I've been tracking my arena stats fairly closely over the past couple of weeks (28 runs, 12 of which got to 7+ wins). I mostly use AntiGrav1ty for drafting guidance:
https://docs.google.com/spreadsheet/pub?...TFE&gid=15
Against Mages, I'm 16-20, my worst record by a significant margin. 2nd worst is Shaman (14-13), followed by Warlock at 6-5 (sample size means I'm hesitant to draw conclusions there).
I've obliterated Druids: 26-4 (87%), and am ~70% against the other 5 classes.
Despite having the worst record against Mages, I've also done terribly when playing them:
Mage: 4.9 wins per run
Druid: 5.7 wins
Rogue: 6.4 wins
Paladin: 7.6 wins (!!!)
All other classes have been played once at most, so I'm ignoring their data.
Questions:
- Recommended sources on how to best draft for Mages? AntiGrav1ty's resources for Paladins seems to work beautifully for me, but I haven't seen the same success otherwise.
- Am I ignoring Shaman too much? I've only played them once, and they seem to be my nemesis in Arena reasonably often.
- Recommendations on how to play against Mages and to a lesser extent Shamans and Warlocks? I suspect the Warlocks are unfamiliarity/small sample size.
- Anyone else have any data to share? Does my experience against Druids match what you've seen?
Some other thoughts from digging into the data:
- The matchmaking effect is very real (I win 79% of the 1st 5 games in an arena run, and 49% of the rest). I suspect that I don't see other trends because of both sample size and survivorship bias in my decks: the better decks are going to go deeper, and so they're better able to cope with the opposing 8- and 9-win decks.
- My luck in matches where I'm 6-2 before the match has been outstanding: I've gone 8-2 in those matches where it's "get win #7" or "go home short". Probably just sample size, but of all the times for luck to come through....
- I see Mage/Paladin the most, followed by Druid/Shaman, and then everyone else is about equal.
If anyone wants a blank copy of the spreadsheets so you can fill in your own data, let me know.