This article was originally published on GameSpot's sister site onGamers.com, which was dedicated to esports coverage.

　　In competitive matches, one of the most mentioned statistics on a competitive league broadcast is the current gold of each team and the lead. I wanted to dig in deep and calculate how much leads matter in the current game, and with the extended match history from Worlds we were able to get the data to do this.

　　

What is the win % of a team with a lead?

The first thing we looked at is just the raw win percentage of a team when they have any lead (even as little as 100g) at any time in the game, this was the resulting graph:

　　Note: After 50 minutes things get wonky because so few games made it that long.This graph shows a few things. First, leads as early as 8 minutes (the second blue/red buff) have 70% win rate, showing that even what is probably very small leads create significant likeliness of victory. Second thing to notice is that things escalate quickly and then plateau at around 80-85% from 11 to 40 minutes. This is a pretty substantial win rate and it is consistent for the majority of the game. The last thing to notice is the downward trend after 40 minutes. While there are significantly fewer games post-40 minutes, this trend leads to the idea that if you have had a lead and haven’t ended it by then, the game must be close and back-and-forth, or there is a throw brewing.

　　How much does a lead matter for the top teams?

　　While we know Samsung White was a dominant early team, there were some other teams like Samsung Blue and EDG who were known as late game teams who didn't quite dominate early but could play well from behind. I wanted to test these theories and broke down the Win % with a lead, the amount of games that a team gets a lead (Lead %), and their win rate when they don’t have a lead for the top 8 teams.

　　These graphs include only from minute 6 on.Here we see definite strengths and weaknesses between teams. SSW, SSB, SHR, and TSM rarely throw leads when they get ahead, meanwhile teams like C9, EDG, NWS, and OMG have a below-average closing rate when getting a lead. However, the only team that has both good closing rate and good “comeback” rate is Starhorn Royal Club, but they get leads in a pretty low amount of games compared to the other finalist SSW.

　　Starhorn Royal were really the comeback kings at worlds, and had the biggest lead that they overcame against OMG where they were behind 9k gold at 33 minutes, a 30% lead for OMG.

　　

What % lead is a “significant” lead?

The last thing we wanted to look at is what exactly was a significant lead. We’ve always theorized in house at onGamers that a 10% lead was ‘significant’, and it was only a “comeback” if you were behind greater than 10%. However, we wanted to dive deeper and see what percentage lead is needed in order to get an expected win rate of 90%.

　　So, here is a graph that represents the percentage lead you need in order to have a 90% win rate:

　　While our 10% guess was not crazy off, this graph did bring a bit of a surprise. At ~15 minutes things start taking a dive down, meaning that 90% of the teams that have at least a 2.5% gold lead at 20 minutes go on to win the game, a pretty surprisingly low gold value. In terms of raw gold, that is only a 750 gold lead (teams average 29.7k gold at 20 minutes). To be clear, a team with only a 750 gold lead at 20 minutes had a 90% chance to win. 750 gold doesn't seem like very much at all, especially at 20 minutes.

　　I also wanted to look at not just what % gold lead is needed to secure that 90% win rate, but how many games did one team have this strong gold lead, or rather how many games were “decided” by each time. And so I created this minute by minute graph:

　　This shows that the competition is a bit more healthy in the early game, especially before 10 minutes. However, by 20 minutes most games are decided, and if a game goes on past 30 minutes things start to get dicey.

　　

Conclusions

So, with all this analysis into leads we circle back to the original question, just how much do leads matter? I think that while gold leads are a strong predictor of who will win the game (especially in the 15-30 min mark), I believe that gold leads are more representative of who the better team is. A 750 gold lead isn't enough to singlehandedly snowball a game, instead its that the better team on average gets more gold by the 20 minute mark, and thus will eventually win the game. If anything, it isn’t the gold which will snowball a lead but map pressure, and unfortunately we don’t have the data to break down map control, yet.