Here's a weird one:
http://pantheon.demigodthegame.com/game/475368/player/11274/

I DCed near the end (have no idea how) when we were beating up on their citadel. Also had a few pantheons that were not recorded.

Heyo, i had posted a few pages back about getting win %'s per demigod to not include mirror matches, and you(bara) suggested using the pantheon filter since that accomplishes it.  Which was an excellent suggestion!

unfortunately, the data isnt calculated properly. When i go to win % i see that almost every demigod is hovering around 10%.

This probably means that the formula to calculate win % is based off the total games played.  I say this b/c when i add up Wins per demigod and Losses per demigod it does NOT equal the total games played, its significantly lower.

for example, Oak with the Pantheon Filter:

Oak - Games Won = 955

Oak - Games Lost = 1879

Oak - Win Percentage = 10%

Oak by my calculations has played 2834 games. his win % should be 33%

Oak - Total Games Played = 9288

if you take the games won and put it over the total games... 955/9288 = 10.2 %

I'm guessing that the total games played data is derived from the people who queue up, but games won/lost is derived only from people who stay connected the whole way through.  This sucks that like 2/3 of people rage quit or crash out of/@ the lobby :/

This might already be fixed though since the data is probably from earlier tournaments(?), so this might disappear when we get the data reset since that will only include new tourny data?  If not, a quick fix would be to change the total games played calculation to be games won + games lost.

btw, does this data include AI's that load up as partners?

UPDATE: so all the DG's have win %s betwen 32 and 35% when i calculate it using the wins/ (wins+losses).  that doesnt seem quite right... or ... maybe it is?

Hi Bara,

Liking the graphs

It looks like you guys stopped tracking custom games data entirely?  I understand they don't count for ranked matches or experience any more, but have you stopped keeping records of the games?

Like my Custom game stats show me with 52 Custom games.  Which was current maybe a week ago...

Thanks

Unfortunately the side-effects of disconnects are still not resolved. 

f1do, I notice that your record is 1-3 now, and I re-ran the job that calculates your win/loss ratio and it didn't change, so I'm guessing that the job isn't the problem but instead that your game is not being recorded.  Unfortunately there are no logs on our end as far as problems preventing games from being saved, so I'm willing to bet it's another disconnect issue (if the host disconnects, no one connects to our servers to save the game, afaik). 

Ke5trl, that isn't exactly the case.  If the host disconnects, there's a chance the game may not be saved at all.  And even if the host stays, something else may prevent the game from being saved if someone quits.

Please be patient with us as we continue to work with GPG in resolving these issues.

I'm not sure what you mean by cumulative win %... A win % can go from 1 to 100%, so if I just add them all up that doesn't really mean much (does having a 13,000 win % after 50 days mean anything?).  Daily win % and experience level is an interesting idea, I may be able to implement those.

That's an error on my part that I have yet to resolve.  The math that I'm doing for calculating that is incorrect, I just need to take the time to rewrite the queries

Thanks for the info (again ), I'll take a further look at the numbers as soon as I have this forum update ready to go.  Until then I wont be pushing another Pantheon update or work on the stats much. The data should not include AI, but some still may.  Late last week I went through and updated many queries to ignore AI.  Some are still not done.  I'll try to get those done soon.

As for your final bit, that all Demigods have a win% in the 30's, that still seems off, and seems like the opposite of what it is now (where every demigod has a win% > 50).

Is that better?

Bara

I am not sure what people have suggested so far, I am too lazy to read through 15pages. But kills/deaths ratios, kills per game, flags per game, average gold per game, favor points per game, etc all the end game stats.

BTW, thanks for the hardwork!

Unfortunately I do not have access to that data so I cannot show it on the website.  Perhaps in the near future this may change.

Bara

So here is what I mean:

Think back to high school:

Daily win %=Test Score
Cumulative Win %=Class Grade

While the daily win % can fluctuate wildly from day to day, a cumulative % will better show overall trends.  The cumulative Win % in also less infleunced by outliers.  So say, the followin occurs:

                   Games Won              Games Played               Daily Win %             Cumulative Win %

Day 1:             3                                  10                                30%                      30%

Day 2:             12                                15                                 80%                     60%

Day 3:             10                                15                                 66%                     62.5%

Day 4:             0                                   5                                   0%                      55.6%

Day 5:             9                                  10                                  90%                     61.3%

As you can see the daily win% and Cumulative win % have entirely different graphs.

Ahh, I see what you mean now.  I may be able to come up with something for that.  Good idea.

Bara

I don't know if they were supposed to be fixed, but I'm still seeing weird anomalies in the Pantheon stats.

I played two games last night and won decisively in both of them (205 favor to 10).  The first one was 1v2; both of my opponents quit after I scored a kill towards the end of the game (one of them explaining first that he was dropping out because I had won).  The game does not appear in my list at all.  The second game was a 1v1; my opponent again dropped out when I killed him and was one my way to knock down his last fortress.  This time, the Pantheon shows me losing 105-0, as if I had dropped out!  I'm actually a bit suspicious that this guy may have figured out how to game the Pantheon by forcing a "blip" in his connection; if you look at his games list, the win ratio is ridiculously high, and believe me, he does not play well enough to explain it.

http://pantheon.demigodthegame.com/game/512827/player/11954/

(EDIT: I found a post on the forums by the guy I played in that game talking about how if you're losing a game you can just quit and it won't count)

One of the upcoming patches will resolve quitting issues.  That along with the stat reset should hopefully prevent this from occurring again.

Bara

Regarding the stats or I guess mine specificly, I have only lost one game and played a handful more, but it says I have played many more with several losses.  When the game has a fatal error and crashes or when all the other players drop and then the AI takes over, then next I leave, beacuse I dont want to play the AI, do those count as losses?  Because if so, that really unprofessonal.

how does the win/loss-count work? some wins (which are now over 25h in the past) havent been counted until now...

We did an update late last week that resolved some problems with the win/loss count.  It should be accurate now.  Keep in mind that the numbers are updated once daily at midnight.

Bara

It may indeed count correctly.  But a large % of games are still failing to show up under "my games", so the numbers are (sadly) still worthless.  I wish I had been tracaking those that were simply missing from "my games," but I would say it "feels" like 1/3 of them are awol.