Near future plans Topic

monkee
Posts: 0 (0)
Hall of Famer
8 192 7 7
Prospect
Quote: Originally Posted By seble on 8/11/2008




Normalization has nothing to do with pace. It has to do with adjusting for different eras.

like I said in another thread - niceties and nuances

what aside from pace accounts for the differences in these specific stats between eras?

FG% is the aspect of the game that differs the most among eras, which is why we decided to normalize that.

based on what mr 'no subjectivity'? did they move the rim? did they take the lid off of it in 67? talk about subjective decisions

The inflated rebound numbers were due to having to estimate team stats.

no kidding - and they are your guestimates - mr elias, tracyr could have told you better - but nevertheless a simple glance at the results could have told you that you had screwed up - who is testing this stuff? - seriously - you wanted to use rbd% so 60s players dont dominate because of relative pace issues and then you make it so they are even more dominant by assigning them guessed numbers that are even more aggressive than the per minute numbers - wtf?

Having now acquired the actual stats, rebound% for those players falls much more in line with everyone else.

no kidding (x2) and it's been 2 1/2 months and you still havent put them into the database even though you could do it without changing a line of code


8/11/2008 11:14 PM
badja
Posts: 1343 (0)
All Star
43 1
Prospect
Quote: Originally posted by badja on 8/11/2008So what number is in use now? What # is used for 2pt and 3 pt? thx.



Let me re-phrase. What stat (FG%, eFG%, FG%+) does the sim look at when a shot is attempted to determine if it goes in?
8/12/2008 2:47 PM
colonels19
Posts: 4399 (0)
Hall of Famer
78 90 3 1 168 2 1 11
Prospect
Quote: Originally Posted By jakotay on 8/11/2008
Quote: Originally Posted By colonels19 on 8/11/2008
Well I blew $10 on a team that went 23-59 because I mistakenly thought that I understood how the sim works...what a shame, huh?
You could really pick any searchable stat item and blame a 23 win season on it, can't you? Has any team you ever built ever stunk because you assembled it poorly? Its always someone or something other than you, isn't it?

The team I built that focused on FG%+ won 51 games. I am befuddled as much as you are (as to why the team did well).

I was using guys per my Cincinnati 73s concept...1973 and prior...thus I thought I was stealing some lower FG% gems like George Mikan and K.C. Jones, and they still blew as in FG%, thus I failed in this season because I banked on FG%+, thinking that it meant something when it actually didn't.
8/12/2008 3:05 PM
seble
Posts: 880 (0)
Hall of Famer
24 44 35 306 121 183
Prospect
Quote: Originally posted by badja on 8/12/2008
Quote: Originally posted by badja on 8/11/2008So what number is in use now? What # is used for 2pt and 3 pt? thx.

Let me re-phrase. What stat (FG%, eFG%, FG%+) does the sim look at when a shot is attempted to determine if it goes in?
eFG% and FG%+ are just for display. They're a way to evaluate the worth of players.

But we don't simply use FG% either. 2 point shots start with the player's 2pt% and 3 point shots start with 3pt%. But there are a number of modifications (e.g. def positioning, team assists, normalization, etc.) applied before we come up with a final adjusted percentage.

FG%+ is useful in evaluating how a player will fare during the normalization process. As it's been pointed out, it's not a complete picture. Right now, we're leaning toward replacing it with the 2pt%# and 3pt%#, which would be the normalized percentages against a historically average opponent. I think it would make sense for us to add a more detailed explanation of how normalization works in the SimEngine to the Knowledge Base. That would probably help clear up these different FG% numbers.
8/12/2008 3:15 PM
ashamael
Posts: 12930 (0)
Hall of Famer
544 25
All Star
Quote: Originally posted by seble on 8/12/2008eFG% and FG%+ are just for display. They're a way to evaluate the worth of players.


Wow.
8/12/2008 3:29 PM
naboimp
Posts: 2 (0)
Pro
11
Prospect
I doubt that was worded the way it was intended.

eFG% is a stat that basically combines 2 pt FG% and 3 pt FG% into one number to reflect that players who shoot and make a lot of threes may be more productive per shot than players with higher FG%. When seble indicates it isn't used, that would be correct, but the underlying data is used.

I'm a little bit fuzzy on how FG%+ is used, but the number represents how a players FG% relates to the average FG% of the season in which it was produced. Under normalization, presumably players with high FG%+ in seasons with low average FG% will shoot somewhat better than they did historically (on average - there are other factors, including basic randomness). People are calling for publication of the FG%# which would allow some prediction of how FG%+ would affect the game's FG%, but I would expect FG%# would vary depending upon the players within a particular game. In other words, if players in a particular game are all from a single season, then FG%# would presumably be the average FG% from that season, and there would be no normalization necessary. I suspect if you had one or two 50's players in a game with mostly 90's players, the FG%# would be much closer to the 90's level and the normalization would increase the FG% of the 50's players, but to what degree I do not know. (I doubt it would be to the extent that the 50's player would be fully normalized to 90's shooting percentages.) Again, if normalization is taking place, while the stat itself is not being used, the underlying data that produces the stat is being used.

As to why normalization may not be exclusively pace based, there have been rule changes as well as other changes in the style of play as the game evolved. The shooting percentages in the 50's were horrendous. Part of this was likely due to lack of talent, but part of it was also likely due to the rougher style of play that existed then. The effect on fouls is statistically apparent - most of the 50's players have a fairly high foul rate, but the effect of all that rough play on the opponent's FG% isn't statistically obvious. In theory, I would expect normalization is meant to allow 50's players to shoot better against less physical 90's players while making 90's players shoot worse when faced with the physical tactics of 50's players.

I'll freely admit I'm guessing about much of this.

8/12/2008 4:56 PM
ncmusician_7
Posts: 1832 (0)
Hall of Famer
70 164 12 18
All Star
Quote: Originally Posted By naboimp on 8/12/2008
People are calling for publication of the FG%# which would allow some prediction of how FG%+ would affect the game's FG%, but I would expect FG%# would vary depending upon the players within a particular game
FG%# would not vary, it is a fixed number. The results from possession to possession would vary depending on various factors (tempo, teammates, etc). Take a look at how the mlb sim uses the # numbers.
8/12/2008 5:52 PM
monkee
Posts: 0 (0)
Hall of Famer
8 192 7 7
Prospect
Quote: Originally Posted By naboimp on 8/12/2008
As to why normalization may not be exclusively pace based, there have been rule changes as well as other changes in the style of play as the game evolved.

none of which form an objective basis for normalizing fg% - wis has repeatedly stated a desire to adhere to objective basis for making decisions like this

you can say something like (for instance): coaches didnt let their players dunk through most of the 50s but that can be countered by a fact like Bill Russell revolutionized the game by actually blocking shots - this didnt happen until near the end of the decade (so yeah they werent dunking too much but no one was making much of an effort to block shots either)

The shooting percentages in the 50's were horrendous. Part of this was likely due to lack of talent, but part of it was also likely due to the rougher style of play that existed then. The effect on fouls is statistically apparent - most of the 50's players have a fairly high foul rate, but the effect of all that rough play on the opponent's FG% isn't statistically obvious.

that's because you are couching a post hoc propter hoc argument - the one thing is not necessarilly the cause of the other - there are a number of reasons for more fouls being called in many of these seasons including differences in pace (after the shot clock was introduced)

the most obvious reason would be that refs actually called the game closer back then (their ruling of 'contact that effects play' was actually much more rigorous than it is today which is part of why that era's games look so wide open) - add to that the fact that FTAs were alloted differently (three attempts to make 2) and the mere fact of more fouls and more FTAs really says nothing about amount of physicality in that era

In theory, I would expect normalization is meant to allow 50's players to shoot better against less physical 90's players

the thuggish players of the 90s (Knicks, Pistons, Heat) were actually allowed a great deal more contact than the players of the 50s - watch recordings

while making 90's players shoot worse when faced with the physical tactics of 50's players.

I'll freely admit I'm guessing about much of this.



anyway to your initial statement: "As to why normalization may not be exclusively pace based"

I never claimed exclusivity - I asked for what other factors there might be - if you watch some recordings of the games from back then you'll see it's for sure not physicality - in fact as an aside the game was much more physical in the 70s and %s were up

if anything your point calls for review of a normalization of PFs and FTAs as there are more objective arguments to be made on that front

I still havent heard a solid one for normalizing FG%
8/12/2008 10:34 PM
naboimp
Posts: 2 (0)
Pro
11
Prospect
Enlighten me, ncmusician. I don't play the baseball sim so I wouldn't know how it is used.

monkee,

First, a point of clarification - when you suggest watching recordings, are you talking about games from the 50's or the 60's. My understanding is that recordings from the 60's are rare and those from the 50's, particularly the early 50's are non-existent. The eras are distinct. I can well believe the 60's were less physical. What I've read suggests the 50's were not though I'll grant you that it is anecdotal evidence.

Actually, the three attempts to make two free throws would have made fouling less appealing, not more. I actually was not basing my statement of the 50's being more physical on the presence of more fouls, but I was constrasting the difference between fouls and defense, both of which I suggested would result from more physical play. Fouls are measurable statistically. Defense is not except in the sense of opponent FG%, which is difficult if not impossible to isolate. I saw the normalization as in part an effort to account for this by adjusting the average FG% for a given group of players for the eras in which they played. (From what ncmusician suggests, this may not be how normalization is being actually implemented.)

Is it justified? I wasn't really taking a stand either way on this. The purpose of my post was to suggest how the FG%+ may be being implemented and why.

Actually, another way of balancing shooting percentages would be to adjust assists for the environment that players actually played in. As I understand it, right now, players get a boost to FG% for the presence of teammates with high assist numbers regardless of what their historical situation was. While I agree that assists have to be beneficial in some manner, this treatment is largely one way and over benefits those players who enjoyed high assist teammates in real life. (They got the benefit in real life in accumulating their stats, and then they get another boost for playing with another high assist player in the sim.)

Ideally, there would be a way to quantify the historical assist environment of the team a player played on. Then, each time the player is on the floor, that historical environment could be compared to the current grouping of players. If it is improved, the FG% should be adjusted upwards. If it has declined, the FG% should be adjusted downwards. Since most 50's players played in a low assist environment, they would generally be benefited to a greater degree. If this was implemented, normalization of FG% would be too much I would agree.

8/13/2008 12:12 PM
tianyi7886
Posts: 4243 (0)
Hall of Famer
51 33 282 40
Prospect
Naboimp, the sim in baseball look at the avg%#, 2b/100#, 3b/100#, hr/100#, obp# etc to determine how a player does. The numbers with # is a normalized stat that stays the same for the player and is normalized against the league avg. Therefore, Arod's monster season from last year might look extremely appealing to use, it's actually not because the modern era is more hitter friendly with higher avg and power numbers. Ofc, Arods numbers still great after normalization, just not as good as it would look straight up.

Therefore, you would want to build a team with players with normalized stats better than their RL numbers.

Normalized stats do not change from game to game. A player who hits .300 in 2007 might get a normalized BA of .294, and it stays the same. Similarly, Bill Russell's .467 Fg% might get normalized to .475 because the league avg in fg% was low that year, but it does not fluctuate from game to game. Rather than thinking you drafted the russel with .467 from the field, think you drafted a russell with .475 from the field.
8/13/2008 3:10 PM
naboimp
Posts: 2 (0)
Pro
11
Prospect
So instead of normalizing to the players actually used, the baseball sim normalizes to a fixed average, representing I suppose average historic conditions (liveliness of ball, size of ballparks, size of strike zone, etc.). I suppose that is simpler overall, but not as appropriate. If you have teams of all current players or teams all of players from the dead ball era, why have them play to some arbitrary fixed average conditions? Then again, even if the sim did do as I was envisioning, I could see where it would make sense to give stats normalized to some arbitrary average just for comparison's sake. Who knows? Perhaps that's even what is really being done in the baseball sim.
8/13/2008 3:39 PM
trtllvr
Posts: 57 (0)
Hall of Famer
8 97
Prospect
How about a "Run up the score" option to use against owners that are really annoying?
8/13/2008 4:17 PM
trtllvr
Posts: 57 (0)
Hall of Famer
8 97
Prospect
(None in this thread)
8/13/2008 4:18 PM
ashamael
Posts: 12930 (0)
Hall of Famer
544 25
All Star
Quote: Originally posted by trtllvr on 8/13/2008How about a "Run up the score" option to use against owners that are really annoying?

lol I LOVE it!!
8/13/2008 4:54 PM
tianyi7886
Posts: 4243 (0)
Hall of Famer
51 33 282 40
Prospect
naboimp, i dont see how its an arbitary average where it is normalized to the league avg in each year rather than a league avg made up by wis.

Wouldn't you agree that batting .350 against Johan Santana in 100 ab is much more impressive than batting .350 against someone like ponson or the terrible Todd Jones? Similarly, isn't it more impressive to shoot 60% and score 50ppg against a defensive beast like bill russell than against a poor defensive C such as the kandiman?

Thats the rationale for normalization and I think it makes alot of sense.
8/13/2008 11:49 PM
◂ Prev 1...6|7|8|9|10...18 Next ▸
Near future plans Topic
△Top△Bottom

Search Criteria

Terms of Use Customer Support Privacy Statement

© 1999-2025 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.