3
TL;DR Yes, both in the observed data and under statistical analysis, but it probably does not actually matter much. There is a fancy chart at the bottom.
MOTIVATION
The topic of tonnage imbalance and the match maker come up time and time again. Some insist the tonnage imbalance skews win/loss, others claim that tonnage imbalance does not matter. The former claim usually revolves around how many more assaults one team had than the other and the latter claim usually revolves around how good players offset raw tonnage. A heavier mech does not automatically make for a more effective mech, even with all other things being equal. Every class of mech has its place on the battlefield and any mech can be effective in the right hands. Nonetheless, I wanted to find out what the data actually can tell us when we look at a large number of games played. Essentially I asked, "what is the reality of tonnage imbalances?"
With a large number of games, things like individual player skill, play style, drop composition, tactics, map choice, game mode choice, events, trolls, groups, etc. should be averaged out. If tonnage imbalance was not a factor, the proportion of wins by heavier teams should be 50% (and conversely, the win rate of lighter teams should be 50%). If tonnage imbalance is no better of an indicator than flipping a coin, likely it does not influence the outcome of the match.
DATA
The data from 734 games were collected from 3/09/2021 to 8/25/2021. The vast majority of these games were played by tier 4 pilots, with some being tier 5, and some being tier 3. So a caveat might be that things are different in tier 2 or tier 1 games. Data were extracted via OCR from match summary screenshots and augmented with mech data that were extracted from https://wiki.mwomercs.com/ (with many corrections) and hand-input by me. There is no selection bias here and no cherry picking: these are my own games. I am a constant and I am nothing special, sometimes playing well-practiced builds and other times trying new stuff.
Tonnage difference is calculated by totaling the tonnage of each team and then subtracting the highest from the lowest. The difference is unidirectional; it will always be positive. Then I identify who the victor was and if the tonnage difference was in their favor or not. In this analysis I did not determine if the heavier team was actually heavier than some kind of average or if the lighter team was below some kind of average. Heavier and lighter are simply relative to the opposing force.
The minimum tonnage difference observed was zero and the maximum tonnage difference observed was 295. The lightest drop observed was 560 tons and the heaviest drop observed was 945 tons, with the lightest possible drop being 240 tons and the heaviest possible drop being 1200 tons. The mean team tonnage was about 779 tons and the median team tonnage was 785 tons.
FINDINGS
Out of 734 games, about 56.4% of them were won by the team that had the heavier drop. A heavier drop is where that team has more tons of mech total than the opposing team, irrespective of the actual composition (which is another analysis I am doing, but I digress). This results is statistically significantly different than 50% (proportions z-test, t-stat 3.498, alpha 0.05, p-value 0.000). The 95% confidence interval is about 52.8% to 60%. Yes, heavier teams tend to win more. Note that causation has not been proven here, which is why I say that heavier teams tend to win more instead of teams win more because they are heavy.
HOW OFTEN DOES THIS HAPPEN
It should not surprise anyone that almost all games have a tonnage imbalance. Only 14 out of the 734 games were equally balanced. There are too many variables to balance to get games in a decent time. But how often are the teams imbalanced and generally by how much? The mean tonnage difference is about 72 tons, with the median being 60 tons. Of the 734 games, 454 (about 61.8%) were imbalanced by between 5 and 80 tons.
MORE DETAILS
A heavier team can be heavier by five tons or heavier by five hundred tons. Does it make a difference how heavy a team is? Yes it does. Using the same statistical method, tonnage differences were binned in 40-ton bins and each bin was analyzed for statistical significance. Why 40? I wanted a reasonable number of bins of a width that related well to mech weights, with a reasonable number of observations for statistical power.
It turns out that at the 5-40 bin, while the observed proportion of heavier team wins is about 53.6%, there is not enough statistical evidence to reject the null hypothesis and accept the alternative hypothesis. In other words, I cannot say that the true proportion is different than 50%.
In the 45-80 and 85-120 bins, the observed proportion does increasingly favor the heavier team, but the results are not statistically significant. They are very, very close to being so and with only a little bit of data can cross that threshold. Their 95% confidence intervals contain 50%, but only just barely. From a practical perspective, the true proportion is very likely to be above 50%.
The 165-200 bin is statistically significantly different than 50%. In fact, the true proportion for that bin could be between about 58.3% and 79.8%. The remaining bins are unreliable due to their small sample sizes. Those tonnage differences do not occur often enough and their confidence intervals are so wide. Ultimately, common sense tells us that if lower tonnage differences are different, higher tonnage differences are as well. Still, a claim of a statistically significant difference cannot be made based on the data available.
CONCLUSION
If you are on a team with more tonnage than the opposing team, based on these 734 games, you have a higher probability of obtaining a victory. The more tonnage difference you have in your favor, generally the higher that probability is. As a practical effect, you might see one additional victory for every ten to twenty games where the tonnage was in your favor, depending on the imbalances. Given that most imbalances are mild (about 60% are 80 tons and under, with 36% being 40 tons or under), in the long run it is probably not as much of a factor on one's WLR (and PSR/tier) as one would think.
Can you do anything about this? Other than not dropping in groups with lots of unused tonnage (or infamous and often very ineffective meme light "hunter" packs), probably not. Just play the game, enjoy yourself, and ignore the stats. If you really care about the stats, focus on winning in mechlab, playing better, or maybe find a group, I guess. Don't let your dreams be memes.
