Analysis: How many matches are surrendered?

We had a discussion on the VainSocial Discord server (shameless advertisement:, wondering whether 5v5 matches are surrendered more often — and how many times.

So we took a look at data from the API and created this chart:

This is from 9 000 matches in 2.12. When analyzing, remember that one team is the winner of the game, so only half as many players surrender as there are matches ending in surrender :wink:

What do you think are the reasons that the surrender rates are like this?

Edit: See also this post Analysis: How many matches are surrendered? for afk rates.


In my personal experience, surrenders happen almost immediately following:

  1. Teammate (or two) deserts match
  2. Enemy team causes ace
  3. Enemy team captures Blackclaw

I’ve found if there’s a large amount of “ALLY HERO KILLED” within a certain timeframe, people are likely to surrender after the tilt of their death is followed by watching their teammates get snowballed.


This is strange. To me, it seems like it is more possible to make a comeback in 5v5 than in 3v3 if you play your cards right.
Maybe in the future, this will change. Right now, most players (especially in lower tiers) aren’t accustomed to the tactics in 5v5.


Hopefully this is just due to a smaller sample size and recent afk issues that came with 2.12

I forgot a zero, it’s based on almost 90 000 matches actually.

It does not look too different for 2.11 either, with a 3 million matches sample:

This is my feeling (and experience) as well, though I have no data to back it up. @shutterfly, o data guru, do you think there’s a way to pick out a comeback from the match telemetry?

1 Like

That is difficult. We need a metric for “comeback”. Is it a comeback when you win the game with less gold than the enemy? Or with fewer kills?

It’s casual so surrenders are going to be high. Also, there is a large disparity in skill tier, ability, and knowledge. There are also some games where the jungler just gets so far ahead that it’s gg. Full CP Koshkas diving under turret with an Ardan to Vanguard is tough. The 12 min surrender is really nice. Taka and Flicker are head aches when people don’t understand vision yet.

I played 2 games today.
The first, everyone was unranked to T5 (according to their last-season rank)
The next game, everyone was T9 or T10

Watching some streamers and in playing 5v5, it seems like in its current state, comebacks seem harder to achieve in 5v5.

  1. The vain crystal regens very slowly, and as soon as one of three armouries are destroyed, enemies can attack the vain crystal (unlike before where all turrets had to be destroyed, giving more time to comeback).

  2. When turrets are destroyed, there’s no acceleration of respawn timers like in 3v3, giving another hindrance to having a comeback.

  3. Another issue is vision; the long cooldowns on the scout cams leaves your base more susceptible to backdoors since supports can no longer spam mines. But with this last point, I think as players start to play more and strategize better this can be solved with better vision control.

Probably it would be best to use something like if significantly behind at some point in the game where significantly behind is quantified in either kills, golds or structures.

I would say coming back from a 10000 gold loss would qualify as a comeback.
Likewise if you are 10 kills behind the enemy
Likewise if you are 5 Structures behind the enemy.

1 Like

The Vain does regen - slowly - otherwise I quite agree with you.

God, the /samples endpoint in the API is sorely missed … It’s a crime that they shut it down.


I actually deel like it’s much easier to know when to give up in 5v5. Games are often pretty one sided, and although comebacks are possible, getting caught out is exponentially easier in 5v5, so it’s also much more demanding positioning-wise to get said comeback. So if the enemy got blackclaw and two of your laners died trying to contest it, people generally know to just surrender and get the loss over with.

My opinion on why surrender rates are higher in lower tiers is lack of experience with moba’s in general + dieing alot (losing hope) + lack of defensive strategic knowledge.

The typical player likes to win/ get kills / not die. When they feel that they aren’t achieving that (which is surprisingly easy to do), they decide to surrender. It’s a combination of factors together, but the most overlooked factor is human emotion.

We make decisions based on emotion, every moment of our lives. So even when deep down we know it could be a winnable situation, the fact that it feels like a loss, we give up too soon.

This obviously gets lower the more experience we gain and understand that winning is actually possible (which is why surrender rates taper down the higher the tier).

This logic can be applied with many situations… emotional intelligence is, imo, the no.1 driving force in good decision making.


Thanks for the clarification - I’ve now fixed it in my original post!

Not enough in my opinion. The games drag on way too long. If your team is getting smashed hard and teammates keep feeding just surrender!

Because it’s a majority vote. Only 4 people need to surrender and you can vote like, 4 times in a row.

Does killing blackwing when it attacks your lane grant gold? Kraken no longer does.

I’ve plotted AFK and surrender rates for a small (~8 000) sample of recent 2.12 matches. Note that AFK is “An ally has deserted” regardless of whether they joined back.

Edit: “Afk rate per match” means that there is at least one AFK in the match. So 11 out of 100 Blitz matches are played with less than 6 players.