War Robots WR Data and Analysis

Could “Hangar Averages” Solve the Matchmaking Woes?

Could "hangar averages" be a solution to some folks' matchmaking woes? Jay runs a simulation to find out!

Player frustration with the new matchmaking/league system doesn’t appear to be diminishing anytime soon, and increasingly I’m seeing proposals and solutions from the community on how to fix this.

One of the most distressing consequences of the new system is that it adds additional challenges for those trying to squad with less-experienced clanmates. The presence of higher-league teammates can drag lower-leaguers in well over their heads, which rarely makes for a positive experience.

This is poor. As a clan leader, I’d like to be able to squad with less-experienced members. Not to serve as a ringer, but rather to promote and encourage them, see how they play, offer guidance and development. And maybe get to know them a little better.

You know, the things people do when they take part in leisure activities together.

In this situation, I’d be more than happy to keep a reserve of “coaching bots” in the hangar. Lower-level bots I could pilot alongside a lower-level clanmate. But aha! That would require the system to matchmake based on hangar strength, and at present the answer to the question of “how can I play in my lowbie bots for fun” is, “play your lowbie bots, and after X many losses or so, you’ll have found your level.”

Yeah, not really feasible.

One of the common suggestions from the community to address this has been the idea of “bot average.” Yes, this gets us back into the paradigm of hangar strength instead of achievement record, but if you’re gonna shoot the moon, why not? I’ve seen this brought up a lot lately, and my first thought was, “hey, that seems simple enough.”

My immediate second thought was “But, but…exponential growth!”

Today I’ll be diverting from the usual By the Numbers content and focusing on one specific question: is hangar average a viable possibility?


To test this theory, I had to recruit a little help.

Let me first introduce the dynamic partnership of Laverne & Shirley, both Level 7 Pattons. In between doing heavy robotics work at the Schotz Brewery in Milwaukee, they also enjoy blasting the opposition with their equally-leveled quad-Magnum array.

Next, meet our other team of volunteers. Both are police detective bots, with the career-minded Cagney coming in at Level 12, and Lacey a Level 2. They’re both well-trained in the use of police-issue Magnums, which are exactly the same as the civilian counterpart, but it sounds cooler to say “police-issue.” Again, weapon and bot levels are equal.

So to recap:

Laverne: Level 7 Patton, Level 7 Magnums

Shirley: Level 7 Patton, Level 7 Magnums

Cagney: Level 12 Patton, Level 12 Magnums

Lacey: Level 2 Patton, Level 2 Magnums

On paper, this seems equal enough- the average level of each duo is 7. But let’s look a little more closely.

First, we’ll want to compare health. Laverne and Shirley’s combined health pool is 93,000*2, or 186,000. That’s a nice chunk of resilience. But what about their opponents? Lacey doesn’t bring as much to the table, just 68,000 health, but Cagney is one tough dame at 127,000. That’s a total of 195,000, and a difference of 9000. About 5 percent- not huge, but certainly relevant.

What about firepower? Well, a quad-blast from Laverne and Shirley will do 1155*8 points of damage, as each Patton has four Light hardpoints (we’re going to assume for the sake of this thought experiment that the two teams will be stationary targets and every shot will hit). What about the cops? Well, Lacey’s again the lightweight of the bunch, doing just 721*4 damage per blast, or 2,884 total. Tough and streetwise Cagney, on the other hand, hammers in for a mighty 7,404. So Laverne and Shirley’s damage per blast is 9,240, against a returning fire of 10,288. That’s more significant- it’s a little more than a 10% damage boost for Cagney and Lacey.

So let’s put this in practical terms, and send these warriors to battle!


Rounds 1-3

And we’re off! Cagney and Lacey take the field on one side, with Laverne and Shirley getting into position on the other. Laverne and Shirley decide to take out the weaker Lacey first, to reduce the incoming rate of fire.

With every shot, big chunks of Lacey’s health bar fall off. The detectives, meanwhile, focus fire on Laverne. Their combined blasts hammer the Patton, leaving her about 1/3rd down on life.

Rounds 4-6

Laverne has lost another third of her life bar, but darkness is beginning to close in for poor Lacey.

Rounds 7-9

Lacey’s had it, and mechs out on turn 8. Laverne is ready to join her at the next stiff gust of wind. It’s now two-on-one…but not for long.

Rounds 10-12

Laverne follows Lacey into the Great Beyond, but not before she gets one blast off at Cagney. At this point, the outcome is a foregone conclusion. Cagney and Shirley hammer at each other, and the weaker Shirley ejects skyward eleven rounds later.

WINNER: Cagney (48,460 health remaining)

Now in the highly simplistic first example, you had each bot pounding away at the opposition focusing fire, but without any coordination. This was effective but inefficient- if only one of the two bots’ firepower was needed to kill off an opponent, the other bots’ output was squandered (“overkill”) when it could have instead started to work on the next target.

But even changing it up a little, with each team coordinating fire and splitting targets when within range of lethal, there’s little change in the outcome. Cagney is the last bot standing on turn 23, and walks off with a bit less health.

What’s perfectly clear is that in this scenario, the outcome is almost certain. The only way this calculus changes is if Cagney and/or Lacey either miss some of their shots, or fall prey to the misfortune of weapon loss that Pattons are notorious for. Of course, both of these hazards are just as applicable to Laverne and Shirley, so it’s essentially a push.

As an alternate scenario, what if Laverne and Shirley try to burn down the tougher Cagney first, instead of the weak Lacey? This will expose them to more incoming fire for a longer period of time, of course, but the idea being that Cagney’s the one doing most of the damage and with her felled first, Laverne and Shirley might make a go of it.

Turns out, this isn’t a very good idea. With Cagney and Lacey focusing on Laverne, she’s sent skyward on turn 10 just as before. Cagney’s taken a beating, but with the opposition’s firepower cut in half, she hangs on until turn 18. Shirley exits stage left on turn 22, with Lacey doing a victory lap with just under 50,000 health remaining.


The problem with the “hangar average” proposition is that upgrades progress weapons and bots on an exponential, not linear path. If each upgrade added +10% to the base statistic of a weapon, for instance, then you’d have a case.

Consider the (fictional) weapon Paddle Thwacker, which does 1,000 damage at Level 1.

Table 1: Paddle Thwacker damage

It does 1,100 at Level 2, all the way up (via linear progression) to 2,1000 at max level. So a Level 2 and Level 12 Paddle Thwacker will do 3,200 damage. And a pair of Level 7’s? Also 3,200 damage. Nice and linear.

 Now look at the exponential column, which is the schedule War Robots uses. A Level 2 and Level 12 combine for 3,953 damage. Two Level 7’s? Only 3,544.

As a result, the more imbalanced teams have an advantage over those more evenly-balanced, and disqualifying “hangar average” as a functional consideration for an alternate matchmaking scheme. Sometimes the simple answers are the best…but not always.

Thanks for reading!

0 comments on “Could “Hangar Averages” Solve the Matchmaking Woes?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: