Question for all of you higher skilled players: Do you generally pick only the heroes that are good, in order to keep your win rate up, or do you try and random every game to get your general knowledge of each hero?
Personally I've been randoming non-stop since I started to learn how each hero works, but it seems from the various discussions that some people regard certain heroes as garbage and don't ever bother picking them for serious games. For example the first few times I randomed sand king I won each game easily; he seemed rather overpowered compared with playing some of the other heroes.
This made me wonder if people just pick the heroes that are good and don't waste their time with other heroes at all, to avoid the risk of losing. In tournament play for example it seems the same set of heroes are picked repeatedly.
For me, having fun is the foremost priority.
-In AP games, I random 75% of the time. In Dota, learning other heroes allows you to kill that same hero better. You become familiar with their CDs, capabilities, nuances, and this can only be truly felt by playing him/her firsthand.
-In AP, if I'm somehow late to first pick, I will situationally last-pick a counter, support, or roamer which perfectly fits what my team needs/
-About 5 years ago first learning Dota, the fastest and best way to learn the game is to random all the time. That helps tremendously. Most heroes are easy to get a moderate grasp of it (Venge, Veno, etc). If I random a complex hero, I'd pick him for next 10-15 games just to 'master' it to the level I like. One time I was really hooked on Invoker. Played him for 15-20 games non-stop.
-New players should play easy, low-ceiling heroes first. Then venture out to randoming to learn the entire roster. I randomed to learn. Now I random for fun. Random is the best (and that sweet extra gold).
-In Captain's Mode (CM), you pick hardcore top-of-the-line heroes and line-up. Well, that's the whole point of CM, people would get mad if you didn't.
-The biggest misconception of Dota 2 is that winrate = skill. It is NOT. The MM gives you games you're supposed to win and supposed to lose. You can theoretically lose easy games, and win hard games and end up with higher rating.
When Dotabuff was relevant, my primary account had 2000 games, 98th percentile with 50% winrate. My secondary account had 700 games, 99th percentile with 49% winrate.
Yes, Valve said Dotabuff's algorithm was bad, but that doesn't mean it's a dysfunctional broken crap. Similar to Starcraft 2 MM, EVERYONE has near 50% winrate except the super elite (0.001%). And they would stabilize to 'only' 55%-65% winrate.