Gold Share is the percentage of the team’s total gold that goes to a particular role/player. The appeal of gold share as a LoL statistic seems clear enough: it can show where the gold goes on different teams.
But before simply assigning a percentage to each player and starting our interpretation, we need to ask some important questions. What is gold share actually measuring? How should we calculate it? And maybe most importantly, is there enough variation between players/teams to make it actually useful?
1) DISTRIBUTION VS. CONTRIBUTION
Fundamentally, is gold share a measurement of a team’s distribution of resources, or does it really measure each player’s contribution to the team’s total gold? Ultimately, it’s both!
In LoL, gold is not just picked up off the ground and passed around among team members. Gold comes from three places: creeps/minions (farm), towers and Baron Nashor (objectives), and kills/assists (combat). (There’s also starting gold and inherent gold per second but we’ll leave those aside for now.)
Farm gold is sitting on the map waiting to be gathered, so we can say that it gets distributed by teams intentionally.
Objective gold sits on the map as well, but it isn’t really distributed around: it contributes gold to the whole team.
Combat gold, like Cho’Gath and Kog’Maw, comes from the void: when a kill happens, gold is created out of thin air. If the kill was assisted, even more gold is created.
So some gold is distributed by the team to specific prioritized team members (by giving them jungle camps or creep waves to farm), but other gold is created and contributed to the team’s overall strength (by earning towers, Baron, or kills).
To complicate matters further, each source of gold could be argued as being both distributed and contributed. How well is each lane CSing (contributing)? Who stands near the turret as it goes down, to get the local bonus gold (distributing)? Who gets the last hit on a kill (distributing)? It’s messy!
What does this all mean? It means that we need to interpret gold share as a measure of both gold distribution and gold contribution, which means we can’t clearly peg it as either of those things. We need to be careful about any claims we make based on gold share.
(We may be able to use some different statistics to separate distribution and contribution, but for now we need to put the distribution vs. contribution question aside and simply acknowledge that it’s a complicated question.)
2) CALCULATION
How is gold share calculated?
For a single game, gold share is simply a player’s total gold divided by the team’s total gold, producing a percentage.
What about when calculating across multiple games? Gold share across multiple games can be cumulative (player’s total gold from all games played, divided by team’s total gold from all games played), or it can be averaged (player’s share of team gold within each game, percentages averaged across all games).
I use average gold share, so that each game is weighted the same (longer games don’t influence the numbers more than short games).
Further to this, I calculate gold share using earned gold rather than total gold. (Earned gold is total gold minus 475 starting gold per player and minus inherent gold generation of 1.9 per second starting at 90 seconds.) See below for the difference this makes in the final numbers.
3) VARIATION
How much variation exists in gold share by role between teams? Is this stat really going to teach us anything if everyone gets more or less the same percentage of team gold?
To answer this question, we need to look at some real data. The table below shows Gold Share statistics by team, using total gold in the calculations.
Gold Share (Total Gold)
Based on weeks 1 to 7 of the 2015 NA LCS Spring Split.
[table “5” not found /]
Using earned gold rather than total gold, therefore producing Earned Gold Share, results in the following changes.
Gold Share (Earned Gold)
Based on weeks 1 to 7 of the 2015 NA LCS Spring Split.
| Team | Top | Jungle | Mid | ADC | Support |
|---|---|---|---|---|---|
| Cloud9 | 19.5% | 20.6% | 22.6% | 26.0% | 10.7% |
| Counter Logic Gaming | 18.4% | 18.3% | 23.9% | 28.2% | 11.3% |
| Dignitas | 22.0% | 16.0% | 24.9% | 27.4% | 9.7% |
| Gravity | 19.7% | 17.4% | 23.6% | 28.9% | 10.3% |
| Team 8 | 21.9% | 18.0% | 24.9% | 24.8% | 11.0% |
| Team Coast | 21.3% | 17.0% | 24.6% | 27.9% | 9.2% |
| Team Impulse | 20.3% | 18.6% | 26.3% | 23.5% | 11.3% |
| Team Liquid | 20.2% | 17.8% | 24.9% | 26.7% | 10.3% |
| Team SoloMid | 19.2% | 17.8% | 25.0% | 26.6% | 11.4% |
| Winterfox | 19.2% | 15.9% | 24.9% | 29.5% | 10.6% |
Basing our calculations on earned gold rather than total gold, the differences between roles are more pronounced, making the table more informative.
For example, look at the ADC role: using total gold, there’s a difference of 3.4% between the ADCs with the highest gold share (Gravity’s Cop and Winterfox’s Altec and Paragon) and the ADC with the lowest gold share (Team Impulse’s Apollo).
But using earned gold, the gap widens to 6.1%, with the Winterfox ADCs pulling away into first by 0.4%. We now have a lot more room to make comparisons.
Conclusions
Gold share is best calculated using earned gold, not total gold, and should be an average, not a cumulative measurement.
Gold share is a messy metric, and we have to be careful how we interpret it. It isn’t a simple measure of distribution, nor of contribution. Analyses based on gold share should probably incorporate some other factors, such as CS numbers, KDA ratio, or share of damage to champions, to provide context.
Have any questions about this statistic, or suggestions for other metrics to explore? Leave a comment or send a note to mag1c@oracleselixir.com.