Last week I introduced the concept of teamfights in Overwatch. Teamfights are far from just the most exciting parts of a match—they’re predictive to winning. For the purposes of this and future analyses, a teamfight is considered “won” when one team scores more final blows than their opponent during the teamfight’s duration. This definition is imperfect, but its simplicity is vital. A well-defined, straightforward rule allows for more complicated statistics to be built on top of it.

Even though this definition of a teamfight win is quite simple, it still unveils some telling results. For example, across both last season and this season, the team that won more teamfights than their opponent has won the map 85% of the time. If we include maps where both teams had equivalent teamfight wins, only 8% of maps have been won by the team with fewer teamfight wins.

Given their importance, it makes sense to study how to win teamfights. Last week I showed some of this year’s teamfight win rates, but how did these teams get those wins? Is there something crucially important that is predictive of the outcome of teamfights?

You might have heard us mention this on recent Overwatch League broadcasts, but not dying first—and conversely, landing the first blow—in teamfights more often than not leads to winning that teamfight. Think about it: suddenly going from a 6v6 fight to a 5v6 is certain to put a damper on anyone's hopes of winning a brawl. That being said, let’s see how badly teams fare when they suffer the first death in teamfights this season:

First-death rates overlaid with win rates after suffering the first death in teamfights.

Generally, you want to be like the Excelsior: they have a league-low first-death rate of 34% and are tied for second in how often they win those teamfights when they do. (The NYXL led both categories in the 2018 season). We can see that this trend maps loosely to the teams at the top of the league standings at the moment: the teams with the lowest rate of first deaths in teamfights or resiliency to first deaths are above or near the playoff cutoff line, and those who are good at both (Vancouver, New York, Guangzhou, San Francisco, etc.) have comfortably positive map differentials as well.

One thing that stands out in a strange way is Gladiator’s third-highest first-death rate, but fifth-best resiliency to first deaths. What explains this? I will revisit this question later in the article, so hold on.

Next, let’s look at the opposite of first deaths: first blows.

Teamfight wins after landing the first blow, overlaid with rate of first blows in teamfights.

This chart shows the importance of landing the first blow in teamfights, especially in the current meta. When an Overwatch League team lands the first elimination in a teamfight, they’ve gone on to win that teamfight 81% of the time on average. This is up from last year’s 72% win rate: while it’s harder to get a first kill in the 3-3 era, landing one is extremely important given the layers of synergy between the lineup’s heroes.

Once again, you want to be like the NYXL here: they have a league-best rate of landing the first blow at 58.8%, and convert those first blows into teamfight wins 86.5% of the time. Also fun to note: the two undefeated teams—New York and Vancouver—are the only two teams whose first-blow conversion and first-death resiliency metrics both rank in the top five. They shrug off first deaths and convert their first blows into teamfight wins better than their peers.

Some of you might have noticed that the rates for first blows and first deaths do not add up to 100%. The reason is because the Burke Model does not actually require a final blow to occur in a teamfight—indeed, there have been plenty of teamfights where multiple ultimates have been used and no players have died. Therefore, it is possible to have a teamfight with no first death or first blow.

We can also present rates for individual heroes, players, or player/hero combinations. There has been an often-quoted statistic on broadcast about Zarya being the least ideal hero to die first in the current meta—well, here’s your evidence to the matter:

Teamfight win rates when a given hero dies first or lands the first blow.

While it’s true that no team has survived after first losing Junkrat, Pharah, or Soldier: 76, these heroes have only died first under five times each, so I wouldn’t put much stock in that trend. With 333 first deaths in teamfights this stage and only a 7.8% teamfight win rate after such an event, I think it’s pretty safe to say that losing Zarya first is the thing to avoid.

However, it’s not unredeemable. Recall that Vancouver is the best team at winning teamfights regardless of which hero dies first, and their teamfight win rate after suffering a first death is only 18.9%. So while teams may lose 92.2% of teamfights where Zarya dies first in the 3-3 meta, they also lose 80.1% of those overall.

Some things to note about these heroes with regard to the current meta is that Lúcio is probably the least-impactful hero to lose, followed by Brigitte, and that while it’s worse to lose Zenyatta and Zarya, Reinhardts have died first far more often than any other hero. D.Va, unfortunately, is hard to track, because losing her mech does not count as a death. Also, in the 3-3 meta, it matters much less which hero lands the first blow as in previous metas.

With these hero averages in mind, let’s revisit our Gladiators question. They currently have a strange ability to win teamfights where they suffer the first death, yet they die first more often than most teams. Let’s go down one more level and start pairing heroes with players to try to solve this mystery, starting with Zarya:

First-death and first-blow rates by Zarya in teamfights, plotted against win rates after securing either in teamfights. A minimum of 100 teamfights is applied.

Immediately, we can see that Zarya is not the Gladiators’ problem in teamfights. Both Lane “Surefour” Roberts and Gui-Un “Decay” Jang have low first-death rates on the hero, and the team does a relatively good job of winning the teamfights where they do die first. Let’s check Zenyatta next:

First-death and first-blow rates by Zenyatta in teamfights, plotted against win rates after securing either in teamfights. A minimum of 100 teamfights is applied.

Jonas “Shaz” Suovaara is also is not dying first on Zenyatta very often. (Also, what do Ju-Seok “Twilight” Lee and Seong-Hyun “Jjonak” Bang know that the rest of the league doesn’t?)

Moving on to the next hero: Reinhardt.

First-death and first-blow rates by Reinhardt in teamfights, plotted against win rates after securing either in teamfights. A minimum of 100 teamfights is applied.

Chang-Hoon “Roar” Gye does die first more often than most Reinhardt players, but is that actually a bad thing? Sang-Beom “Bumper” Park, after all, dies first the third-most in the league, and the Titans don’t seem to be suffering as much as the Gladiators. One thing that the best teams seem to do—that the Gladiators could do a better job emulating—is convert their Reinhardt first blows into teamfight wins.

Roar might be one piece to the puzzle of why the Gladiators have a high first-death rate, but there must be more to why they tend to rally back in those teamfights. Speaking of rally…

First-death and first-blow rates by Brigitte in teamfights, plotted against win rates after securing either in teamfights. A minimum of 100 teamfights is applied.

That’s a bingo! Joao Pedro “Hydration” Goes Telles’ second-highest first-death rate on Brigitte seems to explain a good deal of the Gladiators’ high first-death resilience. It makes sense given what we already know about 3-3 meta heroes: losing Brigitte in a teamfight is less of a problem than losing Zarya or Reinhardt.

To cap it all off, let’s look at the hero who is theoretically the least consequential to lose first: Lúcio.

First-death and first-blow rates by Lúcio in teamfights, plotted against win rates after securing either in teamfights. A minimum of 100 teamfights is applied.

Here, the Gladiators and Benjamin “BigGoose” Isohanni stand out by having a fifth-best rate of winning teamfights where Lúcio dies first. Looking at these individual performances in the first three weeks, we can start to explain why the Gladiators die first often in teamfights but do a relatively good job of winning those fights anyway.

Gladiators fans—and fans of any team that follows this pattern—should feel energized by these kinds of insights. The team is protecting their Zarya and Zenyatta—the two worst heroes to die first in teamfights—at a high level. The team is also already doing well at winning in disadvantageous situations; all they have to do now is not put themselves in those situations in the first place.

Ben "CaptainPlanet" Trautman is the statistics producer for the Overwatch League global broadcast. Follow him on Twitter!