I must preface this entire post with the disclaimer that I am NOT TALKING ABOUT "BONUS RR". It's been addressed by Riot on this sub multiple times, and revealed that bonus RR is based on whether you, "... exceed the expectation we reward you a few extra ranked rating points", which is a direct quote from Riot.

What am I talking about then?

This analysis is looking at raw RR gains/losses, excluding the potential "bonus" RR mentioned above. In the same post linked above a couple of comments down, EvrMoar states that, "You have two MMR's, one for skill and one based on wins. In lower ranks skill MMR matters more, in determining your true hidden MMR. As you climb winning becomes more important. " So, we know this is a real thing; I wanted to know by how much exactly because I feel like I'm getting burned while bottom fragging but playing for the win as controller, while getting rewarded for running around with Reyna.

Assumptions

I'm in mid-silver currently (about halfway through S2). I looked at my last 30-40 games to get an understanding of my average RR gains and losses depending how I performed. This is somewhat subjective, but I labeled games as "high performance" when I had a high K/D (greater than 1.5), and was either the top frag or second place. "Low performance" games were generally when I was either bot frag or second to bottom, and pretty much always negative. Interestingly, I still do see these lower RR gains when I have a around a 1 K/D but bottom frag. Your position in your team and overall lobby seem to have a lot to do with it, rather than purely K/D, so I factored that in as well.

It's hard to perfectly quantify high vs low performance, but this is what I see in my own performance in-game and is what led me to research this. My games generally fall in these two categories, I don't have much in between. If you do this calculation, you may want to throw out data points where you were about in the middle performance-wise.

So, I looked at average RR gains and losses for these two types of games (remember these values are independent of "bonus RR"):

High performance games: Average RR gain on wins: 19, Average RR loss on losses: 15

Low performance games: Average RR gain on wins: 14.5, Average RR loss on losses: 18.5

Well these numbers are pretty much what I inferred from my games; frag more, get more base RR. But how does this actually relate to win rate?

Expectancy

Expectancy is a calculation mostly used in stock trading, to understand if you're actually making money. It tells you what you can expect to make, generally and over the long term, on every bet based on your win rate, and the average size of your winners and losers. In this case, we'll use it nearly the same way to tell if we are "making money" on RR at different win rates. If the number is positive, you are ranking up, if not, you're going down.

The formula is: E = (win_rate * average_win_size) - (loss_rate * average_loss_size)

As an example, assume we have a 50% win rate (and thus 50% loss rate). We'll use the low performance games as an example, where the average winner is 14.5 RR, and the average loser is 18.5 RR. The calculation becomes E = (0.5 * 14.5) - (0.5 * 18.5), which is an expectancy of -2... This means we can't rank up at 50% win rate when mostly bot fragging with a bad K/D, we will be ranking down approximately 2 RR per game.

Expectancy comparison

I did the above calculation for all win rates 1% to 100%, for both the high and low performance games. This table is a snapshot of the most interesting area (nobody really has long-term win rates outside of these bounds anyway). The values given are RR expectancy. In other words, how much RR you can expect to gain or lose every game, generally.

​

It's quite obvious what's going on here. When you are performing well, you can have as low as a 45% win rate and still have a positive expectancy. At a 50% win rate, your expectancy per game is 2RR when you're top fragging and/or performing well! As we saw in the initial example, you would slowly de-rank if you were doing this while bottom fragging. When you look at the values it only gets better for top fraggers as you win more.

Not so good for bot fraggers though... You have to get to a 57% win rate to even break positive on RR expectancy. That's essentially a difference of 12% in win rate between poor performers and top performers to achieve positive expectancy!

Thoughts and interpretation

First I want to make clear that YMMV here, and I encourage you to look at your own average RR gains and losses to get an idea of what this looks like for you and your rank. In the quote above by EvrMoar, we do know this changes to favor winning as you rank up rather than personal performance, however it's not clear exactly where. What does all this mean?

• I think this is bad for the game. At least at my rank, I could tell I was getting punished for not performing well, and rewarded for fragging out, regardless of win or loss. The wins are bigger and the losses are smaller when you're fragging, and the data shows that it pays to occasionally frag out even at the expense of the win. You're better off cheesing kills than surrendering!
• You see advice often saying things like "just play KJ and you can rank up without even fragging". Well, you better hope you have around a 60% win rate or better if you want to do that.

I understand why they do this at low ELO, I'm sure it's largely related to getting smurfs ranked up quicker. However, is it really fair that every rank below Diamond or Immortal heavily incentivizes you to just get kills at any cost, instalock duelists, and bait your teammates rather than playing for the bomb and the win? Obviously the best of both worlds is to play for the win first and play as a team, and also get a lot of frags. But not every role can or should get the frags in conjunction with playing for the team. In fact, it's supposedly not how the roles should work. We're almost playing a different game.

The worst of all of this, is the inability of the system to gauge true impact frags versus worthless frags. It doesn't matter if these kills are all first bloods or multikills to tip the numbers advantage in a round. You can spend the whole round "lurking" while your team dies, and then clean up a few kills, over and over again and you get the same benefit. I know they do measure some metrics for impact frags, but in terms of these RR gains and losses it doesn't seem to change it very much at all from what I can notice.

I hope they adjust this at some point. But for now, I'm very much keeping frags in mind while I'm playing. I still favor to play for the win, but I've been ranking up extremely fast by not playing smokes or sentinals while denying map control and being a complete pacifist. I would prefer to play like that tbh, but Riot isn't incentivizing it in Silver at least currently.