Have we Exhausted V1?

[Beketov 9,025]

After a week off because I didn't need the TPE (I don't remember how exactly) I'm back again bitches! This week in my saga of complaining about things I turn away from what we used to have and more towards "why the fuck have we never done anything about this". It also kinda ties into my last topic about diversity being non-existent. So let's dive in shall we?

 

For those we don't know, we still use V1 of STHS. On first glance this seems probably fine until you realize that V3 is the most current version of the engine. Now I don't know for sure what has changed in the subsequent versions aside from the fact that I know V2 added a bunch of extra stats; specifically stats that relate to defensive play much better than hits and shots blocked do. I'm also guessing that the newer versions make shit way less random, especially in the playoffs, but that is a guess.

 

Now, this isn't a licensing thing. In fact there's only 1 version of the actual program that can be downloaded which is V3. You simply tell it what engine you want to use, in our case V1. So we could literally swap this at any time but we haven't. We've tested it before and found that, if I recall correctly, goalies get completely and utterly fucked or become completely dominant, I don't remember which one. Point is scoring was a problem, mostly because we have stats in the high 90's for the most part. We messed with the sliders but couldn't get them to a normal level.

 

So, what does this all mean? Honestly nothing but it's the topic I came up with given how random the playoffs are seeming to be with Seattle currently tied with Calgary which makes no sense at all. I don't know if the new engine would change this but maybe it's time to test it out again? It would be interesting to see at the very least and I think it's likely worth testing again. The league can grow as much as we want on the board and the portal and what-not but if we remain on the same engine forever our back-end ultimately will never change and that could be a problem.

 

So, thoughts? Should we try it again (in testing purposes only, at most during like a WC, never a full season) or should we just accept that these are the stats we have and this is how our engine works? Weigh in now.

[Enorama 2,038]

Is there any way to load the current database into V3 and run them side by side for a year and see how drastic the differences are? I know this likely means a ton more work for simmers, but it would make the discussion much easier if we have an apples-to-apples comparison across an entire season as opposed to small tournaments that might see different results from the regular season anyways.

[Beketov 9,025]

5 minutes ago, Enorama said:

Is there any way to load the current database into V3 and run them side by side for a year and see how drastic the differences are? I know this likely means a ton more work for simmers, but it would make the discussion much easier if we have an apples-to-apples comparison across an entire season as opposed to small tournaments that might see different results from the regular season anyways.

Yes and no. You can't sim 2 engines at the same time but you can open the same database into both engines. It would involve doubling the workload though as we would have to sim literally everything twice.

 

What would be doable though, and much faster, is doing a straight comparison in one go. So instead of simming an entire season twice over the course of the full season, we could run a test season using V1 and a test sim using V3. Difference between this and the full test would be that lines and such would be the same the entire season but for the sake of comparison that's fine because they're both the same which is what matters. It would also be very clear that these results are purely testing and have no bearing on the actual league. Chances are we wouldn't upload individual games just to make that very clear but the stats would still be present.

[diamond_ace 3,112]

I vote yes. TA and GA!

[Quik 4,113]

I wouldn't mind having a test season run on each of the engines. Would be interesting to see what the results are, and make an informed decision rather than just going off that one time we used it for the World Cup however many seasons ago. If we can have a direct comparison (maybe even run 2-3 seasons with all rosters being the same in each sim if possible), it'd be a lot easier to say "Yeah, let's make the switch" or "No, that shit sucks".

[Beketov 9,025]

3 minutes ago, diamond_ace said:

I vote yes. TA and GA!

I'm completely blanking on what those are.

[diamond_ace 3,112]

5 minutes ago, Beketov said:

I'm completely blanking on what those are.

takeaways and giveaways

[Beketov 9,025]

1 minute ago, diamond_ace said:

takeaways and giveaways

Ah, yes, of course.

 

8 minutes ago, Quik said:

I wouldn't mind having a test season run on each of the engines. Would be interesting to see what the results are, and make an informed decision rather than just going off that one time we used it for the World Cup however many seasons ago. If we can have a direct comparison (maybe even run 2-3 seasons with all rosters being the same in each sim if possible), it'd be a lot easier to say "Yeah, let's make the switch" or "No, that shit sucks".

If @Will is cool with me trying I could probably run a test season or 2 based on current setup and obviously just not save the files to dropbox so nothing gets over-written.

[Quik 4,113]

2 minutes ago, Beketov said:

If @Will is cool with me trying I could probably run a test season or 2 based on current setup and obviously just not save the files to dropbox so nothing gets over-written.

[Will 4,660]

Small detail, but there isn't actually a V3 engine. V3 is just the STHS program itself. 

 

The engines are 1.1, 1.5, 2.0, or 2.1 (we use 1.1):

 

Spoiler

Version 1.1

Advantage:

• Available in the STHS since February 2008 in version 1.1.
• Results are proven to be realistic.

Disadvantage:

• New features since 2010 are not added in the code (Referee/Lineman, Line/Team strategies, Percentage of ice time per line does not apply, etc).
• Some stats are not count and will always show at 0 in the simulator
• A lot of new functionalities are not implemented (For example, 3 vs 3 overtime)

Version 1.5

Advantage:

• Available in the STHS since February 2008 in version 1.1.
• Results are proven to be realistic.
• Some of the new simulation features are added.

Disadvantage:

• Some of the new simulation features are not added in this version of the simulation engine.
• Some stats are not count and will always show at 0 in the simulator.

Version 2.0

Advantage:

• Always has the new simulation features.

Disadvantage:

• The results are not easy to reproduce, somehow “random”.
• Top players stat will sometime behave like normal player stat.

Version 2.1

Advantage:

• Always has the new simulation features.
• Top players stat will be like top player stat.

Disadvantage:

• Fairly new (September 2011) so the results are not proven to be realistic yet.
• Top players stat difference versus the rest of player’s stat can be impressive.

 

4 minutes ago, Beketov said:

If @Will is cool with me trying I could probably run a test season or 2 based on current setup and obviously just not save the files to dropbox so nothing gets over-written.

 

You certainly can, but just know..I've tried it several times and I've never been able to get consistent reasonable results  1.5 isn't THAT bad, but almost none of the features (like TA/GA) are included in 1.5 either so it's pointless to switch to that one. 

 

My testing and research on the STHS site back when I was trying to do 2.0 or 2.1 led me to discover that in the V2 engines, player attributes have A LOT more weight than the league sliders versus in V1. So the reason we have insanely high scoring on those engines is because VHL players tend to have high attributes (relative to say, an NHL rating file that the engine was probably designed around). I did manage to get more reasonable results when i significantly reduced everyones passing and scoring attributes (i did it by a percentage but can't remember what %). That was all before the new update scale(s), but I'm not sure if the scale changes would be enough to bring scoring down to reasonable levels.

[Beaviss 4,957]

Whats the problem with changes in player performance? Isn't that the whole point of changing the engine? I for one would like a hard switch to 2 and the players would adapt.

[Beketov 9,025]

9 minutes ago, Will said:

Small detail, but there isn't actually a V3 engine. V3 is just the STHS program itself. 

I feel like I knew that but forgot when writing this.

 

10 minutes ago, Will said:

You certainly can, but just know..I've tried it several times and I've never been able to get consistent reasonable results  1.5 isn't THAT bad, but almost none of the features (like TA/GA) are included in 1.5 either so it's pointless to switch to that one. 

 

My testing and research on the STHS site back when I was trying to do 2.0 or 2.1 led me to discover that in the V2 engines, player attributes have A LOT more weight than the league sliders versus in V1. So the reason we have insanely high scoring on those engines is because VHL players tend to have high attributes (relative to say, an NHL rating file that the engine was probably designed around). I did manage to get more reasonable results when i significantly reduced everyones passing and scoring attributes (i did it by a percentage but can't remember what %). That was all before the new update scale(s), but I'm not sure if the scale changes would be enough to bring scoring down to reasonable levels.

I believe we have talked about it before and you have tested it quite exhaustively but it would be interesting to at least have the numbers for everyone to see and easily laid out. I do recall scoring being insanely high even with the slider being placed at absolute minimum for scoring. A fully laid out test would at least give us a starting point to decide if the features are worth having to likely completely re-work how high are attributes can easily become right now. I doubt the new scale will make a huge difference because it helps a bit in this case but not mostly just made up for the increase in weekly TPE.

[Beketov 9,025]

3 minutes ago, Beaviss said:

Whats the problem with changes in player performance? Isn't that the whole point of changing the engine? I for one would like a hard switch to 2 and the players would adapt.

It's not a decrease in player performance (aside from goalies) it's a major increase. So there's no adapting, there's just a massive increase in scoring with goalies all having like 3 GAA.

[Beaviss 4,957]

3 minutes ago, Beketov said:

It's not a decrease in player performance (aside from goalies) it's a major increase. So there's no adapting, there's just a massive increase in scoring with goalies all having like 3 GAA.

Turn the scoring slider down a little? that might fix it

[Beketov 9,025]

Just now, Beaviss said:

Turn the scoring slider down a little? that might fix it

That's with the scoring slider at minimum based on our old tests. Hopefully I can figure something out in new ones but, as @Will said, the sliders were given less of an effect compared to the attributes.

[Enorama 2,038]

1 hour ago, Beketov said:

Yes and no. You can't sim 2 engines at the same time but you can open the same database into both engines. It would involve doubling the workload though as we would have to sim literally everything twice.

 

What would be doable though, and much faster, is doing a straight comparison in one go. So instead of simming an entire season twice over the course of the full season, we could run a test season using V1 and a test sim using V3. Difference between this and the full test would be that lines and such would be the same the entire season but for the sake of comparison that's fine because they're both the same which is what matters. It would also be very clear that these results are purely testing and have no bearing on the actual league. Chances are we wouldn't upload individual games just to make that very clear but the stats would still be present.

 

That's more or less what I meant. I just think it would be best to compare how they stack up, like what happens to point totals, sv%, maybe point distribution, etc. between the two, and that would help make an educated decision on which one to use going forward.

[Spade18 1,305]

Yea I would definitely like to see the numbers side by side.  

 

That way we can discuss a possible change, and we can get everyone's eyes on it and any possible problems/changes that need to be made for the possible switch.

 

I'm all for more defensive stats so we can get a real discussion going about the best defensemen.

[hedgehog337 3,483]

54 minutes ago, Will said:

I did manage to get more reasonable results when i significantly reduced everyones passing and scoring attributes (i did it by a percentage but can't remember what %).

 

look, he just took away a record from Cast. what a tragedy @Victor

[diamond_ace 3,112]

41 minutes ago, Beketov said:

It's not a decrease in player performance (aside from goalies) it's a major increase. So there's no adapting, there's just a massive increase in scoring with goalies all having like 3 GAA.

So the VHLM?

[Da Trifecta 1,899]

@Will @Beketov maybe keep the scoring at the minimum and up the goalies somehow if possible?

[Devise 4,475]

Or lower endurance and other associated attributes. There are a bunch of attributes we have auto set to max with our players, that realistically we could allow players to put TPE into, or buy out of the store similar to Experience. I do think with enough tinkering there would be a way to find a similar balance, but also get on pace or similar sim results. 

Edited September 26, 2018 by Devise

[Beketov 9,025]

21 minutes ago, Devise said:

Or lower endurance and other associated attributes. There are a bunch of attributes we have auto set to max with our players, that realistically we could allow players to put TPE into, or buy out of the store similar to Experience. I do think with enough tinkering there would be a way to find a similar balance, but also get on pace or similar sim results. 

I could be wrong but I’m pretty sure our settings are such that endurance isn’t taken into consideration.

[Devise 4,475]

13 minutes ago, Beketov said:

I could be wrong but I’m pretty sure our settings are such that endurance isn’t taken into consideration.

 

So we could change that as well then. I'm just saying, if the argument is that perhaps we have exhausted some of the fun of V1, then we should be looking at our options that the sim provides us, to mess around with the settings, and get consistent results. Obviously that is factoring in, our current TPE numbers and attribute scale. But I don't think the addition of having to update a few new attributes with us finding the balance for where those numbers should be is that crazy either. Both Durability and Endurance run at 99 for us right now. 

 

Obviously with durability we'd have to open ourselves to injuries, which I know is not something that really makes a lot of sense with how we run things. However I do think we may be able to mess with or enable endurance. According to the V2 rating descriptions. Endurance can affect all stats for a specific game, with a fatigued player being overshifted having lower stats for a specific shift. 

 

Obviously if we did include it, I think it would end up being a stat that a lot of people just quickly invest to 99 to avoid the issue. But that would still take TPE. There is nothing saying the update scale for that has to be the same as the other attributes either. Obviously it's still all reliant on finding consistent results with V2. 

 

 

[ColeMrtz 574]

Maybe I’m just an idiot but doesn’t this aspect of randomness kind of lend to some realism? If the team with the best guys won every time the sim wouldn’t be that exciting, but if the underdogs get lucky here or there it certainly makes things more interesting. I mean it doesn’t happen often but every once in a while the WC makes the finals in real life too.

[Fire Tortorella 2,653]

Absolutely no to making endurance a stat that needs TPE IMO. If we are going to continue (rightly) to embrace that the VHL is based on the 6F 4D 1G roster, then endurance should not be an issue for anyone.

 

Takeaways and giveaways would be nice, but not if we have to change everything just for those stats.