War Robots WR Data and Analysis

By the Numbers: The New Matchmaker

The long-awaited matchmaker update is here- and it's turned War Robots upside-down. Jay looks to data to make sense of it all!

It’s finally here.

Pixonic’s been hinting at a matchmaker change for awhile, and for those of us on iOS, we woke up yesterday to a Brave New World.

Or maybe a New World for the Brave? As per usual, horror stories and doomsday scenarios have come hot and heavy. From the volume, you’d be forgiven for thinking that freshly-minted pilots with their Level 1 Destriers were suddenly being offered up like sacrifices to the deities and demigods of the Top Tier. But that’s cold comfort to a lot of players who had become accustomed to a certain way of doing things, only to find all that work had been undone in a stroke. 

Today I want to take a look at the change in the matchmaker system. Like anyone whose paycheck doesn’t say “Pixonic” on it I have no insight directly into what’s going on on the coding level, but we can look at the output and draw some conclusions. It’s no mere hyperbole to say that the impact of the change has been potentially seismic, with some long-term repercussions less obvious than simply, “who am I playing against.”

I’ll begin by discussing the current MM implementation, then dive into observational data.

Header.png

As seen in previous articles, my position on “clubbing” has been somewhat nuanced. In short, I feel that the concern of players gaming the system to give them unfair advantage over junior players has been justified. However, in large part because of how demoralizing and frustrating it can be for those on the receiving end of it, its prevalence appears to be overstated. Only once in the several months I’ve been playing did I put down the game for the night because of an overabunadance of clubbery.

The main problem from Pixonic’s perspective was also one of their great strengths- a “tiering” system that allowed for a diversity of gameplay experience. By making the gear rather than the player the determining factor driving gameplay, you ended up with a Bronze Tier that could be enjoyed for its own merits rather than simply a way station on the path to progression. Not everyone enjoyed it, but not everyone had to. Those looking to rocket to the top could coexist for those looking for a more casual experience.

In that regard, the tiering system worked, but it also left itself open to exploitation. Players discovered that they could suppress their bot level in favor of their weapon level, and the matchmaking system would weigh the former over the latter. The light, nimble yet resilient Gepard was the perfect clubber platform, and many would level it up to 2 and then pack on highly-upgraded weapons to give it a huge punch relative to its competition.

There are many who have expressed the opinion that the revamp of the matchmaker system- which appears to focus more on player performance than gear levels- is a direct result of player outcry.

Does that mean the clubbers are at fault for the new system?

header

Wouldn’t that be ironic? The clubbers being at fault for making Pixonic ruin the game because they couldn’t control their own greed and were themselves ruining the game?

It all depends on whether or not the change was proactive or reactive.

If it was reactive, then it was a response to a threat to the quality of gameplay. An acknowledgement that the previous matchmaking system was not performing as desired and/or intended. With no greater pain point for the playerbase in terms of matchmaking, then some guilt assigned to the clubbers and clubber-clans would likely be justified.

On the other hand, it may well be that this was a proactive measure for some other objective, that had the side-effect of tamping down clubbing behavior. What would be the motive for this? What possible objective to tinker with the game?

It might be that the feature mentioned above- of “tiering,” was in fact a problem. If players could stick around a lower tier for as long as they like, slowly accruing resources for free, it removed some of the incentive for advancement. Perhaps the game worked too well, insufficiently prodding players to move up the ladder to “bigger and better,” which would incentivize real money (RM) transactions.

Nobody’s reaching for their credit card to buy Gold to help speed up leveling their Schutze from level 2 to level 3. But when you start looking at bots like the Galahad and Fury, or weapons like the Orkan or Zeus (or perhaps with the upcoming buff, even the Ancile), players who are pushed to be more competitive may well drop some RM on their journey to excellence. And even that’s just a scratch on the pitch of how much Silver it takes to top them out.

By making game progression lineal rather than categorical, a case can be made that Pixonic is looking to simply advance the playerbase monetization agenda. The clubbers are off the hook!

Of course, it could always have been a case of two birds, one stone. And ultimately, does it really matter? It’s here, and we have to deal with it.

header

The evening before the matchmaker update, I happened to be playing some War Robots and collecting some data. The only change I’d made to previous data collection methods was to add in the Victories stat. This wasn’t so much a measure of player quality so much as player experience. The amount of game play it takes to accrue 500 victories will vary depending on player performance, but we can be sure that player has played at least 500 games.

For the sake of reference, here are my statistics.

Level: 30

Cups: 350*

Victories: 473*

Hangar slots: 5

Bots (Level, type, highest weapon level)

4 Boa 6

4 Boa 6

4 Patton 6

4 Boa 6

4 Gepard 7

*at time of writing

It’s also worth noting that the matchmaker caught me by surprise like the rest of us. As a result, my sample size of 77 is smaller than I usually like.

header

So here’s what I found. I’ll contrast the pre- and post-matchmaking change side by side for reference. Sample size of post-change data is 122. The “+/-” data for Pre measures the change since my last stats post. The “+/-” data for Post is a comparison to Pre.

Average Player Level (Pre): 20.98 (+5.37)

Average Player Level: 24.74 (+3.76)

This isn’t surprising at all. My last By the Numbers article (“Doom Blade versus Counterspell“) saw me in High Bronze, and I’ve moved into Low Silver with the adoption of the Boa “Brickfighting” strategy (excellent Wiki Forum guide here). With a median value of 21, there was a nice concentration of players around this level without a ton of outliers.

I’ve struggled with the notion of “gear versus experience” before when it comes to defining what “clubbing” is. We definitely know it is gear, but how much veteran experience is “too much?” Since my days playing RPG video games, I’ve always been happy to put in time grinding to make the path ahead a little easier. Now that we’re also noting Victories, this will help put things in a little more perspective on what sort of impact that’s had.

The bump in average level following the change is respectable. I have absolutely faced almost impossible opposition in terms of higher-level bots and weapons, but the change in player level is on par with what you might reasonably expect if Pixonic wants me playing opposition that’s closer to my experience level.

Average Victories (Pre): 172.53 (no change, this stat was previously untracked)

Average Victories (Post): 274.36 (+101.83)

So I had a little more than double the average number of victories of my peer class in Low Silver before the new system was implemented. Consistent with what we saw from Level, the new system has bunched me more tightly with players closer to my experience.

I find it difficult to consider this entirely “unfair.” After all, it’s been my intention to play in lower tiers because I enjoy that experience, but I’ve certainly had abundant opportunity to level up my gear. I’ve simply chosen not to. Earlier, we discussed the motive behind the change, and what I’m seeing certainly substantiates the “proactive” school of thought. I’m either going to have to start substantially upgrading my mechs and weapons, or get used to being much more of a non-factor in games than I’ve been accustomed to.

IMG_3020 (1).png
All for one and one for all, right guys? Guys? Guys…??

Average Trophies (Pre): 110.32 (-31.48) 

Average Trophies (Post): 183.60 (+73.28)

This is interesting. With trophies being a measure of how active a player has been in the last 10 days (indirectly, they specifically measure how much damage they’ve output), it’s fascinating that on the eve of the new system change, I happened to be paired with players who were significantly less active than the ones I’d seen in High Bronze.

Not surprisingly, after the change went into effect I found myself up against a much more contributing playerbase. I’ve endeavored to keep my trophy count at around 300, which is still a fair ways higher than the players I’m facing.

So it looks like under the new system, I’m being paired with “players like me,” but relative hangar strength is being disregarded.

Or is it?

Header.png

Nobody would deny that having a larger hangar offers significant strategic advantage. Not only does it give you greater battlefield longevity, but it also extends to battlefield adaptability. The more hangar slots you have, the more tools you can carry in your toolkit for the right situation.

Percent of Players with Three Hangar Slots (Pre): 77.92 (+4.2%)

Percent of Players with Three Hangar Slots (Post): 57.85 (-20.07%)

The slight bump from week to week is a little unexpected, but the large drop post-fix is certainly as expected. Veteran players know that “slots before bots” is the way to go, and spend accordingly.

I’ve set three as the default here, since that seems to be where most players spend a large part of their War Robots careers. It also seems to be a more illustrative look rather than “average player has 3.19 versus 3.40 hangar slots”. What I find really interesting here are the players who have clung desperately to only two slots. Once player I ran into post-MM change had a hangar of a level 7 Raijin and level 1 Fujin. Surely one of God’s beautiful creatures…

Average Robot Level (Pre): 4.14 (+1.87)

Average Robot Level (Post): 5.73 (+1.59)

Make no mistake, this is a significant upgrade in offensive capability in both cases. An average bot level of around 2.00 seemed to define Bronze Tier, and so nearly doubling that for Low Silver is an encouraging marker of progress.

To see that nearly triple for the post-MM environment doesn’t surprise me at all from what I saw playing, and it speaks to some of the ‘system shock’ the Gepard clubbers are encountering. My poor Level 4 bots could little withstand some of the Level 10 and 11 bots I was facing, though to be fair those were hardly the norm.

I’d like to take a moment for a quick aside here. In my piece on clubbing, I likened encountering clubbers to a shark attack: rare, but quite unpleasant. It was my conclusion that they seemed more prevalent than they actually were, because we tend to overemphasize negative experiences over positive ones. Have an amazing dinner at a restaurant nine times, and an abysmal one once…when asked about the restaurant, what experience comes first to mind?

I cannot dismiss the experience of others based on my own limited observations. However, my data does support my suspicion that things are not as bad as they are being made out to be. 

Which is my default position for almost anything.

Again, I’ve kept my hangar underleveled because I enjoy a lower-tier play experience. Overnight, I’ve gone from being a bot-level peer to being outclassed by 50%. I can understand that’s a bit of a shock to the system.

What it does not indicate, however, is that the system is broken. If the contention that player experience is trumping hangar strength, then I would have faced approximately the same level of opposition had I gone ahead and leveled my bots to 6. And then it wouldn’t be such a shock.

Average Highest Weapon Level (Pre): 4.75 (+0.25)

Average Highest Weapon Level (Post): 6.08 (+1.33)

This tracks fairly well- albeit more flatly- alongside the increase in robot levels. That means the average bot I faced in the mayhem of the early MM fix was a 5.73/6.08, up against my hangar of 4/6’s.

Put another way, my enemies were often a little faster and hardier than before, but for the most part they packed the same punch as I was throwing out myself. Sure I may have been spoiled by punching harder previously, but (occasional outlier notwithstanding) was this really much different than me getting a taste of my own medicine?

What Bots are People Running?

This is one of my favorite parts of the stats write-up, getting to see what other folks are playing in the meta. I’m not sure what’s more intriguing here- the ones that had big fluctuations, or the ones that stayed almost exactly the same!

mechspread

header

There are three things as we close that I’d like to stress. First, that the experiences recorded here are mine alone, and this is not categorical refutation of any other person’s differing experiences. I’ve had folks reach out after some of these blogs and mention how they’re seeing something completely different, and that’s perfectly valid. As always, it’s better to back it up with some data rather than “impressions,” but this is supposed to be a fun game, not a second job. As always, your mileage may vary.

Second, there are a lot of folks claiming that the sky is falling over this. Clearly, that’s not universally true. As you can see here, I’ve seen some differences, but they’re hardly game-breaking. In fact, I’m welcoming the challenge of getting back to the top of the tables after growing accustomed to it for awhile. When we get comfortable, we get complacent. And when we get complacent, we don’t grow.

Finally, and perhaps most importantly, this is still brand new. At time of writing, this change is less than 48 hours old. Drawing any conclusions- even the ones I’ve tentatively drawn here- is quite possibly premature. Already I’m hearing reports from some of the most active players that things start to normalize right around 60 games or so. This makes sense, because it does take awhile for an Elo-based matchmaking system to “find everyone’s place in the pecking order.”

It will be interesting to see the long-term impacts of this. For one thing, it looks like the “tier experience” might be a thing of the past, if hangar strength is not a significant part of any matchmaker calculations. Perhaps Pixonic will open up league systems that simulate the tiers of old, so that those of us that really enjoy that sort of play can enjoy it still.

There are certainly going to be some losers in this adjustment period, notably (but not exclusively) those who got very accustomed to clubbing as their gameplay and allowed their other skills to atrophy. There’s something of the “Grasshopper and the Ants” parable here, as those players who invested their Gold in development of their hangar have been rewarded, while many clubbers are paying the price for living in the moment.

There will also be winners. My six-year-old son Liam loves to play War Robots, but has a bad habit of buying things that push him into a higher tier (most recently a Leo). As you’d expect from someone his age, his stats aren’t the most robust, so this system should keep him way down on the competitive ladder. This is surely a positive development for him, as he’ll be classed with performance peers.

Finally, I’ll be very curious to see how pushed incremental improvement becomes. If our opposition is determined by our success, and every time we upgrade our gear we do so in order to increase our success…are we not ultimately just upgrading our enemies every time we upgrade ourselves?

Tonight’s was a long one, and I thank you for reading. This is a monumental change to the game as we know it, and I look forward to hearing your thoughts. Please feel free to comment below and let me know how you’ve been dealing with the change!

8 comments on “By the Numbers: The New Matchmaker

  1. Indricotherium

    My experience has been more on the bad side, though not terrible. I run 3 main lineups, a 6/8 heavy, 3/7 medium (my favorite for sheer fun) and a light lineup (no geps) for when I need to quickly take care of daily liberate missions. Since the MM change, my med and light lineups are simply overwhelmed. Heavy has been a mixed bag. But it’s gotten me down to the point where all I’m running now is my insanity lineup (my 9yo son (who introduced me to WR this past July) challenged me to equip my bots ‘insanity’ style, no matching weapons and/or ranges) because I can fight hard in them and yet lose in a calm zen manner. Here’s to hoping things improve!

    Cdr. Crimmins in game.

    Like

    • Insanity style? That’s a riot! I had mentioned in the WR Forum the other day that the most Zen moment I’d had in the game was when I took the Beacon Challenge, and loaded up on Cossacks with Ecu shields. The first board was Canyon, and I simply sailed around, flitting from place to place, calmly ignoring the combat raging around me and largely being ignored in return. At least, until they caught wind of what I was up to. Felt very peaceful, very tranquil- and utterly hilarious.

      Sorry to hear the MM is giving you woes. Hopefully it’s on a more even keel soon. Thanks for the comment!

      Like

    • After months of not playing I tried it again. STILL DAMN UNFAIR. Level 5 in the game level 2 destrier with lv 2 guns should not be going against a Leo with thunder. And you still get total crap when these specials go on. This game is crap and the developers are motivated by total greed. Shame on you…..

      Like

  2. sirbiscuit

    I really appreciate the analysis you do. I always find it fascinating, especially the bot usage chart. Not surprising that you’re seeing so many more Griffons, I would hazard a guess that it is perhaps the most overall popular bot in the game.

    I wonder, actually, if clubbers have a lot longer of a way to go to become balanced compared to more normal players. I certainly have seen clubbers post brag hangers with 100% winrates before, it seems like it’s going to take a lot of games to get them back to where they should be- and I’m also not surprised that a 100% WR hanger would be shot into what was previously considered “gold tier”.

    For my part, I have been pushing slowly at the edge of high silver for quite a bit before this update, hovering around a 75% win rate with 6/7 bots. Afterwords? I can’t say that my matches seem very different. The biggest change I’ve noticed is actually when I encounter a Leo- because its HP pool is so big, it’s really noticeable when a player is much higher or lower level. There is definitely a greater variance of player upgrade level in my matches, but I would struggle to call it unfair. It does suck to get stomed sometimes, but no matter what the MM is there’s always stomp matches. I actually wonder, do you think that some people are running into difficult matches now and blaming bad MM instead of realizing that sometimes you just get a bad matchup?

    P.S. I’m more a lurker but I always enjoy your blog when you update. Very informative. =)

    Like

  3. Just found your blog today, and I like it very much. Keep it up.

    Wanted to point out though, that 4/6 mediums are high Bronze, not low Silver. Low Silver is actually quite a bit more unpleasant, allowing up to 4/12 lights and 4/9 mediums, and you know what that means….. (I’m on Android, so I still play under the old MM.)

    Like

    • Thanks for the comment, and welcome! I’ve never found a definitive answer to Yamantau, but conventional wisdom seems to limit it to Silver. There does appear a minority opinion that suggests it’s possible to see it on HB. I see snow regularly, so that’s the mile marker I’ve been going by, in addition to the “common heavies” I regularly run into. (Leos, Leos everywhere!)

      Like

  4. Very informative. Thank you. I run five slots, usually light just for the speed, unless missions require a spot of meds or heavies. Last night ran all dest 1/1 just to relax for a bit, get a few beacons. Talk about overwhelmed! Opposition minimum meds 6/7 ranging up to 10/11. Also couple guys with heavies 7/8. No chance at all. On shot and poof. I was definitely the clubee in a world of clubbers. Last time I did this would play on somewhat equal footing. Looks like those days are over.

    Like

  5. After months of not playing I tried it again. STILL DAMN UNFAIR. Level 5 in the game level 2 destrier with lv 2 guns should not be going against a Leo with thunder. And you still get total crap when these specials go on. This game is crap and the developers are motivated by total greed. Shame on you…..

    Like

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: