[DotA] Average ELO gain stat

[LiquidShards]

I'd like to suggest a new DotA stat which would represent the average ELO that a player stands to gain in games they play. Someone who heavily stacks would have a very low stat while someone who plays fair games would be closer to 15. I believe that this would decrease people's interest in stacking because it would ruin this stat. A possible con to this would be that it could further incentivize friends to smurf and stack a team to raise one account's stats but I don't think this would be a widespread problem.

Inversely, the stat could be ELO Wagered and represent the average ELO the player stands to lose in games they play. Low being better and high being worse with 15 again being the middle.

[nabo.]

Isnt this basically....your elo divided by total number of games? (excluding for those who are above 2300 elo since elo calculation is different)

Are you suggesting an additional stats category to be added or use this as a rating?

[LiquidShards]

It would be an additional stat. It's not ELO divided by total games (or maybe you meant (ELO - 1000) / total games) which would show the average rate you gain ELO. (Or if your ELO was below 1000 the average rate that you lose ELO)

Perhaps the title of my post "Average ELO gain stat" was a little misleading. The stat I'm suggesting would represent the average ELO a player stands to gain per game. This number would be between 0 and 30 and never negative. It would be unrelated to whether the player actually won or lost the games but how challenging the games they played were.

Take the user d00m for example (since he's played few games, to keep this simple): https://entgaming.net/openstats/dota/player/d00m/

As of now he's played 3 games:

[ENT] DotA arem us/ca #0 - Stood to gain 15.34 ELO
[ENT] DotA arem us/ca #86 - Stood to gain 16.52 ELO
[ENT] DotA arem us/ca #76 - Stood to gain 6.16 ELO

(15.34 + 16.52 + 6.16) / 3 = 12.44

Two of the games he played had evenly matched ELO while one was uneven. The average ELO he stood to gain is lower because of it.

Take another user, 2nd.Virgin: https://entgaming.net/openstats/dota/?u=2nd.virgin

Also played 3 games as of now:

[ENT] DotA apem us/ca #68 - Stood to gain 13.31 ELO
[ENT] DotA apem us/ca #55 - Stood to gain 19.00 ELO
[ENT] DotA apem us/ca #24 - Stood to gain 14.15 ELO

(13.31 + 19.00 + 14.15) / 3 = 15.43

Two of the games were slightly in his favor while one (#55) was heavily against him. While he lost #55, he stood to gain more ELO from it (19). This caused his average to even out.

Players that play more games that are pitted against them (in terms of ELO anyway) will have a higher score than players that "stand to gain" little.

The addition of this stat would discourage stacking and encourage players to self regulate the difficulty of the games they play.

[nabo.]

How will this stats discourage stacking when it does not affect someone's ranking nor his elo directly?

What is wrong with people playing with one another? Unfair teams? As you may know, there is no matchmaking system in place, so you cant blame people for deciding who to play with. + Everyone has a choice to play with someone or leave the game when dislike.

We used to autobal to prevent "stacking", but after awhile, autobal games became too unpopular and inactive. I do not think your suggestion is better than autobal bots.

The only solution against stacking is to have a client and enable so everyone can only make 1-2 accounts and matchmake games. This option however is not feasible as discussed on several threads in the past.

[LiquidShards]

It would help prevent stacking because it would show if a player "only stacks" to get their elo. The value of your elo score could be compared to the difficulty of the games you've played. This stat could be added to the /games page as well to further assist players in assessing each other's level of skill. Theoretically people would want to protect this stat just like they want to protect their KDR. Good players look at more than just rank and elo.

Since there is no auto-balance or matchmaking system the default is "it's your responsibility to not play games that are stacked against you". However, what happens is that players who are watching the /games page or using !scores in lobby leave and wait for the next game and the inferior team fills with even worse players. The stack gets worse.

While we could just keep the status quo and blame players that get stacked against, I believe that the addition of this stat would encourage players to avoid being the stackers. This would actually give the high elo player a reason not to play a game where they only gain 3 elo. It would be nice to have the ability to gauge the difficulty of games that people play when you look at their stats. It would be revealing to know that some guy who had a 90% win ratio and cool KDR only played games where the average elo gain was 3-6.This would give elo more relevance by giving it something to compare to.

[nabo.]

Theoretically people would want to protect this stat just like they want to protect their KDR.

While KDR may be one's performance during a game, your avg elo gain doesnt reflect your game performance. Why cant you just do the math yourself elo-1000/number of games? Or isnt it quite obvious when you look at someone's total number of games + elo to judge whether a player plays a stacked game often or not?

it's your responsibility to not play games that are stacked against you.

Not really... why is this a responsibility? If you play a higher risk game (ex. against a stack), your reward will be higher, while your loss is low. What teams you may end up next or lobby or next next lobby is not predetermined.

While we could just keep the status quo and blame players that get stacked against,...

Blame? for wat? They decide to play or not, so there is nothing to blame.

[LiquidShards]

It's not average elo gain (elo-1000)/number of games. Sorry for the poor title that suggested this. It also isn't always obvious to me whether a player stacks or not, especially when they've played a high number of games. (I'm also trying to assess every player in the lobby).

People complain about stacking and it's an acknowledged problem. The multiple discussions on this topic make this evident. While there is a higher reward for winning, the vast majority of the games result in no competition and are over in 10 minutes. When you say "They decide to play or not" that is blaming them for playing a stacked game. You are saying that it's the responsibility of the player to avoid being stacked against, which is partially true, but I think we can do better than that.

Since other approaches weren't popular or feasible, as you mentioned, a new stat could act as a deterrent against people who stack. This would also be an additional reward for people who play games that are determined to be difficult (high elo against them).

Unlike KDR, elo is not necessarily based on your game performance. So think of this as "the degree of difficulty" that you gained your elo. Also, unlike autobal bots, adding a stat would not affect the players so directly. Instead of forcing their teams, which was unpopular, they would be encouraged to "play fair" on their own. Also, the stat would rely on the current elo system so it would not be a drastic change like implementing a new rating system.

[nabo.]

If it isnt (elo-1000)/number of games, what is it?

[LiquidShards]

So it would be the amount of elo a player *might* gain from each game / number of games, whether they won or lost.

Again, add up the amount of elo the player COULD gain from each game / number of games. If this number was low then they generally played easy games. If it was high then it meant that they played challenging games.

[nabo.]

"might" gain.....

archived