¡Apagando las luces!

The soft tapping of fingers and beep-booping of a hack completing.

A figure in purple, dissolving into thin air with a sliver of health.

These are the sights and sounds that haunt Overwatch League players. Contenders North America team Fusion University started the conversation by playing Sombra and Doomfist into XL2’s 3-3 in the Season 2 Finals—winning in spectacular fashion. Sombra, with her ability to lock heroes from using any of their abilities, was an easy solution to a complex 3-3 problem.

However, Sombra has not turned out to be the one-hero-fixes-all solution to the 3-3 problem since Fusion University’s win. In the first two stages of the Overwatch League’s 2019 season, several teams have attempted to use a Sombra 3-3 adaptation to counter “traditional” 3-3* by subbing out D.Va for the hero instead. The results have been... mixed at best.

San Francisco Shock main tank Matthew “Super” DeLisi explains that Sombra 3-3 is a crutch

* From here on, Sombra 3-3 will refer to the lineup of Sombra, Reinhardt, Zarya, Brigitte, Lúcio, and Zenyatta, and traditional 3-3 will refer to the lineup of D.Va, Reinhardt, Zarya, Brigitte, Lúcio, and Zenyatta.

Though Super’s comments came late in Stage 2, some teams had already taken—or disagreed with—his advice. The Dallas Fuel and Boston Uprising essentially swapped philosophies by trading Lucas “Note” Meissner and Richard “Rck” Kanerva between Stages 1 and 2, effectively trading strategies as well. Boston played literally no Sombra 3-3 in teamfights in Stage 1, but jumped to 83 teamfights played in Stage 2 after Rck’s arrival. Dallas, on the other hand, went from playing Sombra 3-3 in 205 teamfights in Stage 1 to only nine in Stage 2. Beyond Dallas and Boston, it seems like every team has an opinion about traditional 3-3 vs. Sombra 3-3. Yes, even San Francisco.

The question is still open: which comp is better? Is one comp better? Let’s try to find out.

During the break, I’ve been upgrading my tools to fully integrate team compositions into teamfight statistics. These improvements have unlocked a new level of analysis into the success or failure of hero compositions, which can help us dive deeper into what traditional 3-3 vs. Sombra 3-3 looks like. Let’s start broad with some Sombra 3-3 stats for the first half of the season:

  • Across all teams, versus all lineups, Sombra 3-3 has a 58.7% teamfight win rate.
  • This win rate rises by an average of 7.9% to 66.6% when EMP is used.
  • Against traditional 3-3, Sombra 3-3 win rate drops to 54.5%.
  • Against traditional 3-3, Sombra 3-3 with EMP win rate increases by 6.7% to 61.2%.

These zoomed-out insights already paint a picture of why teams might want to run Sombra 3-3. This season, only the Excelsior (58.7%), Titans (59.4%), and Shock (60.2%) have averaged higher teamfight win rates than the league-wide Sombra 3-3 average. It’s partly why they’re the “Big 3,” after all. Additionally, the potency of Sombra 3-3 was stronger against other comps than against traditional 3-3, but the effectiveness of her ultimate remained mostly the same. The fact that the league’s increase in teamfight win rate only decreased by 1.2% against traditional 3-3 while the overall win rate dropped by 4.2% implies another truth about Sombra 3-3: it fares worse than other compositions in neutral, non-ultimate teamfights.

Why would this be? Remember that to add Sombra to a 3-3 lineup, you have to subtract another hero. Most commonly, the hero that is subtracted is D.Va, and with no D.Va, the team running Sombra 3-3 loses a vital layer of protection from the small but meaningful amount of long-range damage that traditional 3-3 can deal. As a Sombra 3-3 team, you face an enemy Reinhardt that can use Fire Strike on cooldown, a Zenyatta-D.Va-Zarya trio that can quickly break the shield of your Reinhardt, and a traditional 3-3 team that can play far more aggressive on the whole—provided your EMP is not yet ready. In practice, this leads to a lot of “pressing W” for traditional 3-3 teams:

Teamfight first-elimination rates and first-elimination win rates
Teamfight first-elimination rates and first-elimination win rates for traditional 3-3 vs. enemy Sombra 3-3.

This is what pressing W looks like in a statistical sense: high rates of first eliminations. Nearly every team that had a >50% first-elimination rate as traditional 3-3 vs. Sombra 3-3 exhibited a higher first-elimination rate and lower first-death rate in such teamfights compared to their season average. Out of the five teams with a >50% first-elimination rate, only the NYXL fared worse against Sombra 3-3 than their season average; they were also the only team with a >50% first-elimination rate that had a sub-50% teamfight win rate in 3-3 vs. Sombra 3-3. All told, only the Charge, Titans, Valiant, Shock, and Gladiators had positive teamfight win rates in this specific matchup.

The NYXL’s (relatively) poor performance—and many others’ performances at this level of scrutiny—was largely due to who they faced when playing 3-3 vs. Sombra 3-3:

  • The NYXL played 39 total teamfights of this nature, but 30 of those teamfights occurred against the Seoul Dynasty, and they only won 36.7% of those.
  • On the flip side, the Valiant’s great (but confusing) performance in 3-3 vs. Sombra 3-3 came mostly against Houston (33-out-of-55 teamfights), but they only won 48.5% of those. The Valiant instead won most of their 3-3 vs. Sombra 3-3 teamfights against the Hunters, Charge, and Fusion, where they recorded a scorching 71.4% teamfight win rate.
  • The Shock also played the majority of their matchups versus a singular opponent (42-of-74 against the Dragons) but posted a very good 59.5% teamfight win rate against them.
  • The Charge played all 10 of their matchups against the Fuel, winning six of them.
  • Finally, the Gladiators and the Titans did not face one specific opponent who made up most of their total, but had a 56.5% and 57.5% teamfight win rate against the comp, respectively.

*I’m mostly joking—this is an oversimplification of aggressive positioning and engagement timing.

This matchup dependency made me wonder—maybe both sides of the Sombra 3-3 argument are correct. From some teams’ perspectives, maybe Sombra 3-3 is indeed an inferior comp; they know how to counter it and have the personnel to execute the complicated “press W” strategy*. For other teams, it makes sense to keep running Sombra 3-3—just not when they’re playing the “good 3-3 teams,” as Super would put it. Let’s define those as the five teams from above that had a >50% teamfight win rate against Sombra 3-3: the Charge, Titans, Valiant, Shock, and Gladiators. Then let’s filter them out, and turn the matchup around to see how Sombra 3-3 teams fared against weaker practitioners of traditional 3-3:

Teamfight first-elimination rates and first-elimination win rates
Teamfight first-elimination rates and first-elimination win rates for Sombra 3-3 vs. enemy traditional 3-3, with “good 3-3 teams” removed.

It’s all starting to make sense now, and the Dragons are a prime example for the philosophy that appears to be at play here. Their Sombra 3-3 may have lost against the Shock’s traditional 3-3 59.5% of the time, but they excelled against weaker teams—posting a blistering 70% first-elimination rate, which they converted into teamfight wins 92.9% of the time. They split the majority of these matches against the Justice (24 fights, 79% win rate) and the Uprising (28 fights, 57% win rate), but also won 5-of-6 teamfights against both the Outlaws and the Fuel. The story is similar among other teams that have heavily utilized Sombra 3-3 this season.

Let’s zoom back out again and look at the big picture. Some teams, for one reason or another, are weak at playing 3-3. It could be a lack of personnel, it could be coaching, or it could be that their best Zarya player and best Brigitte player are the same person. When these teams look to the future, arguably the two greatest 3-3 teams the world has ever seen stand at the end of every stage’s tunnel. All analytics and eye tests indicate that running a traditional 3-3 mirror into the Shock or Titans is suicide, unless you believe in your heart that your team is better than they are.

But perhaps these teams also know that running Sombra 3-3 isn’t going to work either—and the Shock proved this in a very public way by dunking on the Dragons in the Stage 2 Playoffs. Some teams, like the Fuel, seem to have decided to fight the monsters at the end of the tunnel head-on and have even made trades to facilitate this goal. Others, I suspect, have made the decision that without significant balance changes there’s no reason to try to beat an apple with a smaller, mushier apple. After all, there are far fewer “good 3-3 teams” than “bad 3-3 teams,” if map differential is anything to go by. Instead, perhaps these teams are playing for season playoff berths instead of stage playoff spots. Why not go all-in on the lineup that will help you beat equally strong or weaker teams with more surety?

In practice, only the Dragons and Uprising have maintained or increased their Sombra 3-3 usage from Stage 1 to Stage 2. Most teams, including the Outlaws, Dynasty, Spark, and Mayhem, have all joined the Fuel in focusing their efforts on traditional 3-3, with a smattering of DPS comps thrown in. These teams know more about their own situations than I do—I’m just some dude who makes pretty graphs—and have their own reasons for turning back onto the traditional 3-3 route. Maybe they did try Sombra 3-3, but it failed miserably in their scrims. Instead, I’m more excited to watch Boston and Shanghai’s progress in Stage 3. Will their gambles continue to pay off?

The Overwatch League returns to the big stage for the start of Stage 3 on Thursday, June 6, at 4 p.m. PDT, when the Stage 2 champion Shock (11-3) take on the Reign (7-7). Watch all 2019 season matches live and on demand on overwatchleague.com, the Overwatch League app, our Twitch channel, MLG.com, and the MLG app.