BuzzerBeater Forums

Bugs, bugs, bugs > MVP

MVP

Set priority
Show messages by
This Post:
00
314072.13 in reply to 314072.11
Date: 4/17/2022 9:25:48 AM
Tampines Fusion
SBBL
Overall Posts Rated:
432432
Interesting point. My point difference is actually 7th in the league, so I guess that might hold some truth to it. But further down the chart at second place is a player from a team with -109 PD (12th in PD and 9th in game record), while the 6th placed player actually comes from a team with the second highest PD (tied 2nd in game record but 3rd due to PD), and is 2nd on the scoring charts and 9th in the rebounding charts himself. So it doesn't really tally here.

This Post:
00
314072.14 in reply to 314072.13
Date: 4/17/2022 9:30:00 AM
Wasted Potential
II.1
Overall Posts Rated:
458458
Second Team:
Hazards to Society
You're using one observation of MVP voting and drawing these conclusions haha. PD is only a part of the calculation (although by my estimate it is one of the most important). Obviously other things like efficiency, ratings, and just overall boxscore stats make an impact as well.

I agree that the voting isn't very good right now, but don't draw your conclusions based on one vote. Point differential and record are definitely taken in account.

This Post:
00
314072.15 in reply to 314072.14
Date: 4/17/2022 9:35:24 AM
Wasted Potential
II.1
Overall Posts Rated:
458458
Second Team:
Hazards to Society
If you want a bit more proof, I took the last 30 seasons of my leagues MVP voting data (not that much really, but more than one season) and used a machine learning model to try and predict this seasons MVP. It turned out decently accurate as it had all of the top 5 in its top 8 projected. The model ranked PD as the most important in determining MVP followed by rating, some scoring stats and games played.

While the model definitely wasn't perfect, I think it gives a little more insight then just looking at your leagues MVP from one season.

This Post:
11
314072.16 in reply to 314072.12
Date: 4/17/2022 9:38:47 AM
Venomous Vicious Vipers
Pro A
Overall Posts Rated:
917917
Second Team:
Veni Vidi Vicious
In the last 14 seasons of Pro A, the MVP was in the best team 13 times and one time in the second team. And more than 95% of the players getting votes are in top 8 teams. To me that's some concrete proof lol

If I were coding a game, I would use both player's efficiency and team's record to decide for a league MVP, as it is the case IRL in most sports.

Your example is a comparison between your 80k$ player and 5k$ players.. Come on

This Post:
00
314072.17 in reply to 314072.16
Date: 4/17/2022 8:04:13 PM
Tampines Fusion
SBBL
Overall Posts Rated:
432432
In the last 14 seasons of Pro A, the MVP was in the best team 13 times and one time in the second team. And more than 95% of the players getting votes are in top 8 teams. To me that's some concrete proof lol

That's not how you should look at it. The only way you think that way is because winning is directly proportional to good performance. Did any of the MVPs underperformed worse than all the other players across the board? If yes, then you might at least have an argument. If the MVP is actually one of the top performers in the league, then you don't really have an argument there.

Like I said, if winning was such a huge factor, it won't matter if you have a $80k player beat all the other 5k players, if your record is still bad, you still won't win, or at least not in such a dominating fashion. The fact that one of the worst teams in the conference can have their best player win the MVP shows how insignificant winning record is.

This Post:
00
314072.18 in reply to 314072.14
Date: 4/17/2022 8:10:24 PM
Tampines Fusion
SBBL
Overall Posts Rated:
432432
PD is only a part of the calculation (although by my estimate it is one of the most important). Obviously other things like efficiency, ratings, and just overall boxscore stats make an impact as well.

Fair enough. Tho what actually do you mean by "one of the most important" part of the calculation? Is that in terms of weight age? Coz IMO based on what I see (and it's not only one season really, the only reason why I only use that season because it's the easiest season to pull numbers out, and it's an anomaly, coz how many tanking teams will have the best player in the league), PD at most will only just be a tie breaker should both players are on equal terms. It certainly is important in determining the MVP but significantly less weightage as compared to box score and efficiency.

This Post:
00
314072.19 in reply to 314072.18
Date: 4/17/2022 8:21:40 PM
Wasted Potential
II.1
Overall Posts Rated:
458458
Second Team:
Hazards to Society
I offer evidence based on gathering data and evaluating it through proven methods in response to you asking for "evidence".

Your evidence is just based on what you claim to see (nothing backing it up). I think you have your mind made up.

The model is question can give an importance rating to each variable it uses based on how often it uses it and how big of an impact it has overall on the prediction. It's a random forest model if you know how that works or want to look it up. Like I said, it's not a perfect model but I think it gives a little more insight then trying to discover the pattern just by eyeballing it.

This Post:
00
314072.20 in reply to 314072.19
Date: 4/17/2022 8:23:32 PM
Wasted Potential
II.1
Overall Posts Rated:
458458
Second Team:
Hazards to Society
Additionally, remember that games played is a factor in MVP voting. This means often when better teams hold out their players for the cup they are lowering their MVP odds.

This Post:
00
314072.21 in reply to 314072.19
Date: 4/17/2022 8:59:17 PM
Tampines Fusion
SBBL
Overall Posts Rated:
432432
I get where you are coming from, I'm not totally writing you off, but again, I'm suggesting that there could be sampling bias in your "evidence". How many of them were of leagues where the top teams had the best players in the league? And how many of them had a tanking team with a dominant player? If there is sampling biasness, then no matter how large your sample is, it will still come out with similar/same results, regardless of what model you use.

That said, I totally agree that number of games is a factor (which is also why the only other team with expensive players in my league don't have anyone in the list). But I'm curious as to how many game do you need to make the MVP voting. Could it be possible to see what was the lowest games played for a MVP within your sample?

This Post:
00
314072.22 in reply to 314072.21
Date: 4/19/2022 11:46:36 AM
Wasted Potential
II.1
Overall Posts Rated:
458458
Second Team:
Hazards to Society
I mean even if the top teams tend to have the best players, the model is usually flexible enough to recognize that. If point differential and record really didn't matter it would likely be identified. However, I do agree that my sample wasn't big enough or fully representative. I could collect more data and see if anything changes.

I do agree that the formula could be much better as sometimes the MVP is not a player I would have expected or that seems like the best player in the league.

This Post:
00
314072.23 in reply to 314072.22
Date: 4/20/2022 12:44:10 AM
Tampines Fusion
SBBL
Overall Posts Rated:
432432
Understandable. Well it isn't really a game changing piece of info anyway. Nevertheless it is interesting to discuss about.

Advertisement