How to put the game back on track: an empirical approach
Hello,
Before I start outlining my comments and suggestions, I would like to apologize in advance for any grammatical mistakes AND FOR THE F**** LONG POST. English is not my first language.
Let's get down to business:
The idea of using the stats (statistics) that the game generates is a really useful tool to gauge whether or not the game needs balances, fixes and etc. However, in my humble opinion (not so humble as you'll probably think as soon as you finish reading it), you - Riot - are using statistics in a very deleterious and obnoxious way that, instead of promoting passion and excitement for the game, you are doing just the opposite.
Proofs of Concept:
"The Economic Theory" of LoL It is widely assumed that advances in technology usually comes with a huge burden: countries are, on average, not able to follow up the pace of the market/industry. Thus, regulations over practices that should not happen - because they are harmful to society - usually take years to be approved. When the legislation is finally approved, the market/industry - foreseeing such measure - had already found another solution to overcome such a measure. However, whenever the government tries to prevent such practices before they even happen, it backfires and hinders progress. There should be a balance, which is really difficult to find. (bla bla bla)
"WTF does this have to do with LoL, you moron?" you might be asking right now.
The answer is EVERYTHING.
Riot is acting exactly like the "extremely protective government" whenever they buff, nerf, or change champions, items, spells, preventing the community to work at its peak. This is economic theory dude, anything below the "optimal" point is inefficient.
"Why?" Because, since the game changes every other week (literally) the community tries to overcome the burdens or make use of the benefits.
"What's the problem with that, it sounds exactly like the game is supposed to be, right?" WRONG! Whenever a new patch comes into force, the community takes, ON AVERAGE, 1 month and a half to adapt to this new scenario. During this period, the community is still facing problems, trying to deal with strategies like (Karthus+Nunu / Taric + Yi / etc), losing, winning, flamming, blaming, crying. After this period, in theory, higher elos like challenger (due to the frequency of games) had already seen a good amount of cases in which the strategy either failed or worked. Thus, they will act accordingly.
THE BIG PROBLEM IN THIS SITUATION IS THAT RIOT IS NOT ALLOWING THE COMMUNITY TO HAVE TIME TO FIND ALTERNATIVES AND RESHAPE THE "META", BECAUSE WHAT IS BUFFED IN ONE WEEK, WILL CERTAINLY BE NERFED, IF NOT 1, THEN TWO PATCHES LATER.
"WE ALL KNOW THAT JERK" Yeas, but have you prove that statistically? Have you proved this using numbers? You getting ganked and killed, because you did not place a trinket, in addition to not paying attention to the mini-map and hence blame your team, has the same effect here. (I'm being very optimistic, I know) In one week if a champion/item/ is buffed and its winning rate goes up (which should be straightforward since none had previously dealt with such a scenario), in a space of 1 to 2 patches, on average, RIOT will nerf the champion/item again, creating this infinite loop. (This information, although painful to compile, is publicly available on patch notes Rioters).
Empirical Evidence:
I estimated an LPM model (Linear Probability Model) to support my argument that Rioters are probably looking to the stats wrong. Basically, this model is trying to predict a new patch giving: winning rates, the frequency of champions in ranked games, the frequency of items bought (excluding trivial items like Potions or Wards because it might bias the results), and champions appearing in a different role than those established by RIOT.
DISCLAIMER: SINCE I'M NOT USING OFFICIAL DATA, THE RESULTS MIGHT BE DIFFERENT. HOWEVER, THE IDEA IS TO PROVIDE AN INSIGHT THAT RIOT IS NOT PROBABLY TAKING INTO ACCOUNT THE FACT THAT THE GAME IS TOTALLY BASED ON THE IDEA OF DEPENDENT PROBABILITY. THUS CHANGING ONE CHAMPION WILL CERTAINLY HAVE AN EFFECT ON ANOTHER, WHICH WILL NOT SOLVE THE PROBLEM OF BALANCING THE GAME
The picture below is an estimate of the model that I built to support my argument. The results should be interpreted as probabilities.
As you can see, in the picture (link below), using the data available at op.gg for all ranked games for the past month, winning rates and the frequency of champions are the determinants of a new patch. What is really interesting to notice, is that everything else 0, the probability that Riot still implements a "new patch" is almost 50%. (The Constant term)
https://s15.postimg.cc/5lqtkd6az/LPM_-_OPGG.jpg
Following this logic, MF, for example, has the highest win rate, according to OP.GG as of this moment with 54.64%. Does it mean that MF is broken? Does it mean that it should be nerfed?
No, it means that given the current circumstances, MF has the highest winning rate because players either don't know how to play against it (for now) OR because other champions due to the current situation are really weak to fight against her. This seems to be pretty straightforward since crit items have become much more expensive, allowing lethality builds or other builds to be comparatively strong. But again, I'm showing that with empirical evidence. (at least, I'm trying to)
My Suggestion of Balancing:
-
"Let the market solve its own problems" Give time to the community to adjust to the new scenario and then make corrections based on the deviations from what is expected. You're not giving time for the changes to give the intended results. "Stop it, get some help."
-
Use a Difference in Difference Approach BEFORE IMPLEMENTING PATCHES" I honestly think that a method such as a diff-in-diff or something that can approximate a randomized control trial can definitely work in this scenario. The intuition is pretty simple: How this champion would have behaved had it not had this "buff" or "nerf"? How would this change affect the meta? Other lanes? Other Champions?
"WE CAN'T THINK ABOUT EVERY POSSIBLE SCENARIO" - RIOT, 2018
YES, YOU CAN (or at least try). Information technology era, machine learning technologies, artificial intelligence systems? Does that ring a bell?
How? Here's how:
Let's suppose you are thinking about nerfing MF, for example, because of her absurd winning rates. First: use the period in which she spiked in winning rate and assigned that as the "Control Group", establish the nerfs as the "Intervention", and forecast her winning rates after the intervention. If the target is to balance the champion, anything above or below 50% is a bad "intervention". (Since this is a panel data remember to use fixed effects to remove unobserved effects like player skill and TIME VARIANT EFFECTS)
"Why did you capitalize the last part?" Because one champion or strategy might be strong at the moment due to the fact that other champions or strategies are really weak. Dependent probability, AGAIN.
I know that the balancing team certainly have these and a lot of other issues in mind that I can't possibly think of. I do not really think that I have the solution for this situation. However, with the constraint of information that I have right now, it really seems to me that what is really lacking is basic statistical knowledge (and a lot of other things to the rest of the community). Especially because this is a strategy game with moves and counter moves. Therefore, the probability of nerfing one champion will increase the probability of making another champion stronger as a "backfire". If this isn't logical, I do not know what else is. I don't mean to offend any Rioter or to pose as the "smartest ass in the room", that is not the point, but it really seems that you (Riot) lost this basic notion.
All I ASK RIOT IS TO LOOK HOW: WE ARE PASSIONATE FOR THIS GAME, LOOK HOW FAR WE ARE ABLE TO GO FOR IT (COLLECT DATA FROM A NON-OFFICIAL SOURCE, WRITE A REALLY LONG AND BORING POST, EXPOSING OURSELVES TO CRITICISM, RACISM, VANDALISM, BULLYING, BEING CALLED A JANNA MAIN OR A HEIMERDINGER PROGENY). SO, PLEASE, LISTEN TO US.
Peace