Conquer:Loss ratios

[klapzi]

lol i'm not a legend anymore

This is the unluckiest day ever

I have been last 8 times in a row(really, what are the mathematic chances of this happening?)

I have had territories invaded with half of my number of defenders(5 times)

I think I should stop playing for today.

Edit: 9 times in a row now

Edit2: I'm a legend again, but I had to appeal to 1v1 in the hourglass(I'm very good at it)

How many players were in each of your matches. Based on this, the probabilities could vary wildly.

Five eight players games, two four players games and two two players games

8 x 8 x 8 x 8 x 8 x 4 x 4 x 2 x 2

One in 2097152. I could have .won the lottery

[techgump]

There are a few issues with this way of calculation.

1. Map type and number of players in a typical battle.
2. Moreso, those that quit before games' end will have better c/l.
3. Typical opponent skills. Playing guests and noobs vs. playing the pros.

[Chess435]

There are a few issues with this way of calculation.

1. Map type and number of players in a typical battle.
2. Moreso, those that quit before games' end will have better c/l.
3. Typical opponent skills. Playing guests and noobs vs. playing the pros.

I'll give you that. But at least it's better than wins/losses.

[Ratburntro44]

The only possible good way of measuring how good someone is at Vortex Wars would be to use a set player as the test, then playing a one on one game against them with EVERY possible game type on every size map and recording every full game to determine how much better or worse each person is compared to the set person.

Also, I would say it is relatively normal to have a higher than 1 conquered/loss ratio. Say you play all 8 player games. Then, even if you win 2 out of, say, 8 of them, you still have conquered 14 people's worth and lost 6. Even if you win 1 out of 8, you still have conquered 8 people's worth and lost 7.

[techgump]

Not so sure about that. I think it is all pretty unreliable, as Very Much depends on what map types you play, how many players, and the skill of other players, and silent alliances.

For example, I have improved my win percentage from 33% to 43% by changing over to Full map, HC, random from Conquer, Att, Manual... after first playing about 700 games in the later mentioned mode. That is a huge difference, mostly based on the fairness of start position in the mode I play now vs what I was first playing. Likewise, I play with 7 to 8 people. I like big games because they are more challenging. This means my odds of me winning however decrease dramatically from someone playing with less players... This is also very crucial.

all things even, Standard Win Percentages:
2 players: 50% odds of win
3 players: 33% odds of win
4 players: 25% odds of win
5 players: 20% odds of win
6 players: 17% odds of win
7 players: 14% odds of win
8 players: 13% odds of win

These odds do not consider skill levels of opponents, nor allies. Which in some games I know, 4 player maps have two people that tend to either ally up, or silent truce with end draw, therefore maintaining super high win percentage and C/L ratio.
I think upon a little more study, we can come up with a far more comprehensive formula for this, which includes wins/loss/draw (if we can get draw stats, or kill draws all together which may be best, as allies would be non-existent too; only one could win), c/l ratio, number of players, number of played games (stats become more reliable with more and more data/games played), and perhaps even map type settings. I think player skill will have to be disregarded, as there is no way to track that easily. The more accurate and relevant variables we can include in the formula, the better standings we can produce.

One we can easily correlate is win percentage vs standard win percentage. They are inversely proportional.
IE: as number of players increase my odds of win decreases.
Since my games average 7 players, and my odds of winning in 7 player games is 14%, but my win percentage is 43%, I have a W/S (Win over Standard) of: 43/14 = 3.07 (ie: I am 3 times better than the standard odds). I think this is a very important aspect we can leverage as well.

As for games played, as more games are played, the degree in which the stats are a reliable judge of success (or failure), increases. For example, a new player plays one game and loses. This one game is not a reliable source of data. After 100 games, it is a bit more reliable, after 1000 more much, and so on. So, we can also leverage this to produce a data accuracy %. In general, the more games played, the less the variance. Perhaps a way to do this, and I am open to suggestions, is,
1 over number of played games (1 / X). For the guy that has played 1 game, his is 1/1, which = 100%. 100% Data Variance; meaning it is 100% unreliable; the spread is LARGE; could be very accurate, or very inaccurate, we just don't know. Mine would be: 1/1797, in percentage = .06%

To make things easy, we can covert everything to % for now?
This means instead of calculating c/l, we calculate c/t (conquered over total owned (conquered + lost).
For me:

C/T (conquer/total owned): 63%
W/T (win/total played): 43%
7SWR (7 player standard win rate): 14%
DV (data variance): .06%

To provide in non % output:
C/L (conquer/loss) = 1.80
W/S (win/standard) = 3.07
DV (data variance) = .0006

Now we are getting closer to a clearer picture. Still need an end formula, and this does not again provide data for player skill, silent alliances, draws, or map type. But it is a step closer.

[klapzi]

Everything you said

I think the only real way to judge vortex wars skills is to divide the players in categories.

I will use Player 1 as an example:

Start Type:
Conquer --- Win ratio ~55% 305
Full map --- Win ratio ~45% 205

New Troops:
Manual --- Win ratio 51% - 507 games
Random --- Doesn't play
Borders --- 100% - 3 games

Game Type:
Attrition --- Win ratio 60% - 255 games
Hardcore --- Win ratio 41% - 255 games
1q1 --- Doesn't play

Players:
2 - Win ratio 75% - 35 games
3 - Doesn't play
4 - Win ratio 45% - 200 games
5 - Doesn't play
6 - Doesn't play
7 - Doesn't play
8 - Win ratio 40% - 275 games
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Now you make whatever equations you want to determine an "LVL" for each skill and you can get for example:

Player 1:
Conquer - ++++++++
Hardcore - +++++
Manual - +++++++
Borders - ++++++++++
Attrition - +++++++
Hardcore - +++++++
2 Players - ++++++++++
4 Players - +++++++++
8 Players - +++++++++


This is what is done in Football games. It's very difficult to judge a player in an overall, a defender will never shoot as well as an midfielder but that doesn't mean the midfielder is a better player, he just shoots better.
But if you separate his skills, you can get a better ideia of his capabilities. Being 100% acurrate is impossible, but I think this is the closest you can get.

But obviously, to do that you would need data that we don't have.

[techgump]

Exactly part of the problem, we have to work with available stats. We do not have stats for how many games "Player One" has played in Start Type, or Troop Type, or Game Type. Hence most of us will not be able to provide this, and I believe we are looking for solutions based on what we have, not is what ideal. That said, I can tell you how many players I play with on average.... and I would be willing to bet, most seasoned players can if they are honest. Based on what I am offering, I think it is the closest we can get for now. Now to figure a formula.

[Chess435]

Exactly part of the problem, we have to work with available stats. We do not have stats for how many games "Player One" has played in Start Type, or Troop Type, or Game Type. Hence most of us will not be able to provide this, and I believe we are looking for solutions based on what we have, not is what ideal. That said, I can tell you how many players I play with on average.... and I would be willing to bet, most seasoned players can if they are honest. Based on what I am offering, I think it is the closest we can get for now. Now to figure a formula.

Now I sort of wish for uber-detailed stats for the purpose of metagame analysis, but I don't want to put too much pressure on Lopdo. It would be nice, though for standardized tournament ratings, perhaps via the Elo system. If Lopdo keeps tournaments internally consistent, or seperates different sorts of ratings, then we'll have a level playing field with which to measure player skill with.

[balderr]

4149/2127=1,95...i feel like a pro!

[Lopdo]

I do track more detailed statistics for each player and I planned to make it available to users but I decided to drop because I didn't have time to implement stats screen