Player Ability Ratings

[bruins72]

Also, I CAN imagine a defenseman coming along and putting up 40 goals and 120 points like Orr and Coffey did. It will take a generational player though. Who have been the best defensemen over the past several years?

Also, I think a healthy Lemieux in his prime would STILL dominate today and possibly score 70-90 goals with 90-120 assists. There, I said it.

I just snipped your quote to reply to just what I had a comment on. Honestly, the way today's game is, I just don't see anyone putting up those kind of number. Orr, Gretzky, and Mario in their prime wouldn't have put those kind of numbers in today's NHL. It's a different type of game.

I like the proposal of having a 1-25 attribute range with the 21-25 range being reserved solely for generational players. If we were to do the same with a 1-20 range then it'd be too restrictive in that 99.99% of players (i.e. all of the non-generational players) would be limited to just the 1-15 range. 1-25 would give a little more breathing space.

Another thing that could be done is to really limit the number of 19's and 20's given out. I think by going more conservative in rating players, you could save the 19's and 20's for truly special players.

The new player templates sounds like a good idea. So that when a new player is created his attributes/CA/PA would be influenced by a template. Perhaps each template could be weighted with a probability/likelihood rating? So when the game is creating a new player it randomly picks a template but the weighting makes it more likely that certain templates will be chosen. This would ensure generational players only crop up once every 30 years or whatnot.

Also, the generational player template could be removed/toned-down if users didn't want generational players in their game.

The only drawback with templates is that it could completely skew the game. If it is based on randomly picking a template each time a new player needs to be created then it is possible that an unrealistically large number of very good or very bad players could be created. I suppose at least the principle behind EHM's player regen system (i.e. a new player is created based on the CA/PA of a player who has just retired) will ensure that the balance within the hockey world is maintained - obviously the actual implementation of this by EHM perhaps hasn't been 100% successful

I guess we could keep track of the quality of players who are retiring and have that in mind when creating new ones. While I like the idea of varying quality from year to year we do need some limitations.

Would be nice if countries like Denmark or Norway could have a chance of producing great players since I think that is starting to happen irl.

Okay so how about this...

If we look at the NHL as being the top league in the hockey world we can set up the player templates that way. The templates could based on the player's style and their career peak. For example, you could have "Franchise Playmaking Center" and "Checking Line Winger". You could also have "Minor League Energy Winger" or "Junior Hockey Goon". The career peak part of it would kind of tie in with CA. So when a player retires, have a new player generated of the same player type and have the career peak have a small chance of improving or worsening. For example, if you had a "Second Pairing Offensive Defenseman" retire, the new player that is spawned will most likely also be a "Second Pairing Offensive Defenseman" but there's a small chance he could be spawned as a "Top 2 Offensive Defenseman" or maybe a "Botton Pairing Offensive Defenseman".

To mix things up and have a chance of talent coming out of countries like Denmark or Norway, you could have it so when a new player is spawned there is a 90% chance that he comes from the same place as the retiring player and then split the other 10% between less likely places. So when Brodeur retires, there's a 90% chance that the new player spawned to replace him will be Canadian. There is also a 2% chance that he's German, a 2% chance he's Norwegian, 2% chance he's Danish, 2% chance that he's Swiss, 1% chance he's Austrian, and a 1% chance he comes from a random country (all others). That would give it the potential to mix things up a bit.

Also, I hope this isn't taking things too far off topic but I think the game should look at the USA and Canada a little differently than other countries because of the way that the CHL leagues (WHL, OHL, and QMJHL) determine player eligibility. So you'd have Canada West, Canada Central, and Canada East along with US West, US Central, and US East (or however those CHL leagues break it down). So when a new player generated, they don't have to come from the same birth city as the player they're replacing, they just need to come from the same region. Does that make sense?

[dabo]

Yes, we could do it something like that and then tweak it until it feels ok.

[ArtVandelay]

Also, I CAN imagine a defenseman coming along and putting up 40 goals and 120 points like Orr and Coffey did. It will take a generational player though. Who have been the best defensemen over the past several years?

Also, I think a healthy Lemieux in his prime would STILL dominate today and possibly score 70-90 goals with 90-120 assists. There, I said it.

I just snipped your quote to reply to just what I had a comment on. Honestly, the way today's game is, I just don't see anyone putting up those kind of number. Orr, Gretzky, and Mario in their prime wouldn't have put those kind of numbers in today's NHL. It's a different type of game.

Here's another perspective.

Since the player database will be editable, it's quite likely that people will use the game to recreate bygone eras. Without the ability to create some on the level of a Gretzky, Lemieux, or Orr, it's going to suck.

[bruins72]

Hmmm... you've got a good point there, Art. That would be important in making historical databases.

[BTerran]

I disagree with the premise that there won't be anymore generational talents. No matter what the game of hockey is like today is irrelevant. The game has changed a lot over the last century and it will continue to change, but there has still repeatedly been players who have risen well above everyone else.

It's especially moot because there is a generational talent in the NHL right now, he's just not a forward. I would argue that Martin Brodeur is a generational talent. Yeah, he's been helped by the era he's played in, but the same can be said for Wayne Gretzky. Brodeur holds a number of goaltending records and it will be a while before we see another like him.

Also claiming that there's no bad players in the NHL anymore is a matter of opinion. What defines a bad player is relative. Just because the talent pool is perceived to be better than it once was doesn't mean that a player can't rise above it.

Using baseball as an example for a minute, it's safe to say that baseball players today are significantly better than they were back when Babe Ruth dominanted the league. And yet, even at the height of the steroid era in baseball (ignoring the morals of it for a moment, it did increase the overall abilities of the league), there was still one player that rose above them all in Barry Bonds. So even when the talent pool increases (in baseball's case through artifical means) through leaps and bounds, it's still possible for one player to rise well above everyone else.

[archibalduk]

It's true what you say Archi, but if the game has thousands of players in the database, it's very unlikely that some day there wouldn't be any good players in the world. Sure the average ability of active players can change but I don't think it's gonna rise or fall dramatically.

I agree in principal - as we could weight templates, etc. However it would be a statistical possibility that a large number of good players could retire and their replacement regens could all be very poor. The more seasons of a game you play, the greater likelihood of this happening. Clearly it would be a slim chance but there would still remain that possibility that a user's game could be ruined/hampered because all of the talent suddenly dries up. That is assuming we used some sort of random CA/PA generator.

Having said that, I think the ideas proposed by Dabo and Bruins72 could address this (slim) possibility. Based on what they have suggested, why not have the regeneration of players (by regeneration I mean one play retires and a new player is generated) made up of three parts:

1) Firstly, the regen player's PA is the same as the retired player's CA plus something like 15% or minus something like 5%. Remember that when a player retires they are likely to be past their peak and so their CA will not represent their peak CA. To compensate for this, the regen's possible PA range should be much higher than the retired player's CA or only slightly less. The game would generate the actual PA by randomly choosing a number between this range. If the PA was simply the same as the retired player's current CA then slowly each generation of players would have lower and lower PA's

2) We could assign a template based on the retired player's position. So if the retired player's primary position was a winger then the template use for the regen would be a forward's template (e.g. sniper, two way, etc).

3) Nationality. Although we do have the occasional superstar from countries where they wouldn't ordinarily be expected, the balance of talent ought to be kept fairly balanced. Perhaps we need some sort of neighbouring countries entry in the database for each country? So if a retired player is Canadian then his regen will be from either Canada (say a 90% likelihood of this, as suggested by Bruins72) or from a neighbouring country (10% likelihood) - in this case it would be the US. Take another example - retired player is Swedish so there would be a 90% chance of the regen being Swedish and a 10% chance of him being from a neighbouring country (i.e. Norway, Finland, Denmark).

Perhaps it would be a good way of ensuring a reasonable balance of talent and of talent from different nations whilst allowing for some variation? Perhaps I have complicated things a little?

I like the proposal of having a 1-25 attribute range with the 21-25 range being reserved solely for generational players. If we were to do the same with a 1-20 range then it'd be too restrictive in that 99.99% of players (i.e. all of the non-generational players) would be limited to just the 1-15 range. 1-25 would give a little more breathing space.

So when using the 1-100 range generational players would also use the 101-125 range.

Yes

[axwel3221]

I personally think that B72 suggested the best way to do this.

Also the 1-25 scale sounds good, about the 20-25 assigned to new generations of hockey.

[bruins72]

1) Firstly, the regen player's PA is the same as the retired player's CA plus something like 15% or minus something like 5%. Remember that when a player retires they are likely to be past their peak and so their CA will not represent their peak CA. To compensate for this, the regen's possible PA range should be much higher than the retired player's CA or only slightly less. The game would generate the actual PA by randomly choosing a number between this range. If the PA was simply the same as the retired player's current CA then slowly each generation of players would have lower and lower PA's

Hmmm... that could be an interesting idea. I like the idea of randomizing the new PA. So with the method that you described, if a player with a 154 CA retired, the regen's PA would be somewhere between 146 and 177. I wonder if the 15% might be a bit too much? Maybe scale it back to 10%? Anyhow, I like the general idea of this.

2) We could assign a template based on the retired player's position. So if the retired player's primary position was a winger then the template use for the regen would be a forward's template (e.g. sniper, two way, etc).

I really think the templates are going to be key. Once dabo gets to the point where he's getting ready to implement these, we should compile a list of player types that we would need templates for and pick which attributes are important to that type of player.

3) Nationality. Although we do have the occasional superstar from countries where they wouldn't ordinarily be expected, the balance of talent ought to be kept fairly balanced. Perhaps we need some sort of neighbouring countries entry in the database for each country? So if a retired player is Canadian then his regen will be from either Canada (say a 90% likelihood of this, as suggested by Bruins72) or from a neighbouring country (10% likelihood) - in this case it would be the US. Take another example - retired player is Swedish so there would be a 90% chance of the regen being Swedish and a 10% chance of him being from a neighbouring country (i.e. Norway, Finland, Denmark).

I think your "neighboring countries" method could work too. The only thing it doesn't really allow for is the random odd talent from a country like Japan or something like that.

Perhaps it would be a good way of ensuring a reasonable balance of talent and of talent from different nations whilst allowing for some variation? Perhaps I have complicated things a little?

I think these things are always going to be complicated if they're going to work. Plus, brainstorming is good!

[dabo]

Once dabo gets to the point where he's getting ready to implement these, we should compile a list of player types that we would need templates for and pick which attributes are important to that type of player.

I was gonna suggest this myself, perhaps we could start sooner rather than later since I will be needing players for early testing of other features.

[empach]

Remember that when a player retires they are likely to be past their peak and so their CA will not represent their peak CA.

One way to counter this problem is to store the players top CA over his career in a variable and use that variable rather than the current CA for player regeneration.


These ideas sound pretty good guys.

[archibalduk]

Remember that when a player retires they are likely to be past their peak and so their CA will not represent their peak CA.

One way to counter this problem is to store the players top CA over his career in a variable and use that variable rather than the current CA for player regeneration.

Thinking more about it, we could actually base the regen's PA on the retired player's PA because a player's PA never changes. The PA indicates the maximum CA a player can reach at the peak of his career.

So my earlier suggestion was complicating things unnecessarily. Instead, why not base the regen's PA on the retired player's PA but with some sort of random element to add some sort of variation? E.g. the regen's PA could be a random number plus/minus 10% of the retired player's PA?

This in essence is what empach suggests above. The PA is actually a reasonable record of the peak CA a player will achieve at the peak of their career. Obviously a player's CA may never reach his PA (in fact I wonder how often players actually do) but it would be good enough for our purposes.