4Game

The DotA2 Tench is a hard place to be. Speaking as somebody who has been there for 5+ years, it is an under served population with a TON of anecdotal advice surrounding it. With most of the statistical analysis and game play advice being geared toward higher level play, I collected 1,300 sub 1k to low 1k trash tier games to analyse and compare to high level games to figure out once and for all how different trash tier is.

Do Trench games go on a lot longer than high skill games?

Yes. But not as much as you would think.

    Average pro game = 33 min
    Average trench game = 41 min

But there is a difference for sure. Anecdotally, you should be picking heroes that can go into the late game(ie. carries). But does doing that actually increase your win rate?

Does more carries = win?

No. But not having enough carries does. When we first look at the number of carries vs your chance of winning it looks like it does not make much of difference. But if we break down the data and look at Radiant carry vs Dire carry count match ups, we can see that not having enough carries lead to a higher chance of losing but when roles are a bit more even things even out. This holds true at high level play as well to a certain extent. But the difference between high level and trench is you are more likely to play in one of these mismatched games in the trench. Maybe the reason the trench feels so bad is because nobody actually sticks to their role?

Do trench players pick much more greedily than high level?

No. If anything they pick a bit too passive. Take a look at this chart showing high level carry counts vs trench. 50% of games will have 3 carries. This holds true of high level as well. But you are twice as likely to be in a game with 2 cores in the trench…which isn't great.

Thanks for reading this post. Feel free to let me know if there are more myths I should look into.

Some notes on the data:

I manually marked what heroes I thought were carry vs support. I could be wrong on the margins. The trench games all had 10 players and were ranked. Same with pro. I collected trench games from my friends who I know try in their games so that should help with some data variance. The data isn't perfect, but the fact I didn't see anything too crazy tells me it's not too far off.