Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

[tbbucsfan001]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

@elknavo I can not seem to get your Draw Foul Tendency to work correctly in Excel. Can you show me a screen shot of the formula?

[The Mac]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Im having trouble here, it could be just because i'm an idiot, but...

For Shot Inside rating your formula is 4*(Inside FG%-30) / 3+60

I'll use Dirk's stats from last year (hoopdata) as an example... he shot 71.2% at the rim, now plugging that into your formula gives you an answer of 114.93...How do you get a rating from that? Is that a 99 rating?

[tbbucsfan001]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Im having trouble here, it could be just because i'm an idiot, but...

For Shot Inside rating your formula is 4*(Inside FG%-30) / 3+60

I'll use Dirk's stats from last year (hoopdata) as an example... he shot 71.2% at the rim, now plugging that into your formula gives you an answer of 114.93...How do you get a rating from that? Is that a 99 rating?

Everything above a 99 is a 99. He explained that the inside rating is messed up somewhere in the first post.

[elknavo]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Dang I loved your rosters last year too bad I got 2k12 on 360 this year

Wow! I never even released any rosters last year and you still loved them? I must be a more talented roster maker than I ever even imagined!

Edit: Whoops I thought that was addressed to me, never mind, I'm dumb lol

[elknavo]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

@elknavo I can not seem to get your Draw Foul Tendency to work correctly in Excel. Can you show me a screen shot of the formula?

Here is the formula copy-pasted from my spreadsheet.
C6= Inside Shot Tendency
C9= Close Shot Tendency
C12= Med Shot Tendency
C15= 3pt Shot Tendency
B18=Draw Foul Tendency

FTA/FG =((0.5*C6+0.5*C6*B18/100)*C6/(C6+C9+C12+C15))/C6+((0.2*C9+0.05*C9*B18/100)*C9/(C6+C9+C12+C15))/C9+((0.2*C12+0.05*C12*B18/100)*C12/(C6+C9+C12+C15))/C12+((0.3*C15+0.075*C15*B18/100)*C15/(C6+C9+C12+C15))/C15

Hope this helps!

Also it won't work it any tendency is set to 0 since it would have to divide by 0... for 0 tendencies just put in 0.000001 or something like that.

[elknavo]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Don't be perplexed. Companies don't care about realistic stats or simulation gameplay. They just want the best they can have while paying the less they can.

And yet the joke's on 2k, as I'd do a much better job than the droolers they have doing it now for next to nothing.

[elknavo]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

I would love to see this done with the classic players

I've started applying my formulas to some classic teams and it's working extremely well so far. A few insights on rating classic players:

-While OffReb% does need to be adjusted for era, DefReb% need not be, because the top-end numbers posted in this area are pretty consistent throughout history, and the overall era differences in DefReb% are much smaller proportionally than the ones in OffReb%.

MINI-GUIDE TO RATING CLASSIC PLAYERS FROM THE 3pt ERA

-For classic players from the 3-pt era, instead of just giving them 99 Inside Shot, I give them whatever 2k gave them +15. This seems to produce a distribution of Inside Shot ratings very similar to the current league.

-The next issue you'll notice with these players is that, like their default current players, midrange shot ratings are generally too high, so drop that by 5.

-Next you want to fix the 3pt tendency. You should set it equal to:
(Inside Tendency + Close Tendency + Med Tendency)*3FGA/(FGA-3FGA)

-If you want to adjust the other tendencies go ahead and do so, but make sure they still add up to the same number or you will have to recalculate the 3pt tendency. Some players will definitely need some edits to Inside/Close/Mid tendency (Dennis Rodman can't get up to his real FG% with his default tendencies unless you give him 80+ shot ratings), but most players' default inside/close/mid tendencies aren't really too far off.

-So now, into your spreadsheet that outputs FTA/FGA and Expected FG% with the formulas from my guide, plug in the shot location tendencies, the 3pt rating , your shot inside rating (2k's +15), 2k's shot close rating, and your shot med rating (2k's-5).

-Compare the FT/FGA to the player's actual FT/FGA from that season and adjust the draw foul tendency until you get a close match. I do it in increments of 5 and it's pretty easy.

-Compare the expected FG% to the player's actual FG% from that season. If it's too low or too high, add or subtract points evenly to the players' close/mid shot ratings (don't add any if the player has 0 tendency for that type of shot though, like mid for some bigs). Once you have got those shot ratings nailed down, subtract 5 from 3pt rating if the player is from '95 to '97, due to the shortened 3pt line those seasons.

-Fix Shot Tendency, Pass, Block, Steal, and Rebound ratings according to my guide. Be sure to adjust offensive rebounds for era.



MINI-GUIDE TO RATING CLASSIC PLAYERS FROM BEFORE THE 3pt ERA

Do everything detailed in the above guide EXCEPT:

-Don't subtract 5 from everyone's midrange rating. This was the heyday of the midrange shot and it was focused on like the three is today.

-For '70's teams, add 1 to everyone's pass rating for every 14 points above 30 in the (AFTER applying the Pass rating formula from the main guide).
-For '60's teams, add 1 to everyone's pass rating for every 7 points above 30. These adjustments are necessary because the requirements for having an assist counted in the old days was more stringent than now, and less assists were awarded per made field goal.


-Obviously you don't have 3pt data, so just leave 3pt tendency as is. When you're adjusting shot ratings to get closer to FG%, you can adjust 3pt along with Close and Mid, since we don't know the actual 3pt%.

-You don't have data for Usg%, so for shot tendency use:

(FGA/36 + 0.5 * FGA/36 - 15) * 2 + 50

-You don't have offensive rebound or reb% data 'til '74. For '71 and '72 use this formula:
5 * (TRB/36 - 2.5) + 25

This will give you what the player's average rebound rating should be. Look at his 2k OffReb/DefReb ratings and adjust them evenly until the average between them is equal to the output of the above equation.

In the 60's rebounds were even more plentiful so we should use
0.45 * (TRB/36 - 3) + 25

-And finally, for players drafted in the 70's and earlier, you should increase height by one inch, because players were measure barefoot at this time rather than in shoes.

Using this method I've re-rated the '96 Bulls and with proper rotations, I'm getting simulated stats incredibly close to the real '96 season.

Here are the ratings I used for the main rotation:

Name Ins Clo Med 3pt Pas Blk Stl ORb DRb OVR
Jordan 99 87 83 84 62 40 87 41 52 99
Pippen 99 91 82 76 68 48 75 45 51 90
Rodman 99 71 59 47 45 40 43 99 99 85
Harper 99 89 81 61 53 45 83 36 37 72
Longley 99 72 68 30 46 84 39 51 53 65
Kukoc 99 96 88 80 62 40 57 46 44 77
Kerr 90 84 92 99 51 26 59 25 28 69
Wennington 95 85 84 37 40 46 45 46 47 59
Buechler 99 78 79 87 46 36 73 49 41 63
Simpkins 97 52 41 36 41 39 39 70 53 57

And Tendencies:
Name Sht Ins Clo Med 3pt DrawF
Jordan 77 39 100 99 39 68
Pippen 59 50 46 70 82 0
Rodman 31 28 10 4 4 0
Harper 40 30 18 24 19 0
Longley 46 28 10 23 0 0
Kukoc 53 32 33 39 40 0
Kerr 36 9 10 27 44 0
Wennington 43 20 15 29 0 0
Buechler 45 20 10 27 34 0
Simpkins 43 37 2 0 0 0

Anyone on PS3 or 360 looking to "test out" my formulas can try these numbers out.
Other than certain unsolvable issues with the engine noted in my original guide(i.e.player and team offensive rebounding, too many FTA for lowest FTA players)you should get stats that are pretty damn close to that real Bulls campaign. The only issue I've found is Jordan tends to shoot about 1.5% greater FG% on average and I'm at a loss to explain why. Other superstars with really high ratings seem to go as expected, but Jordan shoots from every range as if he's rated about 5 points higher. The Close/Med ratings posted for him are 5 points below what my formula would give but result in realistic stats. Jordan appears to be a unique special case- I believe it's due to his hidden Shot Off Dribble rating of 97, which is 12 points higher than any current NBA player.

[Tomba]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

I like this thread it's like

[socharliemac]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Since you came up with all these good formulas do you have a good formula for the coach profiles, I been trying to use stats off of hoopdata and basketball-reference. I been trying to come up one with using the teams teams pace, team shots locs, Off eff, or what ever I could use. I'm not sure if it'll work but trying couldn't hurt, I just won't the CPU to stop wasting the whole shot clock down and start shooting more.

[elknavo]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

socharliemac, sounds to me like you're talking about gameplay. I only deal with simulated games- I think the gameplay is unsalvageable due to the complete lack of basketball basics like double teams, offensive fouls, and a realistic number of offensive rebounds.

[Po Pimp]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

I've read your formulas and it looks like good research. However, you didn't take into account how a players vertical affects their performance. I've done my own research, and while it doesn't have much of an effect on blocks on default rosters, it definitely affects the rebounds.

One thing I'm interested in, is if you lower certain players vertical as a whole, would it increase the blocks in this game. For example, Kobe is 33 years old with an 83 vertical. Let's say you drop his vertical down to 60 (he's still 6'6") and just overhaul the whole vertical scale, while leaving athletic bigs with a decent vertical and block rating...wouldn't those block numbers increase? Just asking...

[tbbucsfan001]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Here is the formula copy-pasted from my spreadsheet.
C6= Inside Shot Tendency
C9= Close Shot Tendency
C12= Med Shot Tendency
C15= 3pt Shot Tendency
B18=Draw Foul Tendency

FTA/FG =((0.5*C6+0.5*C6*B18/100)*C6/(C6+C9+C12+C15))/C6+((0.2*C9+0.05*C9*B18/100)*C9/(C6+C9+C12+C15))/C9+((0.2*C12+0.05*C12*B18/100)*C12/(C6+C9+C12+C15))/C12+((0.3*C15+0.075*C15*B18/100)*C15/(C6+C9+C12+C15))/C15

Hope this helps!

Also it won't work it any tendency is set to 0 since it would have to divide by 0... for 0 tendencies just put in 0.000001 or something like that.

Thanks, man. I was putting in the "All Tendencies" part wrong.

[chinatownpwnage]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

what is your thoughts on potential elknavo?

[Darkh]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Since your a pc modder, you could use tool like cheat engine or art money, i think that the maximum allowed value for shot inside is around 110, this might work, but the game may just equate it down to 99. Worth a try i think

[iTofu]

Re: Comprehensive Simulations Stat Mechanics Guide - ALL ROSTER MAKERS SHOULD READ

Incredible post. I like your approach, results and summary. When reditor is finally released for 360, I'll be definitely trying my hand at using your formulas.