There are a number of things that can cause this:
- Hero matchup data is "eventually consistent", meaning that for example a win for ES might appear either before or after the corresponding loss for Bane. The window for this difference is usually quite small (minutes) but can be up to several hours in certain conditions.
- It's possible for both heroes to receive losses, such as in the case of an abandon on the winning team. This happens quite frequently, meaning that the sum of the win rates for both heroes versus each other will almost always be less than 100%.
- Pages can be cached at different times. Check the footer at the bottom of the page. For example, right now the page for Bane is almost an hour old, where the ES page is only 10 minutes.
Thank so so much for this response! Really clears up my question. I understand the second two points very clearly, however could you please elaborate on the first. Is it to do with the steam software registering the win late or is it something to do with your code that gets the data from steam?
The eventual consistency of hero aggregate data is on our end, a trade-off we make for better performance and higher availability.
Будь ласка, увійдіть до системи для написання коментарів.
Can anyone explain the small percentage difference between hero's winrates against each other. For example, taking two heros basically at random: bane and earthshaker. On the bane page, vs ES: winrate = 45.23. However, on the ES page, vs Bane: winrate = 53.27. These numbers do not add up to 100. Now this is true for virtually every hero matchup and I am curious as to why.
Also interesting is that the number of games can be different. Why?