You need to cs better; unless you play jungle, in which case you need to gank more

tl;dr Bronze junglers farm too hard.

The LeagueMath.com dataset has ballooned to 415,860 ranked NA solo-queue matches and now provides better coverage than ever of players of all skill levels. There are so many possible analyses with a dataset like this, but it is still probably best to start with some basics. Today I'm going to look at cs/min, gold/min, and wards placed/min per league, and for each role.

When doing analyses like this, it is fun to compare the actual data with your expectations. I thought I knew what these tables would look like, and in many cases I was correct. However, there are some nice surprises here as well, and I think we all stand to learn something today.

Here is the first table, for solo top laners:

All three stats are monotonically increasing as one ranks up. This is perhaps some small evidence that the ranking system works. Better top laners kill more creeps, earn more gold, and place more wards. Though, what surprised me here is how close these numbers are. Averaged over a large number of games, a challenger level top laner only kills about 1 more creep per wave than a bronze player, and earns only a few hundred more gold per game (probably less than a full item).

I suspect this comes down to match-making. The difference in skill between a challenger player and a bronze player is tremendous, but the game (at least these days) does a good job of matching players up with opponents of comparable skill. So, a challenger top laner gets about one extra cs per wave, but against an opponent who is that much better at stopping them from farming. From that point of view, I think this table makes more sense.

Here is the table for junglers:

This is one of the more interesting tables in the history of this blog. First, gold/min is going up, while cs/min is going down. Skilled junglers are getting more gold without killing more creeps. Actual decision-making for junglers is going to depend heavily on champion choice, and this table obviously washes a lot of that out in a big average, but look at cs/min for gold league and above; the values are essentially identical. Skilled junglers quickly converge to some (optimal?) level of farming, and stay there. After that, climbing ranks depends on something other than what is quantified in these simple statistics.

The other thing worth mentioning here is that the best junglers place almost twice as many wards. That is a pretty staggering difference that highlights the importance of playing around vision in League of Legends.

Here is the table for solo mid laners:

This table is less surprising. Higher ranked players farm better. It is interesting to note that the difference in gold/min between ranks is fairly well predicted by the difference in cs/min. It would be interesting to look at some other stats, perhaps kills/min, deaths/min, game length, etc... to see what else could be said about players at different skill levels. I still think that playing carry-style champs and improving farming mechanics is a good way to rank up, but these tables are less dramatic than I thought they might be.

Here is the table for marksmen (adc):

The best ranged ad carries place fewer wards. Ha!

In essentially every league, marksmen kill the most creeps of any role. This makes sense, since the champions typically played in this role have the best kit for last hitting.

Again, a difference of 30 gold/min here does not adequately capture the difference between a challenger and a bronze ad carry. We would have to look at some other stats, or perhaps slice the data some other way to tease out and quantify the differences. Regardless, it is still interesting to compare yourself to these numbers to see where you stack up.

Personally, I had games where I sustained greater than 6.5 cs/min, but that is certainly above my long-run average.

Finally, here is the table for supports:

The stats for cs/min and gold/min here are not super interesting, but perhaps it is funny to see that bronze supports take more farm. The gold/min varies essentially not at all as one ranks up as a support, confirming that the role's contribution lies outside merely getting rich. (late edit: It would be neat to see assists/min.)

Again, the best players here are placing almost twice as many wards. Concentrating on farm is all well and good, but the strategic elements of the game that revolve around the acquisition and communication of information may be just as important.

We find ourselves again here at the bottom of the article. All good things must come to an end. Sadly, the use of platitudes is not a good thing and thus may never end. But I digress. Kind thanks for reading these articles.

This particular post was, I think, enjoyable and edifying. I hope you are happy and satisfied. Get at me on twitter and let me know, the ground is fertile for more analyses. Data inquisition continues, the next post should be the first to source data from over a half million matches.

Go back to the homepage and read another article, or just hover the poro with your mouse for a while. Get up on the rss feed so you can know within seconds of future posts going up.

League + Math = LeagueMath.com.

No MongoDB aggregation functions were harmed in the making of this article.

Peace.