Saturday, April 4, 2015

Wrapping up the Experiment

Earlier this year we announced the DotaCoach Progress Experiment. The idea was pretty simple: we felt coaching was a great way to improve your Dota skills, but we didn't have any proof. So we gave free coaching to nine students and measured their progress over eight weeks. Now that the experiment has concluded, we'll be publishing some data analysis on the results over the coming weeks.

We're interested to hear what type of data you would like to see on this. Obviously we'll be showing things like MMR, KDA, GPM, XPM, and similar stats and how they changed over time. We also have a control group that we tracked -- applicants to the experiment who were not selected; we'll be including data from these players as well (anonymized, of course). But we'd like to know if there's anything non-obvious you'd like us to take a look at. We have all the match data that Dotabuff and the Dota Match History API provide for each of the players, so there's potential to do some interesting things. For example, one thing we're looking at is specifically how GPM improved for people when playing carry heroes, since GPM is more correlated with playing well for those heroes than it is for support heroes. Another might be: for the players who improved the most, which stats correlated most highly with their improvement; i.e. can I become better quicker by focusing on objectives, or decision making, or is improvement in last hitting a better precursor to in-game succcess?

We will be in contact with the individual players to get some qualitative feedback to help answer these questions in addition to the standard quantitative analysis, so even some slightly fuzzier questions are fair game. For example, we've already asked them about what type of coaching (live game, replay analysis, skills training in private lobby, etc.) they thought had the biggest impact.

You can leave any thoughts/suggestions you might have in the comments here, or email us: Looking forward to hearing from you!

1 comment:

  1. Bring me in dude... My steam id: WLmikasa... PleAse add and ty