We really don't know what makes someone good at video games. We can speculate, but we don't know. What we need, frankly, are better stats.
In StarCraft we bear witness to the immaculate precision of a perfectly-timed Zerg ambush, but we don't know how that compares to the average Zerg ambush.
In a Call of Duty multiplayer match, we can watch someone noscope a kill cross-map, but we don't know whether or not that shot will regress to the mean.
That knowledge of how good a given player is in a competitive game should be attainable. It's all numbers. It can all be calculated. But it is locked away in byzantine mazes of data hidden within the first-person shooters and real time strategies that we and thousands of pro gamers . As games are slowly being taken more seriously on competitive levels, however, those mysteries are starting to vanish. People are starting to get that knowledge. With that, a sense is emerging about which stats can tell us how good players of the most popular multiplayer games actually are.
You could make the argument that a sport isn't a sport until it has numbers backing it up.
Eivind Fonn is not a professional StarCraft player. The 29-year-old spends most of his time working on his Ph.D in applied mathematics somewhere in Zurich. But in his off time he has built Aligulac, a website that tabulates a ridiculous amount of competitive StarCraft 2 data into an interface that allows its user to predict completely hypothetical matches. It's perhaps the first usage of analytics in eSports.
Fonn's website takes a player's wins and losses, strength against particular races, and other, more abstract computations to calculate their cumulative rating. With that number in hand, Aligulac can predict matches using the difference of two player's ratings. For instance, with a few clicks on Aligulac, I can see that HerO has a 20.96 per cent chance to beat Jjakji 2-0 in a best of three.
"I got into StarCraft when it was released 12 years ago, but I didn't start thinking about deconstructing the game in numerical terms until a couple years ago," Fonn told me. "I was looking to find a way to predict matches but I found the existing resources kind of wanting. So I just developed my own."
How accurate is it? According to Eivind's maths, very. A line graph on the site shows that Aligulac has predicted StarCraft matches correctly about 80 per cent of the time. As the number of games in a match goes up, so does Aligulac's accuracy. A best of five? The site gets those matches right nearly 95 per cent of the time.
"StarCraft is a difficult game to predict," Fonn said. "It's not like tennis or chess. But I did enter the Team Liquid prediction contest and I did quite well. My numbers could beat most everyone else competing."
Fonn is just one of a growing number of people inclined to bring the advanced stats we see in traditional sports into competitive gaming.
Looking at video game performance through statistics isn't anything new. Goldeneye told us our kill/death ratio. Halo calculated our assists. The fighting game scene has always been beholden to frame-counting.
But in gaming and in sports in general, we're entering the era of widespread advanced analytics, the natural product of macro-computing. This year, the NBA installed dozens of player-tracking cameras in each of their arenas. Want to know which player moves the most on the floor? It's Nicolas Batum, at 2.6 miles per game. Figures like that were mere postulations a few years ago, but now it's only a few clicks away from everyone. Aligulac's algorithm might not be at that level quite yet, but the promise is there, and it's the direction that competitive gaming — eSports — has to move, supporters say, in order to be taken seriously on a mainstream level.
"eSports is the one place where everything the player has done is recorded by the computer," says Joerg, noting that could lead to "bullshit-free analysis...better conversations, better players, and better games."
Some people even believe that competitive gaming can get more out of stats than any conventional sport can. After all, what kind of competition is more quantifiable than one that's run not on a field or on a wooden floor but on a computer? What kind of sport should be able to more defined by stats than eSports?
"The dream is the end of bullshit," says David Joerg, owner of the StarCraft statistic website GGTracker. "eSports is the one place where everything the player has done is recorded by the computer. It's possible — and only possible in eSports — where we can have serious competition and know everything that's going on in the game. It's the only place where you can have an end to the bullshit that surrounds every other sport. You could have bullshit-free analysis. You'd have better conversations, better players, and better games. There's a lot of details needed to get there, but the dream is possible."
GGTracker arrived on the scene about two years ago. It allows any StarCraft player to upload his or her replays to a database that spits out stats like actions-per-minute, spending skill, and win rate. The site has processed over four million replays since its inception. It's not perfect, but it informs a player of some solid numerical efficiencies and inefficiencies to their play-style. GGTracker can reveal bad habits, missed opportunities, and hidden turning points. It's the first step towards articulating eSports through meaningful data, the first step towards the dream.
Right now, stats in eSports are fairly unregulated. Unlike something like Player Efficiency Rating in basketball, there are only a few stats in competitive gaming that enjoy widespread recognition. A company like StarCraft-maker Blizzard doesn't focus on developing metrics, so most of that work is left up to the community, for better or worse.
"There are some stats in every video game that are directly visible to the player, like kill/death," GGTacker's Joerg said. "Everyone will use it because it's right in front of their face, and then people will say that stat doesn't tell the whole story. So then a brave soul will try to invent a stat that's a better representation of a player's value, but that leads to a huge uphill battle trying to get people to use it correctly and recognise its importance."
Sometimes this works out. Spending Quotient is a stat that emerged from the community over two years ago, and it remains one of the go-to stats to track a player's StarCraft efficiency. You plug in your collection rate and your average unspent resources, and out comes a tidy little number that represents how good your economy is. It's broad, but it works, and its adoption by the community is proof that players are looking for a better way to understand the game they love.
Spending Quotient is fairly crude. It only identifies a player's inefficiencies on a superficial level. It gives a player a loose idea of where they need to improve. It answers the what, but not the why, like evaluating a baseball player's lackluster offence, without knowing if he's a bad baserunner, or unable to hit for power.
"We're a ways off from baseball nirvana," Joerg said. "I don't think anyone truly understands StarCraft outside of the top talent."
Ideally, eSports games and the growing number of people who watch competitive games will have access to stats that measure the precision of a player's unit control, or the average effectiveness of their ganks. Those numbers may someday be available, but eSports just isn't at that level yet. It doesn't help that the games people play in eSports change over the years and what people compete in remains fairly new, adding new rules and introducing new gameplay possibilities — new plays, as it were — that can make old stats irrelevant and necessitate the need for new ones.
"Before we were guessing, and now we're doing a little better than guessing," Joerg confessed. "Take basketball. When was basketball invented? And when did the stats for basketball really get solidified? For any sport, the answer is always between 10 and 50 years. So what video game is going to be played at a high level for 10 to 50 years for the community brains to nail down what numbers are important?"
Regardless, it's safe to say that statistics are a key element to the future of eSports. They add legitimacy, they provide context, and most importantly, they tell a story.
"Stats make competition much more impactful," Whalen Rozelle, director of eSports at Riot Games, developers of the hugely popular League of Legends, told me. "It changes the air of the entire experience. Even right now, if we track a win streak, that's the foundation of what drives narrative and [appreciation of the] talent of the players. Narrative is what makes sports fun and interesting."
Better stats create more possibilities. To wit, League of Legends' Rozelle says: "We've been thinking about fantasy eSports."
One of Rozelle's top priorities is to emphasise stats during eSports broadcasts. That will prove how stats can make eSports more interesting for their audiences. Tools like GGTracker are built to help players get better. They're personal investments, numbers that don't necessarily fit well in a livestream. Stats that can be better-integrated into a livestream that enlighten a spectator will help eSports grow.
"It's about giving wider access to our community," Rozelle said "You look at armchair quarterbacks and sabermetricians who are doing their own analysis, and it even goes further than that. We've been thinking about fantasy eSports, which we think would be a really cool experience. Obviously the game is core, but League of Legends is more than that. It's about being a fan. It's about watching streams. Stats unlocks another aspect of that."
You could make the argument that a sport isn't a sport until it has numbers backing it up. Until someone can point a series of statistics that clearly designate a player's superiority, there will always be doubters. If that's true, then it's true for eSports as much as it was for baseball, football and any other sport when it was young. For gaming, those metrics remain hidden in the computers running StarCraft, League of Legends, Call of Duty and any other game being played in high-stakes tournaments. Slowly, though, we're starting to discover how competitive gaming truly works. We're starting to find the numbers that tell the story. That's exciting.
"We're constantly throwing things back and forth with each other," Rozelle said. "Like how cool would it be if we could track how successful a team is at tower-diving before five minutes? It will be something that develops over time. And I believe strongly that as our statistical knowledge gets more sophisticated that we will be able to use those numbers to figure out how good a player really is."
It's a lofty goal, but if there's one thing we know for sure, it's that communities rarely back down from a challenge. Eivind Fonn is a Ph.D student, David Joerg works in the financial sector and raises a family. They're making time to benefit their scene. With people like that, it's hard not to be optimistic that someday we'll all know the stats that define how good a competitive gamer actually is.
Top image via Shutterstock.