August 28 release - engine changes Topic

mamxet
Posts: 1154 (0)
Hall of Famer
215 48 26 2 436 11
Rookie
  • New engine version with the following tweaks:
    • reduced single-game variance on rebounding
    • reduced single-game variance on shooting
    • reduced single-game variance on fouls
    • reduced single-game variance on turnovers
    • reduced single-game variance on free throw shooting
    • changed the performance hit from fatigue so that it won't be as linear, but instead performance will decrease more rapidly as a player gets tired
    • reduced offensive rebound odds on free throws
    • various minor logic improvements

    The end result should be that single-game results match more closely to what you would expect and there will be fewer upsets. There will still be upsets though, so don't expect to win every game you're favored to win.

    This will affect all worlds immediately.

8/28/2010 2:10 PM
mamxet
Posts: 1154 (0)
Hall of Famer
215 48 26 2 436 11
Rookie
what does "reduced single game variance" really mean?  does it mean that possessions are no longer independent events and that - for example - if a guy who should hit 80% of his FTs hits 8 in a row that the 9th will not have an 80% chance?

if it doesnt mean that, what could, what does it mean?
8/28/2010 2:12 PM
rednu
Posts: 2673 (0)
Hall of Famer
185 2 5
Pro
I think it refers to situations where Team A and Team B played one game and Team A won by 40 points, then met three nights later for the return conference game and Team B won by 20 points.

Based on previous discussions, I'm interpreting that line to mean that the game-to-game swing of two teams playing each other more than once won't be as radical as it has been previously. 
8/28/2010 2:17 PM
tmacfan12
Posts: 336 (0)
Hall of Famer
356 4 2
Prospect
Right before Knight Post Season Tournaments... this could be great or terrible
8/28/2010 2:18 PM
seble
Posts: 880 (0)
Hall of Famer
24 44 35 306 121 185
Prospect
It basically means what mamxet suggests.  Within a single game, outcomes to the various listed events will no longer be completely independent of previous outcomes.  It's an attempt to combat the inherent variance of a small sample size (i.e. one game).  The end result of that should also do what rednu mentions, limit the swing in outcomes in games between the same teams, assuming all the inputs (player ratings, game plan, etc.) are the same.
8/28/2010 2:33 PM
seble
Posts: 880 (0)
Hall of Famer
24 44 35 306 121 185
Prospect
Posted by tmacfan12 on 8/28/2010 2:18:00 PM (view original):
Right before Knight Post Season Tournaments... this could be great or terrible
This was the best time to release it, where no worlds are currently in the postseason.  You probably won't even notice a difference right away, but in the long run you should see more consistent results.
8/28/2010 2:33 PM
oldresorter
Posts: 646 (0)
Hall of Famer
1110 1
Prospect
Posted by seble on 8/28/2010 2:34:00 PM (view original):
Posted by tmacfan12 on 8/28/2010 2:18:00 PM (view original):
Right before Knight Post Season Tournaments... this could be great or terrible
This was the best time to release it, where no worlds are currently in the postseason.  You probably won't even notice a difference right away, but in the long run you should see more consistent results.
funny you mention that, I was going to chime in, the change is probably working best if it is somewhat difficult to notice,

a somewhat related question, to you compare real life variation among things like final score,  reb, assist, to's, shooting % to that in the HD game?
8/28/2010 2:37 PM
dahsdebater
Posts: 10043 (2)
Hall of Famer
284 292 18
Pro
I'm not sure I like the idea of having separate game results not be calculated independently.
8/28/2010 3:30 PM
reinsel
Posts: 927 (0)
Hall of Famer
1 130
Rookie
We'll have to see as always, but I think a tiny bit more predictability is a good thing.
8/28/2010 3:34 PM
seble
Posts: 880 (0)
Hall of Famer
24 44 35 306 121 185
Prospect
Posted by dahsdebater on 8/28/2010 3:30:00 PM (view original):
I'm not sure I like the idea of having separate game results not be calculated independently.
We're talking about events within a single game, not completely separate games.  In other words, if John Smith has shooting odds of 45% normally, but is 0/6 in the game, his odds going forward will increase above 45% until he reaches that 45% for the game.  The same would happen the other direction if he were 5/6.
8/28/2010 3:43 PM
Weena
Posts: 579 (0)
Hall of Famer
227 1 1 1 531 1 27
Rookie
Will the overall season results still match up with real life NCAA percentages as they have in the past?
8/28/2010 3:51 PM
cornfused
Posts: 1492 (0)
Hall of Famer
2 235 11
Prospect
Posted by seble on 8/28/2010 3:43:00 PM (view original):
Posted by dahsdebater on 8/28/2010 3:30:00 PM (view original):
I'm not sure I like the idea of having separate game results not be calculated independently.
We're talking about events within a single game, not completely separate games.  In other words, if John Smith has shooting odds of 45% normally, but is 0/6 in the game, his odds going forward will increase above 45% until he reaches that 45% for the game.  The same would happen the other direction if he were 5/6.
So if an opposing team shoots 60% from 3 in the first half, they'll shoot worse in the second half... AND the engine will adjust us to guard against the 3?




Honestly, I'm not sure I like the idea of events not being random.
8/28/2010 3:59 PM
mccabemi
Posts: 65 (0)
Hall of Famer
45
Prospect
I don't see this turning out well.   Upsets are good for the game and the extent to which they were occuring was way overblown by a few. 
8/28/2010 4:23 PM
professor17
Posts: 366 (0)
Hall of Famer
222 13
Rookie
Posted by seble on 8/28/2010 3:43:00 PM (view original):
Posted by dahsdebater on 8/28/2010 3:30:00 PM (view original):
I'm not sure I like the idea of having separate game results not be calculated independently.
We're talking about events within a single game, not completely separate games.  In other words, if John Smith has shooting odds of 45% normally, but is 0/6 in the game, his odds going forward will increase above 45% until he reaches that 45% for the game.  The same would happen the other direction if he were 5/6.
I don't like this at all. Each event should be independent, not based on the outcome of prior events. You are forcing outcomes. Isn't there some other way to decrease standard deviations without artificially forcing a reversion to the mean? 
8/28/2010 4:33 PM
antonsirius
Posts: 2567 (0)
Hall of Famer
167 135 9 20
Prospect
Color me very nervous about this.
8/28/2010 4:39 PM
1|2|3...11 Next ▸
August 28 release - engine changes Topic
△Top△Bottom

Search Criteria

Terms of Use Customer Support Privacy Statement

© 1999-2026 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.