A compilation of International hero picks, bans and win rates puts Earthshaker and Mirana near the top for total picks, while some older heroes barely appear. What can we learn about long-term popularity and changing metas from numbers like these?
What do the all-time TI hero pick stats actually tell us?
Answers
Raw pick totals are a decent starting point for spotting heroes that have stayed relevant across different tournaments. In the compilation, Earthshaker has 489 picks and Mirana has 487, with reported win rates of 54% and 55% respectively. That's a lot of appearances, not just one hot year.
But total picks don't tell you whether a hero was consistently popular or had a few huge tournaments. I'd compare picks by year and, where possible, use the share of tournament games rather than just the count. That makes tournaments with different numbers of games easier to compare.
Reply to this answer
Don't read low picks as a lack of interest without checking bans. A hero can be highly contested and still rarely make it into games because teams keep removing it in the draft.
Naga Siren is a good example in the supplied figures: 833 bans and 252 picks. Pick plus ban presence, expressed as a share of games where the hero was available, would give a better picture of draft priority than picks alone.
Reply to this answer
Win rate needs some context. Meepo's reported 67% looks great, but it's across just 18 picks. A few results would move that percentage quite a bit, and rare picks may be chosen for particularly favorable drafts rather than being strong in every situation.
The 50-pick cutoff in the win-rate table helps avoid the tiniest samples, but it doesn't make those rates a clean ranking of hero strength. Patch changes, opponents and how teams use a hero all get mixed together in an all-time number.
Reply to this answer
The individual peak years are probably more useful for exploring metas than the all-time ranking. Tiny's listed peak is 82 picks in 2022, for example, while Rubick's is 77 in 2012. Look at those alongside each year's bans and win rates to see whether a hero was broadly prioritized or just picked often.
That still won't explain why the meta shifted on its own. You'd need patch and tournament context before attributing a spike to a particular change.
Reply to this answer
One caveat before comparing the whole roster: the compilation excludes several newer heroes, and older heroes haven't necessarily been available for the same number of tournaments either. It's a ranking of the included heroes' accumulated appearances, not a fair measure of popularity per opportunity.
Reply to this answer
Add your answer
Answers are reviewed before publication.
mechServo~Offbeat:
Ban rules and draft formats can change too, so I'd be careful comparing that presence across years without noting the format.