Preseason 5: Support item considerations

What difference, if any, is there between the different choices for support items in terms of gold earned? This question led to the following experiment:

1. Look at over fifty thousand solo-queue games played by Diamond, Master, and Challenger elo (MMR) players in North America (NA).
2. Filter by players buying support items
3. Group by which support item they bought
4. Sort by popularity in each group
5. Measure gold earned
6. Make tables

If only writing the code were as easy as making that list. Well, it is pretty easy, but...

Before we get to those tables, here are some quick stats. In a given game, you should expect to see an item from the Spellthief's Edge / Frostfang / Frost Queen's Claim line approximately 73% of the time, an item from the Relic Shield / Targon's Brace / Face of the Mountain line approximately 62% of the time, and an item from the Ancient Coin / Nomad's Medallion / Talisman of Ascension line approximately 36% of the time.

So, the blue support items are the most popular, and they're about twice as popular as the yellows. These numbers add up to more than 100% because there is typically more than one support in each game, and they don't quite add up to 200% because some fraction of games have no support on one or both teams.

The popularity of the blue support items follows from the current meta-game (meta), I think. They work best on AP mage type supports, and with many compositions being AD heavy (AD jungle and or AD mid and or AD top), teams are looking to the support role to find magic damage. Correlation is, of course, not causality, but one might wonder if the meta is in fact flowing backwards from these support item choices. To attack that question we will need a bit more data.

Without further ado, here is the table for the Spellthief's Edge / Frostfang / Frost Queen's Claim line of support items:

Janna is by far the most popular support pick at the moment, and the blue support items are the most popular choice for her. In agreement with player's choices in general, she is about twice as likely to choose Spellthief's Edge over Ancient Coin. From this table we can see that she earns on average less gold than the other popular blue support item picks (except Soraka), but you do not need to be rich to have a big impact on Janna. Her disengage is essentially unrivaled, and that utility in team fights (coupled with a bit a groupthink among top players) is plenty to explain her popularity.

The gold earned here is quite even over this spread. With a variance of ~10% there seems to be good balance in this area. Zyra's deviation (the standard deviation of her gold earnings over thousands of games with a blue support item) is the highest, but not significantly higher. My intuition is that she is a bit more likely to become super-rich than some of these other picks, but I do not think we can say anything concrete here.

Now, here is the table for the Relic Shield / Targon's Brace / Face of the Mountain line:

Thresh is the most likely to buy a Relic Shield. This is interesting because his basic attack is not considered ranged so the execute passive on the green support items is wasted on him. However, the healing that these items provide gives valuable sustain that is otherwise missing from his kit.

Again, the gold earned here does not vary widely, and is quite similar to the numbers above for the blue support items. There are no obviously overpowered choices from this vantage point. With a much larger dataset one could split this data over game lengths, to see if there were more variance in short, medium, or long games. Though, the length of the game is difficult to predict from champion select, so it is not clear that there is any significant advantage to be had.

Lastly, here is the table for the Ancient Coin / Nomad's Medallion / Talisman of Ascension line:

These are the only four champions with a significant number of games played with a yellow support item. I think it is interesting to note that this list is a mix of the two lists above. Though the other two support item lines are more popular, there seems to be a valid stylistic choice in going with an Ancient Coin on these popular support picks.

None of the support items correlates strongly with winning or losing, and none of them seems to give a profound advantage in gold, so the meaning inherent in the choice is mostly one of passive vs. aggressive play. The yellow support items encourage a more passive play style, since all their benefits can be had without interacting with creeps or enemy champions. I personally prefer the mini-game of trying to hit basic attacks with Spellthief's Edge to watching to see if my marksman makes us both rich when I buy an Ancient Coin. Of course, the extra engage that comes with Talisman of Ascension is super-fun and potentially essential if your team does not otherwise have good engage.

So, to answer the question posed above, I think the support items are well balanced with respect to each other, and are not designed to shape the meta. If Riot did something silly, and (for silly example) doubled the gold earned by the green support line, I would expect to see many more melee AD supports. That could drive up the use of mages in the solo lanes (much to the dismay of Zed and Yasuo), but I don't think they would do that.

There are no obvious changes to be made here, which is a testament to the game's overall balance and goes a way towards explaining why there haven't been any recent changes to these items.

This article went a little deeper into support picks, and some considerations about the support-specific items. We didn't find any super-op freelo tactics this week, but we did see that there appears to be a good balance among the support items, and that the Ancient Coin offers a neat stylistic jumping off point for mixing up support play in the bottom lane.

Did I miss anything in the analysis?

Want to see something similar or related?

Then get at me.

Regardless, I'll be back with more league'n and math'n here at LeagueMath.com, why not subscribe to the RSS feed?

No cannon cs was flubbed with a Targon's Brace during the making of this article.

Peace.