Are there hidden intangibles in the engine that make a player better or worse than they look on paper? 
I think there are based on a few things:
1)  I have players with lesser stats across the board who consistenly (over 3-4 years) outperform teamates with better stats.
2)  Comparable players in the same position can have strikenly different stats on the same team
3)  The scouting searches sometimes say things like:  "very hard to stop inside" and you look at their core stats for inside and they are average at best.  Then the player seems to do pretty well at the next level.

If your talking about the H.S. stats of a recruit they are useless except the free throw percentage. If your just talking about ratings then other things matter as well when it come to stats. For instance who you play against, how they gameplan and IQ's. All those things could lead one player to being a little better then another when you expect different.

Along with that I do believe there is some randomness to the engine so when nothing else can be explained sometimes it's just the randomness and nothing else.

admin has consistently said that there are no hidden attributes.

So that, when the games are simmed,  the data that is used to feed the sim engine and ultimately generate a playbyplay is just the player ratigns that you see.

So,  why does one player consistently shoot a better percentage, or grab more rebounds, or have fewer turnovers than a comparable player?
well, sometimes there are reasons, such as the player's teammates, or whether or not he is a starter, or the player's complimentary ratings (for instance his ATH will impact his rebounding totals), or, if the compared players are on two different teams,  the SOS comes into play.  anyway,  you get the idea.  there are alot of reasons why a seemingly weaaker player can outperform a seemingly stronger one.

Other times,  as hard as we may look,  no explanation can be found.  in those cases,  one possibility is that random variation has favored the weaker player due to luck.  for instance.  lets say the ratings would suggest that playerA would have 6 to 12 rebounds per game.  and playerB would have 4 to 10 rebounds per game.  based on everything we can see,  playerA should be the better rebounder.  but sometimes chance will favor playerB such that he will average 8 rebounds per game and playerA will average 7.

Although, i must admit, i  find that the random variation explanation  usually does not fit the facts for me.
What i mean is this:

lets suppose in the case above that playerB averages 8 boards per game over the first half of the season,  while playerA averages 7.   if we really believe the randomness theory, we would bet that playerA would turn things around and outrebound playerB over the second half of the season.  but there have been many times when i see this kind of thing,  but i have a feeling things will not turn around,  and, in fact , they dont.

so,  i can tell you what admins have consistently said ----- "there are no hidden ratings"
but i can also tell you my intuition says maybe there is something behind the ratings that we cant see,

Posted by oldave on 7/21/2010 9:57:00 AM (view original):
admin has consistently said that there are no hidden attributes.

So that, when the games are simmed,  the data that is used to feed the sim engine and ultimately generate a playbyplay is just the player ratigns that you see.

So,  why does one player consistently shoot a better percentage, or grab more rebounds, or have fewer turnovers than a comparable player?
well, sometimes there are reasons, such as the player's teammates, or whether or not he is a starter, or the player's complimentary ratings (for instance his ATH will impact his rebounding totals), or, if the compared players are on two different teams,  the SOS comes into play.  anyway,  you get the idea.  there are alot of reasons why a seemingly weaaker player can outperform a seemingly stronger one.

Other times,  as hard as we may look,  no explanation can be found.  in those cases,  one possibility is that random variation has favored the weaker player due to luck.  for instance.  lets say the ratings would suggest that playerA would have 6 to 12 rebounds per game.  and playerB would have 4 to 10 rebounds per game.  based on everything we can see,  playerA should be the better rebounder.  but sometimes chance will favor playerB such that he will average 8 rebounds per game and playerA will average 7.

Although, i must admit, i  find that the random variation explanation  usually does not fit the facts for me.
What i mean is this:

lets suppose in the case above that playerB averages 8 boards per game over the first half of the season,  while playerA averages 7.   if we really believe the randomness theory, we would bet that playerA would turn things around and outrebound playerB over the second half of the season.  but there have been many times when i see this kind of thing,  but i have a feeling things will not turn around,  and, in fact , they dont.

so,  i can tell you what admins have consistently said ----- "there are no hidden ratings"
but i can also tell you my intuition says maybe there is something behind the ratings that we cant see,

Your 2nd to the last paragraph is wrong mathmatically.  Independent trials.  If you flip a coin and get 100 heads in a row, your chance of getting a head on the 101st flip is still 50%. 

. . probability does not work that way.  The 'law of averages' is a statistical fallacy, the same one that you coudl call 'The Gambler's ruin"

http://www.google.com/search?sourceid=chrome&ie=UTF-8&q=law+of+averages+fallacy

since snakes had legs, admin has said there are no hidden variables

on the other hand, that does not mean that there are not odd, non-intuitive effects and interactions of variables - does weak ballhandling and passing by some team's bigs cause their guards to shoot a lot less well than one would think from their PER, IQ, ATH and SPD?  does great ATH and SPD and a clever -2 setting make a guard who has crazy low PERI an efficient scorer and shooter if measured by FG%?

who knows

no hidden attributes (cue tmac telling me i am wrong)     

Are hidden intangibles a double negative?

Fellas, Fellas!  slow down and read.

if you read closely,  you would realize that i used an assumption that player A was a better rebounder.  All i was saying is that if our original assumption was correct (playerA was better rebonder)  then we would expect him to outrebound playerB during the second half just the same as we would have expected him to outrebound B  during the first half or any other half.  Im not saying he will gain back the rebounds lost, just that he will again have a good chance to do what we expected him to do in the first place (unless, of course, it turns out that due to hidden or unknown attributes he is really not the better rebounder)


in other words,  here is the litmus test ----
playerA - 90RE, 60ATH , 7.0RE per game through game 13
playerB - 80RE, 50ATH, 8.0RE per game through game 13

given that info,  and if we could somehow assume that all variables are equal (I know they arent... but lets pretend)
If I asked you to bet on which player will have the most rebounds per game from here on out.... who would you bet on?
Would you bet on playerA becuase the ratings say he is a better rebounder?
or
would you bet on playerB because he has proven himself bettter so far?


am i making more sense now or do i get another lesson in statistical probability?

Yes you woudl bet on it. . but just because it doesn't happen doesn't make the results automatically suspect.

huh?

so, i pretty much agree with oldave's first post. anyway, in regards to the part where he says, well, sometimes there is no logical explanation so we have to point to stat variance... i think that is part of it. what is the other part, hidden attributes? more likely, what is hidden is, the details of the sim engine implementation.

so how about something like this. you have guard A, who is a pimp mac daddy. you have guard B, who is still great, but without the flare. other teams have a pimp daddy defender, but their second defender is suspect. their A defender goes after your A guard and puts him in his place (those flashy guards never live up to the bill anyway, against a great defender :), while guard B gets to abuse his counterpart. Guard A and B are in the same class, so this continues for 4 years. its just that in say soph year, its the best backup defender on A, and the pathetic worse second string backup on B. And in senior year, its the best defensive player on A and the other, inferior starting guard on B. but consistently, guard B is the beneficiary of a weaker defender. so something like that can explain consistent and repeated poor performance from a player. on a side note, similar logic also goes a good way to explain the dreaded senior slump.

A few quick thoughts:
1)  Disclipline/personality used to be a part of the game as far as I understand.  It was appearantly removed, but vestiges of it remain (pyschological testing, and quotes on character from the recruit and in evals).  Perhaps, a few stray bits of code remain in the game engine that still takes into account a player's discipline.  Thus, a loose cannon/party type player may have more bad games than their attributes would suggest.

2)  Even though admins say there are no "hidden" attributes, maybe we take them too literarly.   Conventional wisdom says that the eval notes such as - "teen wolf"/shoots a lot; "found a $20 bill and turned it in"; "supports his teammates even on the beacnh" - are meaningless.  Maybe they are not.  Maybe the provide clues to how a player will affect his teammates around him by making everyone better or worse by how he plays.   Thus, while it's not a metric people use to judge a player, it is not a truly hidden characterisitc becuase we receive information on it.     I can see the stuff that has no bearing on basketball (Ladies man; finding the money) being filler, but the stuff that is basketball related (ball hog; team player) might have more value than we currently put on it.  The admins went to the trouble of including dialogue on it for a reason.

3)  During testing, admin refused requests to allow us to sim the same game multiple times in a row to see how minor changes affected the game against the identical opponent.  Admin said they didn't want us to learn more about how the engine worked.  Maybe we would have discovered a factor/variable that isn't obvious on first appearance

4)  Perhaps the "randomness" admins talk about when explaining a result is actually a hidden factor that they don't want us to know.  Otherwise, the game becomes a math problem where the person with the best spreadsheet and equation can win by figuring out a winning formula.  Or perhaps, the randomness isn't as random as they state because the code is written that way to have a predisposition to certain events.

Posted by dukenilnil on 7/21/2010 5:42:00 PM (view original):
A few quick thoughts:
1)  Disclipline/personality used to be a part of the game as far as I understand.  It was appearantly removed, but vestiges of it remain (pyschological testing, and quotes on character from the recruit and in evals).  Perhaps, a few stray bits of code remain in the game engine that still takes into account a player's discipline.  Thus, a loose cannon/party type player may have more bad games than their attributes would suggest.

2)  Even though admins say there are no "hidden" attributes, maybe we take them too literarly.   Conventional wisdom says that the eval notes such as - "teen wolf"/shoots a lot; "found a $20 bill and turned it in"; "supports his teammates even on the beacnh" - are meaningless.  Maybe they are not.  Maybe the provide clues to how a player will affect his teammates around him by making everyone better or worse by how he plays.   Thus, while it's not a metric people use to judge a player, it is not a truly hidden characterisitc becuase we receive information on it.     I can see the stuff that has no bearing on basketball (Ladies man; finding the money) being filler, but the stuff that is basketball related (ball hog; team player) might have more value than we currently put on it.  The admins went to the trouble of including dialogue on it for a reason.

3)  During testing, admin refused requests to allow us to sim the same game multiple times in a row to see how minor changes affected the game against the identical opponent.  Admin said they didn't want us to learn more about how the engine worked.  Maybe we would have discovered a factor/variable that isn't obvious on first appearance

4)  Perhaps the "randomness" admins talk about when explaining a result is actually a hidden factor that they don't want us to know.  Otherwise, the game becomes a math problem where the person with the best spreadsheet and equation can win by figuring out a winning formula.  Or perhaps, the randomness isn't as random as they state because the code is written that way to have a predisposition to certain events.

this is quite a bit of wordiness. anyway:
1) there are non rating factors that influence a players' decision, but supposedly, only outside of game situations. could there be bits of code left? after the sim engine rewrite, i would bank on not.

2) admin's don't say there are no hidden attributes. they say, the only thing that goes into the outcome are the ratings and settings (including hca and home/awayness).

3) seble was absolutely, unquestionably correct not to allow us to sim the same game multiple times. the reason being, you could break down uncertainty to an unacceptable level, i mean seriously some of us would have taken the sim engine outline (the steps that show the decision making process) and assigned actual formulas to those steps. it would jeopardize the integrity of the game and give a ridiculous advantage to those with the mathematical and analytical ability to exploit the opportunity. things like, for a guard at 50 ath 90 spd 60 def 80 per/bh/pass, how many points of ath is 1 spd would become facts, not opinions, and that is a terrible thing.

4) the randomness is the product of the game being a series of a relatively small number of coin flips (not even weighted coins). for example, there might be 800 coin flips in a game, and if you expect to win 420, its still pretty likely you would win 399 or less. on the subject of a hidden factor, dubbed "upset factor" by some users, both admins deny its existence. however, something as simple as, randomizing the game-long effect of HCA is mathematically equivalent to a hidden upset factor. so personally, i am very open to the idea an upset factor does in fact exist. i do not think it is coded as such, but that really doesn't matter from a user's perspective.

my  conspiracy theory is that the more time you spend on the site, and the more pages you view in a single day, the better your team will do.