I used the deck builder to look for a good mortar deck. I selected Evo Mortar and it showed me different decks.
The current meta deck with hero knight and hero musketeer has a rating of 57, but the same deck with the normal knight has a rating of 65. How is this possible?
That’s simply what the stats say, based on all the battles we have loged. It means the variant with normal knight is more popular or has better winrate.
There’s a bit more context for what “rating” is here:
Thanks for your quick response! I have read the blog post and do now understand what rating means. However i strongly disagree that it denotes the strength of a deck as my example has shown. Another Example: If i search for decks with new Hero Wizard all decks have a rating of 1, which makes sense based on the formula you use because there arent many players using the new card for now. But that gives me no information about the strength of the decks.
Have you ever considered using another method to calculate the strength of a deck? I understand that this is not an easy task but maybe you have something in mind?
We have considered, but the results would be subjective and below our standards. We prefer to stick with stats, which seem more reliable and objective.
We do have other tools to find unusual decks with high performance, such as our deck leaderboards. In that case, the deck strength is proven by a specific player using it.
The point is that i want a tool that compares the strength of slightly modified decks with the original ones. For example i use a homemade mortar deck which gave me a 87% winrate to reach UC but bc nearly noone uses it the rating is low but that doesnt reflect the real strength of the deck. If more people would use it, it would have a higher rating aka strength in your terminlogy. The deck itself uses some substitutes for the meta mortar deck, evo valk for knight/skeleton king and arrows for log/barb barrell to better counter the annoying firecracker. And this modified version should not have a sigficant lower strength/rating then the original one in my opinion.
There seems to something wrong with the number of battles aka popularity of 90k.
If i add up the top 20 popular decks the sum is estimated 4.5k which is not even close to 90k
90k battles = the number of battles with golden knight + pekka. The top 20 popular decks only represent a percentage of all the battles. Do you know how many card combinations there are? This is totally normal.
In your screenshot, you can clearly see that the top result represents 0.8% of all the battles.
