This project is dedicated to World of Warcraft and was started to analyze and provide latest statistical information about player activities, achievements and ratings.
2016-03-05 | Database Size exceeds 1 Million Active Players |
2015-12-28 | Percentages and History of All Achievements |
2015-12-28 | New section: Raiding Achievements |
2015-12-28 | New section: PvP Achievements |
2015-08-17 | Website redesign: News and Intro combined. |
2015-08-17 | Arena and RBG Faction Fractions over Rating + History. |
2015-08-17 | Arena and RBG Race Fractions over Rating + History. |
2015-08-06 | Arena and RBG Cutoff History. |
2015-08-04 | Personal Percentiles for Arena and RBG Ratings. |
2015-06-29 | Unrated Battleground Win Ratio History for Regions and Countries. |
2015-06-15 | Win Ratio History for BGs and RBGs |
2015-05-27 | Honorable Kills Graph. |
2015-05-20 | Website redesigned to support responsive web design. The content now scales depending on screen resolution (mainly to better support mobiles and tablets). |
2015-03-15 | wow-stats.info is now also registered as wow-stats.com. |
2015-01-13 | Achievement Points Distribution and History. |
2014-12-11 | Improved update rate of Character Activity by Level Graph. |
2014-11-03 | Item Level Information. |
2014-10-09 | Unrated Battleground Win Rates by Time of Day. |
2014-09-15 | Unrated Battleground Popularity Graphs. |
2014-09-01 | Website update: sidebar navigation |
2014-08-30 | Added statistical background |
2014-07-23 | Content update with
|
2014-07-11 | WoW-Stats.info goes online |
2014-01-15 | Start of data collection |
All information on this website is collected via the Blizzard Community Platform API which provides data about characters, guilds and realms. Some of the data available via the WoW API can directly be analyzed and graphically presented, because it provides exhaustive information (e.g. region arena ranking lists).
However, most of the data presented on this website is based on single character information which needs to be collected in a statistical manner. Because of the limited number of samples, this inherently results in a limited precision. Therefore, in most graphs you will find error bars representing the standard error of the corresponding value. In most graphs the standard error is fairly low. With an unlimited number of samples errors could totally be avoided. Unfortunately, the number of requests on WoW API is currently limited by throttling to 864.000 per day.
The WoW API provides a number of complete datasets which provide exhaustive information about a certain limited field. Such are for example:
Most of the information on this website (like leveling information and battleground win rates) is gathered in a statistical manner. Therefore, a database of characters is maintained, analyzed, processed and frequently updated.
To generate an unbiased and representative sample of characters played in WoW, all characters added to the database are selected independently of their individual properties. There is only one requirement for a character: it has to be played actively. Because of the random selection process, a lot of twinks, bank characters and otherwise inactive characters are added over time. Therefore, characters not being updated or not having killed anything (creature or player) for a certain period of time as well as all characters below level 10 are removed from the database.
The characters successfully remaining in our database are updated frequently (typically once a week) to gather and store development and history of those characters. This way, we can extract information like level or item level development, battleground win rates as they change over time and so on.
To enable us to analyze unrated battleground win rates by time of day, a special subset of characters is scanned several times per day. By using the resulting information plus the logout timestamp, the BG win/lose result is correlated with a time.
The subset of characters used for this analysis is statistically idendepent by using a random selection of characters.
Where applicable, precision of the underlying statistical analysis is presented via error bars in the corresponding graphs. The precision is estimated using the standard error of the sample.
The standard error SE is calculated from standard deviation SD by
SE = SD/sqrt(n)
with n being the number of samples. The SD itself is calculated as the corrected sample standard deviation. It depends on the root mean square value of the absolute difference between all individual samples and the sample mean.
As the above formula describes, the standard error is anti-proportional to the root number of samples. Therefore, quadruplication of the number of samples leads to doublication of the precision.
The underlying character database for this project is collecting data since January 2014 and is continuously growing and optimizing itself. The following table shows its current key values: