Simulating statistics in NHL2K

[nikethebike]

Quote: Originally Posted by nikethebike

I realized I had a bug where the editor did not write back the checking tendency, but instead it wrote the checking stat as tendency. After changing this, which actually made most players have much lower checking tendency, these were my top scoring defenders: 1. Hedman, TBL: 19+53=72 2. Karlsson, SJS: 24+45=69 3. Rielly, TOR: 21+48=69 4. Carlson, WSH: 20+47=67 5. Giordano, CGY: 20+45=65 6. Makar, COL: 20+43=63 7. Hamilton, CAR: 22+35=57 8. Dunn, STL: 20+37=57 9. Pietrangelo, STL: 19+38=57 10.Burns, SJS: 21+34=55 Now this is much more what I was aiming for points wise! Still no defender on the top 50 list though, Hedman just missed out on fewer goals...

I think I need to give up on the hits and blocked shot stats. Messing with these which also included aggression and composure made the total scoring drop to where the top point scorer had like 94 points and it really didn't make a real difference to how many blocks and hits that were simulated.


I could probably make it a little better but there would be a couple of consequences, one being I'd need to redo the entire rating system to make up for the lost point production. Second is that the stats would still be so off that I don't think it's possible to get it even close anyway. It's just not worth the effort. Believe me I tried.



I had some good results though with changing stamina, this made forwards get more realistic ice time.


I would still like to make defender scoring more consistent and the total scoring a little bit greater but right now I have a roster file setup that is behaving pretty realistic.


I need to boost some individual players in order to get better default lines and I think in the long term I will also need to reintroduce multiple positions so that second line centers can end up on the first line powerplay etc. Malkin really suffers from playing behind Crosby and I've made Draisaitl and Point wingers to make them consistent scorers.


Edmonton will probably need to be boosted, especially Draisaitl but I suspect the entire team needs to be better for him and McDavid to start dominating the charts like they've done these last two seasons.


Chicago are too good. They get to the playoffs every save and I don't think they are at that level right now. Crawford was boosted too much in my opinion by the goalie attributes I imported from that other statistics thread.


I'd also like to figure out what really decides the goal scoring amount. I have a hunch about it actually being the offensive/defensive stats the game presents in exhibition that can have an impact. When me lowering hit tendency which make defenders generally weaker made defenders actually score more points it got me thinking that it might just be that higher overall defenders compared to forwards mean less goals even if the stats boosted for defenders are the offensive stats, which is basically what I have done. But it could also be that hitting tendency is a defensive attribute which make it harder for defenders to drive the offence.


It's been kind of fun to do this analysis but I think it is about time to wrap this up and move on to other things soon.

[nikethebike]

Man I was in the zone today...



Wrote code to change the overall of players dynamically so now I am finalizing a tool that sets overalls for players based on a depth chart. The overall calculation is way off right now though but it is good enough so that it sets overalls so that players end up on lines similar to the depth charts.


I also used this tool to recalculate keepers globally and I now know that the easiest way to change amount of goals is to make the goalies worse. It's either this or to make the shot tendency globally higher. I even made a global edit where I had players perform Gretzky-like stats so i think it could be possible to setup an 80s scoring league if one wanted. Too many teams though.



I will try to tweak the scoring using this knowledge.


It's really fun to use statistics and data about players to set attributes.
There are a bunch of not so easy to set variables but I am thinking of perhaps letting size (weight/height) have effect on agility and perhaps also acceleration, balance and strength.



Would make sense lighter/smaller players have better agility and can get up to speed faster. Perhaps they should also hustle more. Strength and balance could also be effected by this and be higher for heavier players. Could be a good way to get differences in stats since HUTDB/NHL20 seem to have spd/agl/acc pretty much similar for all players. I could make a boost/decrease here based on the real life data.



Courage can be based off hits and blocked shots. Problem is I don't know what the attribute is for though really...

Aggression can be based off hits and PIM.
Discipline can be based on PIM reversed as in the less amount of PIM the higher the discipline.
Clutch I base on GWG.
Ice time is also a factor you can use in many combinations as a better player will get more ice time.


Well I am trying a lot of stuff and getting interesting results such as by mistake putting Draisaitl on the second line behind Ennis as a LW and many of his stats dropped so much that he was put on the third line but he was still on the first powerplay putting up some points...


I would like to achieve default lineups that have powerplays with a forward on the defense some way. Don't know if it is possible.

[nikethebike]

The player types are really messing things up for me when it comes to line generation and overalls. Playmakers are constantly underrated if they aren't superstars. Im really tempted to just set everyone as two-way players in order for the overalls to be calculated the same way.

[nikethebike]

I managed to generate a save where Ty Barrie had 83 pts and Karlsson 78 and Burns 75...


Still tweaking...

[nikethebike]

Man is it tough to get defenders to have many shots but low sh%...
It is a lot of trial and error and so many parameters effect eachother. I kind of understand why EA in their old games took the "we use the old stats as base for simulation" approach. :P


As it looks right now D-men will end up with pretty low shooting stats. I will need to see how these run in a match.

[nikethebike]

http://www.naturalstattrick.com/


After finding this site, with all the stats it has... I might have to rewrite my stat to attribute conversion.

[nikethebike]

I know I'm mostly talking to myself these days but I'm going to keep documenting my findings here.


I simmed a bunch of franchise seasons with the latest released roster and realized that there were almost no missed games for any players. However generated rookies missed games and I realized they had much lower durability so I looked at the durability stat. Instead of having it in range of 65-95 I changed it to 50-85. Now there are much more injuries and missed games.
Still some teams can go a full 82 game season with the same 6 defenders though so perhaps I need to lower this even more. Perhaps the maximum should be even lower to allow for even more random injuries like what can happen in real life.



The interesting thing is that this actually seem to have increased scoring averages which make sense because more players will get chances on better lines during the season.


Had a couple of players with 120 points and Makar finished a season with 84 points and 60+ assists.


This makes for a more dynamic franchise mode.

[tinpanalley]

Quote: Originally Posted by nikethebike

I know I'm mostly talking to myself these days but I'm going to keep documenting my findings here.

You're not talking to yourself. If it's NHL2K related, I'm interested. To be honest, I only lose interest after 2K10.