Preseason 5: Popular team compositions

Patch 4.21

As preseason 5 goes on, more and more data accumulates. For the first time, and by analyzing over 50,000 diamond, master, and challenger ranked solo queue matches, we can see specific team compositions being played more than once on recent patches.

From a naïve mathematical viewpoint, I initially guessed that fifty-thousand matches wouldn't be enough to see any specific group of five champions played together more than once. The number of possible teams can be calculated as a binomial coefficient of the number of champions (123 at present) and the size of a team (5). Working out 123 choose 5 gives 216,071,394 (216 million) possible teams. Fifty-thousand matches is not going to put a big dent in that (even with two teams per match, ha!).

Though, of course, players do not pick champions at random. There are two primary mitigating factors that drastically shave the number of different chosen compositions. First, players (especially the high-ranking players in this dataset) typically play in-meta compositions. That is, each team often strives for two solo lane champions, one jungler, and a ranged-ad and support for bottom lane. The other major factor is that players tend to pick flavor-of-the-month champions. Thresh is picked 60 times more often than Urgot, and Lee Sin is picked 50 times more often than Urgot.

So, without further ado, here are the most picked comps in preseason 5 ranked solo queue diamond, master, and challenger matches. I will begin with the most picked comps in the entire dataset. There were two and each was chosen seven times.

With 6 wins and 1 loss, this is likely a very good composition; if not the best solo queue composition on this patch. The ability to smite champions has definitely benefited Jarvan greatly, and the other picks are strong team-fighting champions with lots of damage.

Interestingly, this composition does not have a lot of hard crowd control, though it does have a lot of displacement. Also, Janna + Lucian can bully many bottom lanes, which can give your team an early advantage. An early lead is important for solo queue team psychology.

The other team that was picked seven times in this dataset, (with 5 wins and 2 losses), is this scary team. The only champion in common with the previous one is Orianna, and this composition has better synergy with her skills. Thresh, Irelia, and Lee Sin all go in, which is good for Orianna's Shockwave ultimate.

This team also has a lot more hard crowd control than the previous one, though besides Irelia's stun, it is all skill shots. The high-level players in this dataset are all experts that can make great use of these difficult-to-play champions. It is not surprising to see them being picked and performing well.

Next: The teams that were picked six times.

With 4 wins and 2 losses this hard-engage comp looks quite menacing to me. Syndra is also a popular pick right now in high-level play, and the thought of this team diving in is actually quite horrifying.

All of these picks can also win their individual lane match-ups. The players in this dataset are all skilled at climbing leagues, and it would not be surprising to find out they all think they can do it by winning individually. The differences between solo queue and competitive play are very interesting.

With 3 wins and 3 losses, this is the first popular team composition in this dataset that doesn't have a winning record. For me personally, I find this interesting since this comp looks much less threatening (to me) than the ones above.

There is not as much synergy, and team-fights look like they will come down to exceptionally skillful plays, or landing really good Ahri charms. There are almost no point-and-click skills on this team.

Now: Teams that were picked five times.

Both of these compositions were undefeated with 5 wins and 0 losses.

The top team looks like an ideal poke/siege comp with absolutely insane range. If thresh lands a Death Sentence, or Xerath a Shocking Orb, the follow-up damage is going to be lethal. Also, this comp is going to shred buildings like none other. No surprise that these picks were able to finish out games.

The bottom team looks a bit more subtle. Top-lane Lissandra is strong if she can get some early surprise ganks with teleport. Zed is another high-skill champion that shines when played at this level. Lucian is also appearing in many of these successful compositions, perhaps his popularity is not unwarranted.

With 3 wins and 2 losses each, these two similar compositions also fared well in the matches I analyzed. Neither of these relies on any special synergy apart from Lucian + Janna as described above. Both rely on solo-lane dominance, and landing skillshots to finish out the game.

It is a possibly-interesting coincidence that we only see LeBlanc twice in this entire post, and they're right next to each other.

With 2 wins and 3 losses these comps are popular, but perhaps less successful. Of course, we have to keep in mind here that the sample sizes are not large enough to say anything concrete. There are some different picks here in Rumble, Pantheon, and Katarina.

I think all we can say for sure here is that these are some popular champions in the current meta.

With only 1 win and 4 losses, these teams fare the worst of the popular compositions in this dataset. The bottom team here resembles the successful poke/siege comp that we looked at above, but it has Azir instead of Xerath. I personally think Azir is a strong champion, he certainly does a lot of damage (and from relatively long range) but I think it will take time for players to master him.

The top team here is made up of strong champions, but may be too reliant on Thresh hooks and Syndra stuns to get good fights. The comp scales well, so I would not imagine it would lose many long games. It may be interesting to look at the lengths of these games (though I didn't in this analysis).

So, there we have it. Some popular compositions from preseason 5 of League of Legends. This, like many of the other analyses I have done here at would be fun to do on a much larger set of matches. In recent weeks I have ramped up my data collection, which allows primitive analyses such as this one, but I think I'll always be wanting for more data.

Take care, and do let me know what you think. Until the next installment of league'n and math'n, friends.

No third tier marksman picks were harmed in the making of this article.