On the 17th of December there were 211,086 players
On the 25th of December there were 211,012 players
On the 2th of January there were 210,868 players
(Please be aware that since then I have read on Reddit that there were problems with this website showing inaccurate game scores)
First, I removed all of the duplicate names within the datasets. This is problematic, because it has the unintended consequence of removing many barcodes, i.e., IIIIIIIII, still this is a necessary evil. Also using multiple accounts with the same name across different regions are excluded.
On the 17th of December there were 149,919 players
On the 25th of December there were 149,518 players
On the 02th of January there were 141,666 players
Because I have 3 datasets, I can compare Ladder time 1 to Ladder time 2 and Ladder time 2 to Ladder time 3.
A dataset containing names which corresponded on both Ladder time 1 and Ladder time 2 was created. This dataframe consisted of 83,965 players (I wonder how the difference between time 1 and time 2 can be so high! probably due to removing duplicates and looking for unique names consistent across time 1 and 2). Thereafter, I added the wins and losses columns to get a measure of overall games played.
At time 1, an average of 62 games were played per player.
At time 2, an average of 71 games were played per player.
At time 3, an average of 80 games were played per player.
I then removed any player whose number of games played at time 1 was equal to number of games played at time 2 – essentially removing any player who did not play a game. The final dataset contained 33,815 players. These fit the criteria of having played at least 1 game over the two week period and only having 1 unique name on the ladder at time 1 and time 2.
This is the sample which I will use to describe the propensity to rank-up your league.
First I changed the league from levels of a factor to numbers, in other words, Bronze was changed to 1, Silver was changed to 2, etc. At this point I checked which players had ranked up from Time 1 to Time 2. The logic is if league 2 > league 1 score the person ranked up, if league 2 < league 1 score the person ranked down and if league 2 = league 1 score the person remained in the same league.
From time 1 to time 2, 412 player’s league standing went down, whereas 182 player’s league standing went up.
The same steps were undertaken for ladder time 2 to ladder time 3. From time 2 to time 3, 235 players standing went up, while 182 players went down.
These results do appear rather weird, but please be aware of the assumptions undertaken during the analysis. For instance, removing duplicates, making sure that players played across the two weeks. I have also made my analysis, which was undertaken in R, and data files available if anyone would like to check for errors - I am sure there are flaws in my code.
R
Time 1 data
Time 2 data
Time 3 data