The relative value of gold and dragon leads in pro LoL over the past 3.5 years

Turret Plating and Early-Game Variance: How well did Riot manage the metagame for the 2019 spring split?

Turret plating was introduced to League of Legends for the 2019 season as a way to improve and extend laning phase. That change, along with some other factors, had the potential for huge impact on the pace and metagame of pro LoL. After a full split of domestic and international play, it’s time to evaluate the results to see whether Riot’s changes produced the intended effects, and just as importantly, whether or not they created any collateral damage to the game’s balance.

Spoiler: things worked out pretty well!

Where did the shift come from?

In a development update going into the preseason, Riot stated:

“[Turret plating provides] a more protected and slightly longer laning phase, but still rewards those early push or strong lanes types with the opportunity to destroy a lot of barricade segments and reap the gold rewards.”

Essentially, the goal was to make the first 15 minutes of the game less volatile, but still give it a high influence over the outcome of the game. There was also a sub-goal of bringing down game lengths, partly seen in Riot saying that they wanted to see “decided games resolve faster.”

The goal was not to make the laning phase or early game a more influential part of the overall game flow, and that is the most important aspect of what we’ll evaluate below.

Decreasing game lengths

Game lengths are the simplest piece to measure, so let’s begin there: the 2019 Mid-Season Invitational produced the shortest average game time (AGT) for a major international event (MSI or Worlds) going back as far as the Oracle’s Elixir database reaches (which is Worlds 2014).

Click for larger image

In fact, a look at pro scene game lengths shows that game lengths have been fairly consistently decreasing over the past several years, other than occasional short-lived spikes.

Click for larger image

Shorter games can be beneficial for esports viewership, especially if it means that there is a higher pace to the on-screen action. Measurement of that on-screen action goes beyond the scope of this article, but as an aside, there has definitely been an up-tick in action levels and combat, producing some very exciting games.

As far as the goal was to bring down game times, mission accomplished!

Accelerated incomes?

The creation of turret plating had the potential to inject gold into the game that simply didn’t exist before. That has, in fact, taken place. The amount of gold entering the game has been climbing fairly steadily over the past several years, and that growth continued during the 2019 spring season. Overall, it manifested more as an extension of the trend, with a big spike during the first two patches of the year before a bit of normalization afterwards.

Click for larger image

This chart only shows the trend for 15:00, but the pattern was consistent for 10:00 and 20:00, as well.

Early-game variance

Regardless of how much total gold is in the game, or where it comes from, the most interesting question to me is how much early-game variance there is between teams. In other words, what is the average size of the gold lead for the team that is ahead? If the early gold leads are increasing, then that suggests that the early game is becoming more influential over the game outcome.

Lead sizes did grow over the course of the spring split, but they were not really larger than the leads we had been seeing over the past few years.

Click for larger image

Again, this chart reflects 15:00 but the trends hold to the same pattern for 10:00 and 20:00.

What this suggests is that even though more gold is entering the game, there is not necessarily more volatility in how much the game is getting decided early on. But to be really confident in that conclusion, we need to also explore how influential the gold amounts are over the game outcome. Conveniently, Oracle’s Elixir has a tool that directly explores this question.

I developed the early-game rating (EGR) model in 2015 as a way to convert game state into a win probability for each team. The model allows us to see the relative win probability value of a lead in different game metas, as shown in the chart below.

Click for larger image

The value of a 1000-gold lead did not noticeably increase in the 2019 spring split, compared to the previous three years of play. As a side note, the value of an early dragon was the highest it has been since early 2017, presumably due to the addition of higher stat bonuses from the buff on the first dragon of each type.

Overall, this is a very encouraging result, because it means that the changes Riot put in place achieved their goals without introducing drastic, unwanted side effects.


Through the introduction of turret plating and some other adjustments, Riot successfully lowered game lengths and increased early gold flow, which creates additional acceleration into the mid-game and invites more action. In doing this, they crucially did not create a game meta where the first 15 minutes were more decisive than desired in determining the game’s outcome.

I can honestly say I didn’t expect to find such positive results when I began exploring the data from the past split, but I’m excited to see that League of Legends continues to move in a good direction in its watchability and game balance, and that Riot is doing an effective job of intentionally managing its game to be a better esport, and not just a better game.

Join the Discussion

This site uses Akismet to reduce spam. Learn how your comment data is processed.