BuzzerBeater Forums

Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital.

Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital.

In your opinion what are they concealing?

I was waiting for his answer... something like: "Team ratings are well known as quite useless"

I would have responded talking about MR ....

Ok effort, ok random, but i think that this match is outside of all logic.

Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital.

In your opinion what are they concealing?

Ratings, they played CT.
Ofc there is a bad luck for you in this game, but like everywhere, a batter team doesn't win every time..

Cheers!

Ok, it's shit
but I still think (in BB) it's ridiculous

Cheers

You're thinking it's ridiculous and you are right despite the usual attempts by Manon and others to defend indefensible blunders without any evidence.

Also for the guy who mentioned GSW@DEN and his supporters who voted him up:
GSW: EFF 123, GmSc 92.4 (EFF=Efficiency as measured in Buzzer-Manager, GmSc=Game Score one of Hollinger's stats)
DEN: EFF 160, GmSc 112.8
One can see why Denver won: +37 EFF and +20.4 GmSc, there is a massive positive difference in favour of Denver. If they replay that game 10 times, GSW may win every time, but the statistics of the specific games will reflect that. The problem with this game is that the statistics do not reflect the outcome because of gimmicks in the GE.

What happened here:
HOL: EFF 72, GmSc 41.3
ITA: EFF 103, GmSc 59
Holland -31 EFF and -17.7 GmSc. I've never seen a team in BB winning a game with -31 Efficiency, it should probably go in the record book. Note that EFF is used in Buzzer-Manager and it's not great, but it's very rare for a team to win with lower efficiency. GmSc is also not perfect but much more robust.

Want to look how a tight game would look in real life using these basic efficiency measures? DET 102 @ 101 TOR
DET EFF 106, GmSc 77.2
TOR EFF 112, GmSc 78.2
Toronto lost on a buzzerbeater 3 pointer with +6 EFF and +1 GmSc.

There is a reason why you won't find teams losing with +20 GmSc in real life. If you go through the play-by-play for this game and analyse it you will understand what happened. Count the number of possessions, break down how the fouls impacted the teams, check the referee's decisions, rebounds going out of bounds etc.

It seems to me the astrophysicist did a remarkably poor job in modelling Effort. Not only he boosted shooting beyond any recognition in the PP100 ratings (Holland got the highest shooting efficiency in the game from the SF spot which had the second lowest PP100 in the game at 57.7), but he also stacked possessions due to offensive fouls, referee decisions, rebounds ending out of bounds etc in favour of the team using more effort.

BBs: if you can't change the GE so that it's realistic and you want to keep using some gimmicks (9 rebounds ending out of bounds for Italy, really?, offensive fouls 5-0 to Italy when the total foul count without the intentional fouls was 15-20, really?) at least give us some meaningful ratings. Or remove them altogether.

Last edited by Lemonshine at 2/14/2017 6:02:16 PM

If you just use the ratings the way you did here (selectively if anything) then you could assume that the multiplication effects of efforts kill your advantage pretty quickly. Assuming CT is 1.20 and TIE is 0.80 (too lazy to google forums) then
inside scoring = 11 to 9.5
outside scoring = 10.5 to 12
rebounding = 8 to 7
game flow = 8 to 7

Italy by 2 in that case. And that doesn't even include how the actual performance of each individual player is effected by game strategy, position and game shop.

This isn't like the 76ers beating GSW in 7 but more like Twolves beating GSW when Kerr was out with back spasms and Curry was more interested in dancing on the sidelines. Tactic just beat mono-monsters (Come on - Thib can outcoach a Kerr on pain meds any day!)

Last edited by Randy Maus at 2/14/2017 3:21:39 PM

Repeat for the people who feign ignorance or cannot understand. The ratings are one thing the statistics and what was actually simulated by the GE are another. And if you cannot understand this, then I urge you not to post, especially if you are trying to explain to our Italian friends how -18 Game Score can produce a win without stuff that does not go in the box score because it does not exist in real life. Like, for example, those 9 rebounds ending in the stands after being grabbed or offensive fouls being called only on one team.

If you can't understand why a statistically MUCH better team ON THE NIGHT should win a game both in real life and in this game then there is nothing anyone can do for you. Just trust those who tell you you are completely missing the point. Ratings are useless? Remove them, but by god at least have the decency of letting the statistically better team on the night win.

Last edited by Lemonshine at 2/14/2017 6:39:57 PM

Seems some that have posted here would like to see a game where higher rating always wins no matter what.
I can't think of a more boring game then that.

We want ratings that are meaningful. Tell us what you understand from the ratings for this game, if you think they are ok.

Replay this game a hundred times and I am sure Italy wins 99 of those games as a minimum.

You don't know that. And I will tell you more: Marin and Astroguy don't know that either.

But to see that number become 100% would be a very boring game.

This is fairly simple: the ratings should give an idea about what's reasonable to expect in normal conditions given the match-up, so they should be the expected average of the statistical distribution used for the events they describe. This means that:
- All players on one team outscoring their expected average
- All players getting more rebounds that their expected average
- All players getting more assists than their expected average
- etc
- etc
is extremely unlikely. It is also blatantly evident that the current Game Ratings and PP100 are not useful to draw conclusions from games. And I'm not talking about a small divergence because some skills are not accounted for.


That said, it is not even the point for this game as I have tried to explain already. What you are saying is very similar to: even in the 0.01% of cases were a team overwhelmingly outperform their expected output across the board and are statistically superior and more efficient than the opponent, it's fine if they lose due to imperfections in the GE.



Last edited by Lemonshine at 2/14/2017 6:41:06 PM