Originally Posted by pyrogene

Hmm... so no differentiation when looking at stats for different ranking sub-groups? I do believe there would be a significant difference in these stats if you look at say, 300+ only games. And it might be more meaningful to look at game balance for more "competitive" players rather than the entire player group.

I was wondering the same thing. Powerlevel need to be evaluated at a competitive level. If a hero isn't played in the 300 rating that's saying someting about the power level of cards available to that hero.

It's statistics, there will always be errors...

Think about it this way: If Elad has a 3% fta across all matchups by all players in the game, that just means that that 3% should be viewed as a remarkably high number (in context).

There are two items in question here: the statistic and the interpretation of that statistic. No one care's about the statistic per se; it is only once one makes the interpretive leap to "power level". What we care about is how much more powerful is elad when going first as compared to her power going second.

3% sounds like a small number, and a small number makes it seem like elad's the power difference is also small. However, power level is a potentiality, whereas the statistics are actual outcomes. But the gap between power potential and actual results can be quite large. It is consistent to say that elad is massively more powerful when going first vs. when going second while also saying that she only wins, overall, slightly more often when going first. Why? There are at least 3 factors. A less skilled player with a great deck can waste potential power by misplaying. A skilled player with a poorly constructed or experimental deck can lack opportunities to take advantage of potential power on account of card choices (i.e. elad with all neutral support abilities will have a 0% win rate, but this does not meant that her power level is any less). And finally, a skilled player with a well designed deck can fall prey to bad luck in draws.

So there are many factors that mask a hero's power level. One could interpret the fact of 3% fta by admitting that this number is low but interpreting it's lowness as a testament to how much of elad's potential power (with fta) is wasted. Looking at it the other way, we might say that although 3% sounds like a low number it is actually incredibly high given that her fta is still 3% after so much of her potential power is wasted suggests.

A true judgment on power level is only available statistically if you assume you have players of perfect skill with perfectly constructed decks playing against one another for enough games to make an adequate sample size. Then, if you measure win rates, the interpretive element of statistics is reduced to a minimum.

On a tangential note: does anyone else get the feeling that baduruu might be the only hero who actually has a SECOND turn advantage? My stats and my in game experiences seem to suggest that this is true. And I speculate that this is because putting a bow in play on t4 is much more game-altering than on t5 combined with the fact that baduruu's ability costs a card which is relatively more expensive when one goes first and starts with one less card to begin with.

No reply by Mojumbo on this.

Originally Posted by pyrogene

No reply by Mojumbo on this.

If by "this" you mean the question on average rating, that's a Kyle question.

Originally Posted by Gondorian

If by "this" you mean the question on average rating, that's a Kyle question.

No, was referring to basically this entire thread and the discussion it spawned.

Originally Posted by pyrogene

No, was referring to basically this entire thread and the discussion it spawned.

OK. The general approach we try to follow is to let you guys discuss things yourselves rather than have us weigh in too often, other than to clarify things, which I did a few pages back.

That's most likely why Mojumbo hasn't replied here. But if you have a specific question for him, now that the discussion has died down, post it here and hopefully you'll get a reply.

By the way, we have stats just for high rating, and they are very similar to the overall ones.

Hmm, ok. I guess the main questions from this thread would be:

1. Is it better in his opinion (and also what is currently being done) to look at stats for the entire player base or only for higher ranking players for balance decisions?

2. Are there differences in stats between subsections of players with different rating? Specifics would be nice but I understand if some data cannot be revealed.

3. It's mentioned that no heroes have a win rate of above 55% but are there heroes with win rate under 45%?

Originally Posted by MacLeod

Another more lively example for the same point.

Let two monkeys play SE. No matter which heroes you assigned to the monkeys, the win rate will be about 50%. And it also doesn't matter which monkey goes 1st!

But does that mean SE is perfectly balanced?

you just totally made my day mate

OMG, win rate from QM? The ratinq system is made for win rate around 50% !!! Everyone climbs up till moment when he starts lose, and than most of us have constant 50% win rate. Even very best players reach rating 330 very fast and than they have around 50% rating against other best players... I think this "Trueskill" rating is about to set u somewhere in the field where u have 50% win rate. So more u play,... closer to 50% !!! Only players who play very little or players on edges can have win rate far from 50%.

I'm quite happy for overall deck balance and variety, what I'm unhappy with is the number of expensive cards in store that look like bombs but are still unbalanced and unplayable that I feel are just there to trick TCG noobs. In a small tcg with limited amount of cards, you really can't afford to have many competitively unplayable cards.

Originally Posted by pyrogene

Hmm, ok. I guess the main questions from this thread would be:

1. Is it better in his opinion (and also what is currently being done) to look at stats for the entire player base or only for higher ranking players for balance decisions?

2. Are there differences in stats between subsections of players with different rating? Specifics would be nice but I understand if some data cannot be revealed.

3. It's mentioned that no heroes have a win rate of above 55% but are there heroes with win rate under 45%?

1. It's important to pull stats from all levels of play really. You want to get an idea of what newer players are attracted towards playing and how they do with the heroes they choose to use. Naturally pulling stats from the top tier players makes sense because it gives you an idea of how balanced the game is in the higher levels of play which helps determine the health of the tournament scene. Lastly, it's a good to pull the overall stats so you can look for "glaringly" overpowered heroes. Let's say a hero, overall, was holding a 60% win percentage overall. That would be pretty telling that there might be an issue with that particular hero. Overall stats are also good for determining the raw popularity of heroes.

2.For the most part, hero stats are rather consistent across the ratings so far. You do get minor differences in first turn advantage win rates, overall winrates, and hero usage but Eladwen, for example, holds a similar win rate and usage % both when broken into subsets and taken overall (her first turn advantage fluctuates between 2-4% depending on the data being looked at).

3. Yes, very few heroes have a win rate below 45% but some do exist. Generally, the heroes that aren't as popular (the few that get played around 1-2% of the time) also harbor a win rate around 40%. Jericho, for example, holds a 42% win rate and is played 1% of the time among high ranking players. It's important to keep in mind that this isn't necessarily a reflection that Jericho's power level is lacking but rather that he is unpopular. In fact, Kyltz recently used Jericho in the Wulven Challenge Tournament and won with him only dropping 1 game the entire time! It should be noted, however, that we are always looking for ways to increase the popularity of our lower rated heroes to get them some time in the spotlight. The increased exposure can then be turned into a higher win rate

If two players are both using Eladwen, one will lose and one will win.. does this water down the stats you're looking at? Can you look at it without the mirror?

Originally Posted by Drakkon

If two players are both using Eladwen, one will lose and one will win.. does this water down the stats you're looking at? Can you look at it without the mirror?

Yes we can

Thanks for the reply! Good to know that there is more analysis done behind the scenes.

If they only nerfed Eladwen a little bit more... we have a nice balanced and playable meta, way more than Yugioh and MtG; but it's a shame to still see kris-puwen-aldon-oracle-nova-nova-fireball-fireball at every level. And that's why Millstalker still reigns supreme

Originally Posted by Pandevmonium

If they only nerfed Eladwen a little bit more... we have a nice balanced and playable meta, way more than Yugioh and MtG; but it's a shame to still see kris-puwen-aldon-oracle-nova-nova-fireball-fireball at every level. And that's why Millstalker still reigns supreme

I don't think Eladwen is popular just because she is OP. It's probably also due to relative ease of play and the fact that the deck is cheaper to build compared to item heavy decks. And even without Eladwen, I doubt the less popular heroes like Jericho would see much more play.