/27/2015 7:35:55 AM

That's not how statistic works

That's exactly like it works. You have data points above and below the average. It was clearly 27 points not 47 though. It doesn't change the point since with 27 it's still a blowout for the guests. Both 6 and 27 would be outliers around the mean of the exact same magnitude.

Besides that, I don't you can adjust PP100 that way because of actual limitations on Points per shot.

Are you saying that the ratings were not exactly indicative of performance? In that respect I agree, but it's not what the manual says.

What would you want those ratings to display then?

The sensible average calculated based on the numbers the GE will use in the simulation. Ratings are the result of analytical formulas and I believe that the 98.6 number is incorrect as it weighs skills like it correctly did with a slightly different GE.

It's no mystery that they fiddled with the GE in the last few seasons, so if they haven't changed how ratings are calculated (both team's and scoring) we're looking at incorrect numbers and people will continue to ask why the ratings do not reflect games and how they lost although they had better ratings. Conversely, it may happen that you know you've been dominated, then you look at the ratings and it seems you would had won the game 8 times out of 10.

you can conclude they had a bad offensive game, an outlying result.

An equally valid conclusion would be that the rating is incorrect and the player had a perfectly average game.

some meaningless random statistics produced as a byproduct of the randomly generated games

Ok so if ratings (team and scoring) are meaningless why even analyse a game through them? That's my point: either the ratings are reasonably correct and we can take them as an indication or we can't. It's ok if there is a margin of error and variance around the mean, but not to the point where you conclude that they are meaningless

Sorry I misread or you edited, but I thought you said the next time the PG's should score 47(27) points. That's obviously not the case, but yes the PG's could score 27 points on 17 shots, probably with the same chance as the 6p on 17. If all factors and outcomes stay the same and just that 1 outcome is overturned, yes there would be a big difference in score, but that's perfectly possible too. It's just less and less likely that all the factors are in favor of one team to cause a blowout if the teams are about equal in strength. I think there's so many statistics that are calculated that it's hard to have them all line up and cause a huge anomoly. Maybe there's a limiter in the GE to preven that from happening, I don't know. I do know that players can have bad games.

Even if the PP100 aren't entirely accurate, I don't think it differs immensely from what it should be. The changes in the GE aren't that drastic, so why would the PP100 calculations change drastically? Even if it's 10% difference and the PG's would have 90 PP100, the 6p on 17s is still a pretty big outlier.