39 replies to this topic

[Grief-Code]

Might you are interested. Since im building some cool commands for our bot i need to update some small things on my database.

I was a bit surprised by the results but its however funny also

 

Most played heros in 23,095 games:

1. Trax (Drow Ranger), Play Count: 8933
2. LifeStealer, Play Count: 8337
3. Centaur Warchief, Play Count: 6729
4. Dwarven Sniper, Play Count: 6067
5. Oblivion (Pugna), Play Count: 6006
6. Obsidian Destroyer, Play Count: 5196
7. Acient Apparation, Play Count: 5174
8. Skywarth Mage, Play Count: 5167
9.  Phantom Lancer, Play Count: 5116
10. Stealth Assassin (Rikimaru), Play Count: 5086

Last 10:

1. Enigma, Play Count: 381
2. Death Prophet(Kröb ), Play Count: 389
3. Holy Knight, Play Count: 393
4. Nerubian Weaver, Play Count: 397
5. Demon Witch (Lion), Play Count: 440
6. Storm Spirit, Play Count: 474
7. Batrider, Play Count: 476
8. Keeper of the Light( KotL), Play Count: 495
9. Guardian Wisp (Io), Play Count: 528
10. Witch Doctor, Play Count: 566

 

Here are the top winning heros with win percentage:

1. Ancient Apparition, WinPerc: 56,4%
2. Earth Spirit, Win Perc: 55,7%
3. Centaur Warchief, Win Perc: 54,8%
4. Disruptor, Win Perc: 54,1%
5. Obsidian Destroyer, Win Perc 54%
6. Skywarth Mage, Win Perc: 53,88%
7. Oblivion, Win Perc: 52,74%
8. Lifestealer, WIn Perc: 52,65%
9. Rubick, Win Perc: 52,2%
10. Gyrocopter, Win Perc: 52,2%

And again the last:

1. EarthShakre, Win Perc: 36,2%
2. Nerubian Assassin, Win Perc: 36,84%
3. Goblin Shredder, Win Perc: 37%
4. Lord of Olympia, Win Perc: 37,23%
5. Pit Lord, Win Perc: 37,38%
6. Undying, Win Perc: 37,44%
7. Templar Assassin, Win Perc: 37,77%
8. Nerubian Weaver, Win Perc: 38,28%
9. Tidehunter, Win Perc: 38,34%
10. Morphling, Win Perc: 38,37% 

 

But in all im surprised mostly on the last heros.

Storm Spirit is actually a really good hero with some good statistics and gains, also short cast animation.

Guardian Wisp is just awesome, bad gain but 0 cast time is perfect for low mana skills and trolling all over the game

 

What do you think about this statistics?

Edited by Grief-Code, 15 September 2013 - 09:01 AM.
More statistics

[antiweltteilchen]

where u has this database of 23k games from? the stats itself make some sense imo. for the total amount of picks its likely that newer heroes like oracle and earth spirit wont show up, which makes the win% more reliable as a stat imo. talking about oracle i expected him to be in the top winning list, and i didnt expect pugna there. also: WHERE THE HELL IS URSA

[Grief-Code]

This is our database. This are statistics of 2 month?

There we had the last server reset.

Yes we are hosting 500-600 games currently per day, 20 games instant running at the same time.

 

True ursa is missing

[Meatmuffin]

I wan't to play when rgc is off, via garena in your communtiy, but garena keeps telling me that i'm using hacks and throwing me out of game and garena. I don't use hacks wat do

[antiweltteilchen]

stop hacking. huehue.

[Meatmuffin]

lolu

[Grief-Code]

Remove all .mix/.mixtape files like manabars or fog click detector.

 

Then you could also join on server.eurobattle.net, this could be done for example with our modded gproxy:

http://chevelle1.net...msg2884#msg2884

 

For any furthere questions you can also contact me via PM

[Meatmuffin]

but i liek mana bars :C

[Meatmuffin]

Omg i just joined a game it started im playing boom, they throw me out

[Grief-Code]

Lol, why that?

[Meatmuffin]

cuz "im using hacks"

[MYSTATSOWNFILIP]

very weird that io is played so little.... cast time > statsgain anyday for me also IO has awesome, if no turnrate:)

[antiweltteilchen]

just an idea i had: is it possible to parse the skills picked in an equal manner like heroes? for lod it would be a nice statistic seeing the skills leading most likely to a win or loss.

my bets for win would be multicast and shukuchi, for loss it would be remote mines/land mines

[ResQ]

just an idea i had: is it possible to parse the skills picked in an equal manner like heroes? for lod it would be a nice statistic seeing the skills leading most likely to a win or loss.

my bets for win would be multicast and shukuchi, for loss it would be remote mines/land mines

 

code for that exists in lod since a few versions per grief-code's request, it's not 100% done yet though (I think it misses 6th spell still)

[MYSTATSOWNFILIP]

I think aftershock will score high, and so will grow, blinkstrike, stormbolt and ess. I think one of these skilz is in every game that is won. 

 

owh and how can i forget, the oh so popular CHARGE =)

[Grief-Code]

Yea the skills are saved.

Well i have as ResQ said only 5 skills currently in the database. There some issues as already mention but in general it is working.

The problem is the translation, because the skills are saved as skillid, and not as name. I can create a simple top list with the most picked skills but all in ID's, and i dont have the time currently to 'translate' this

[Grief-Code]

Hm i created  a list for this, but it is a bit weired...

See the result here:

http://ohsystem.net/test/skill.php

 

And yea all datas are correct, i checked it manually

[ResQ]

1. Fury Swipes

2. Shadow Strike

3. Telekinesis

4. Searing Arrows

5. Wind Walk

 

..

 

76. Refraction

77. Poison Sting

78. Fervor

79. Smoke Screen

80. Marksmanship

 

 

except for fury swipes and wind walk, quite surprising..

[Grief-Code]

The last skills are understandable imo.

 

But the searing arrows on 4? Wtf....

hm generally i created a new site with all statistics also from players /ff/votekick/drops etc:

 

http://ohsystem.net/statistics.php

 

Working on items currently

[antiweltteilchen]

the last skills arent understandable at all.

refraction is baws in almost every build.

fervor is a nice carry addition.

markmanship is great for impetus builds

 

while in return shadow strike is a rather bad nuke/slow

searing arrows are just wacko.

and telekinesis is a rather bad disable

 

e: may i ask why the abilities are parsed over their ids? its quite easy to implement it directly for the skill name

e2: the skill statistics are wrong for sure. srsly, just look at those numbers, they make no sense at all.