WTC 2023 Statistics

Warmachine
1

Congrats Sweden Asgård for taking the title! Thanks to Longshanks we have some nice statistics from the event. Take a look from here:
https://warmachine.longshanks.org/event/11059/stats/

My take for the most significant results:

  1. First player won 57% of the games. This is no surprise to anyone who plays the game competitively. With the new unit movement rules you can get very far up to table when going first and that puts the one going second into a tough spot. Especially many of the best armies in current meta benefit a lot for going first. Hopefully this will be addressed somehow in the new Steamroller.

  2. Mk4 armies didn’t dominate - quite the opposite. Storm Legion won 51% games and all others went well worse. So the best armies are currently legacy. This is very healthy for the game when some releases are still missing and cadres are still to come.

  3. A few casters might need a tweak downwards. The sample size for many of the armies is way too small to draw any conclusions and is more about the player, but Retribution Shadows and Grymkin stand out at high win percentage and many games played. Sadly we cannot see the casters played but it is almost certainly mostly Thyron and Old Witch. These two stand out even after almost every team trying to build lists against.

What’s your take?

7 Likes
2

That’s actually just a little more than in chess, so I’d say it’s surprisingly well balanced. At least it shouldn’t warrant any drastic measures, or the scales could easily tip in the other direction.

This has been my gut feeling without any solid statistical evidence all the time. My guess is, that’s because the Legacy armies have more options for versatility, and the new Mk4 units are, well, simpler, with fewer tricks and special rules. They may pack more punch to a certain extent, but less finesse.

1 Like
3

I’m on the camp that 57% advantatge for first player is pretty solid and good (chess seems to be at 52-26% using a 3-minute google search :stuck_out_tongue: ).

4

Well the chess reference doesn’t work at all because in chess you specifically play both pieces the same number of times (either against the same player or balanced throughout a tournament). This is not the case in Warmachine. So maybe let’s not continue that.

57% advantage is on territory of significant and certainly should be looked at. The game should reward making the correct decisions and currently going first is not a choice at all but pretty much mandatory given you win the roll.

5

I caution anyone from drawing conclusions based on amateur statistical analysis. :slight_smile:

…Unless we have a professional statistician in the house who can share their professional expertise, then, by all means. :slight_smile:

For all we know, it’s actually a case of “First player has 57% win rate…when the second player is hung over from partying the night before, and due to the position of the table, the light from those really huge windows over there is shining directly in their eyes, making for quite a distraction.”

2 Likes
6

Guard towers and bunkers are the kind of thing to balance the going first tool, but in these parts they have proved very unpopular.

7

57% vs 43% is a huge advantage. Try to multiply .57 together or .43 together a few times, and see how the odds of winning a series of matches changes if you are lucky or unlucky. In many matchups the starting roll is the most important roll you make in the game, and you have no way to control anything about it.

Most armies can now run units (almost) to the middle of the table turn 1, leaving the opponent the choice of not grabbing much - if any - table space, or getting alphaed, both are bad. I had several matches at the WTC where I knew that if I won the die roll I had good chances, and if I lost I was already in trouble.

Many armies can run, then defensive / control feat plus run, scoring zones on their side while leaving the second player scrambling to not lose on points top of 3. I had one game where I had to move AD units backwards bottom of 1 to not get charged. It’s ridiculous, Steamroller has to be fixed for Mk4.

Mk4 is amazing though, best edition ever, miniatures are awesome, keep it coming. Just fix Steamroller please!

1 Like
8

For the sake of pointing this out:

Neither situation described is new to Warmachine.

This was the de facto way Advance Deployment worked in MKs I and II: player 1 would AD to the ~17” line (or greater) and player 2 would AD to the ~20” line (or greater). Casters with defensive feats would charge one of their own models and feat on turn 1.

MK III reduced AD to only an additional 6”, though one still saw the same effect under certain conditions.

9

I played mostly Mk3, where this was less of an issue - no feat and run on the caster, no 2" + base extra unit movement twice. Going second often did give a good chance to score first. I don’t see why it’s relevant what the rules were more than 10 years ago. The question is if going first is too good right now, in Mk4. I think it is, Steamroller needs fixing.

1 Like
10

A professional statistician would probably consider a single tournament’s data as anecdotal evidence, and would refuse to draw any conclusions based on that alone.

It’s also worth noting that a team tournament is a different matter from single-player tournaments because the metagame level has an extra layer in it.

1 Like
11

Pjolterbeist is describing the situation 100% correct. This discussion so far clearly shows who has played competitive Warmachine lately and who hasn’t.

It is totally different that in mk2 or so you could get charged by some Advance deployment units while these days you can get charged by almost the whole enemy army if moving forward. The new unit movement both gives additional speed to cover ground after running on turn 1 and if one model is within charge reach you can get everyone in. This is a drastic change increasing the starting player advantage.

And just for the record in mk2 it was very common to decide going second (but many armies preferred to go first also then). It was partly because of the Steamroller back then was race to 5 control points and not 5 control point difference. Getting to score first was important. And of course the unit movement was different.

Some changes to reduce the 1st player advantage could be done. One might be to disallow charging and running for the first player 1st turn. Then you migth be able to increase the deployment zone a bit too.

2 Likes
12

Interesting how Michael keeps denying any data he doesn’t like while providing no data on his own (like in the Stormsmiths thread). Even bad data is better than someone’s gut feeling and I fail to see how a stance of “I have made the belief that the game is perfect a part of my identity so much that, unless you totally overwhelm my gut feeling with copious amount of data and evidence I will always fight any notion to the contrary” is rational.

4 Likes
13

In this case it is the statistics AND the gut feeling of the vast majority of competitive player base…

I mean if you are not interested in competitive Warmachine there’s nothing that forces you to even touch the Steamroller package. You can play with your friends just fine without.

2 Likes
14

Whether the result is significant depends entirely on how many matches have been played, how large the winrate is, and the base chance of winning. Did the math here: Lightning Shroud: Going first advantage of 57%: The math - as mentioned I am not a statistician, but I am fairly sure this is how it’s done.

TLDR: The probability of having a 57% winrate for going first happening just by random chance, with 292 games played, is 1.1%.

1 Like
15

AD was actually reduced to 6" at the start of MK.2 and was kept that way through the entirety of MK.3

16

I could add to this that Shadows of the Retribution going 34-17 has a 1.2% chance of being due to random chance. So if you believe that going first does not matter, you should also believe equally strongly that Thyron unit spam is fair and balanced after winning 2 out of every 3 games - even with most players being aware of this army being completely broken and trying hard to tech into him.

2 Likes
17

Exactly. :slightly_smiling_face: Personally, I’d find a player-by-player analysis of W/L when going first or second much, MUCH more reliable of a measure. I am not a professional statistician, but I know enough to say that I’m confident the data set is insanely limited and all the data necessary for a good analysis are not recorded.

(But, whatever. Hand-wringing shall commence, regardless.)

1 Like
18

Please correct me if I’m wrong*, but that analysis is correct only with the implicit assumption that the probability of winning each game is exactly 50/50, and that the only confounding factor** was the starting roll.

But that isn’t a correct assumption, is it?

The whole WTC team format lends itself to stacking the odds in your team’s favor as much as possible, right? Nobody’s trying to stack the odds so every single match is exactly a 50/50 outcome, they’re trying to play their strongest list/strongest player to secure the easiest/most likely wins they can, while minimizing their team’s number of losses, yeah?

Interestingly enough, your Shadows of the Retribution example is probably a much stronger case of something being out of balance. Presumably, unless teams all decided “Screw it, this is too hard, throw this one so we can stack our chances elsewhere” when this matchup came along, they were trying to stack the odds in their favor, and we still see a rather non-trivial imbalance in the W/L here.

So, again, just as a general note on critical thinking: check those implicit assumptions, because they’ll lead you astray. Statistical analysis is difficult to do properly, and, to say it in the form of an aphorism, amateur work creates amateur results.

* And I’m not just saying that as a rhetorical device. Seriously: share the statistical expertise if you have it, so we can all learn.

** Is that the correct term? It’s been a while, I’m not going to look it up right now, and again: not a professional statistician. :slight_smile:
19

OK, I’ll try to explain it another way.

The probability of a player winning a game is always 50%, because half the players win, the other half lose each game. That’s where the 0.5 in the formula comes from, and it will be for any game where there are two players and no ties, whether it’s flipping a coin, armageddon chess, Jenga, beer pong, or Warmachine.

There are many factors affecting who actually wins each game. But these do not matter in the calculation. If a player is better, drunk, was matched in a particular way, or has a better painted army or cool shirt is irrelevant. A game begins, you roll off - this roll is completely independent of all other factors coming before it - on average in the WTC the starting player now won 57% of the time.

So far, we have just been counting. Here, we start with statistics.

If starting had no effect on winning, we would now expect to see starting players win approximately 50% of the games. The further we get away from this number, the greater the odds of it not being by chance.

Starting players really did win 57% of the games, and we can calculate the chance if this happening by using the binominal test. The chance of the starting player winning 57% or more of 292 games by random chance is 1.1%. Or, in other words, if starting does not matter, and we play the WTC with 24 teams for a hundred years, we would expect to see this win rate one year, and not see it the other ninety-nine.

Saying statistics are hard and that you or I don’t understand statistics is not an argument. We can roll pairs of dice many times and calculate that the odds of a 7 is 1 in 6. This is statistics. We both know this and it’s not hard, but it is still statistics. The binominal test is not hard, either. You count the things that happened, stick the numbers into the formula, and get the chance of the result happening by chance.

I did actually study statistics and math at uni a long long time ago, but I hardly see why this should affect whether the calculations are correct - they are correct if they are correct. If I did something wrong in the calculations I’m of course happy to be corrected, it’s been a while since I used this and I could have forgotten something.

4 Likes
20

Why are there 51 games recorded for Shadows of the Retribution? (I guess those unfair elves stole one from the poor Soldiers of Fortune, who already didn’t have a particularly good tournament.)

Some of the numbers don’t add up (to 300, the expected number of games).

Elves (both armies) did pretty good, as did the Undead (Banes) . So why didn’t the undead elves succeed? (The obvious answers might be that the army wasn’t fully released yet and that Warmachine doesn’t work that way, but that’s just no fun. )