Revisiting match value, WAR and the Capcom Pro Tour
Time for some adjustments
You can subscribe for 20 percent off for a full year when you hit the button. Thanks for your support!
On a podcast I listened to recently, there was a person who talked about coining as many phrases as possible just to see what sticks.
The biggest reference was FOMO and how there was a second acronym introduced at the same time in 1996, it was just that FOMO took off and the other was left behind.
I think about coining phrases in esports and it’s really commonplace. There are some famous phrases that have taken off and have melted into other games.
There was a time years ago when I would try coining a phrase and would get slammed for it, and the phrase would die. These days I’ve changed my approach and not care about it.
I bring this up because since the end of August I’ve been heavily talking about the match value concept and what it means in relation to the Capcom Pro Tour Top 8 matches.
My initial approach to match value was more about the characters. Since switching characters is allowed during a match, that means it’s important to know how much each character contributes to the player winning or losing a match.
Over time as I dived into player stats, I realized it could be used in other forms. It’s not necessarily our version of wins above replacement, but it is a good starting point of what players or characters do in specific situations or against certain characters. It could eventually be built over time to include other factors and become that WAR number.
With thousands of CPT Top 8 matches in the database with every win expectancy total noted for every round, match value won’t be going away anytime, and I plan to talk about it when possible.
That did bring me to a couple of points regarding what works and what doesn’t, which I’ll highlight in the next few sections. Some of these seem obvious, others do not.
It works for matches but not beyond that
In the play index, you’ll see two expectancy totals: match win expectancy and tournament win expectancy. Match win expectancy is what I use to create the match value term.
Tournament win expectancy is a little different. That’s based on where a player is not only in terms of the score of the match but also where they are in the bracket.
I tried playing around with creating a tournament value stat, using the same concept as match value. But it does not work.
When it comes to match value, the total is direct, same as WAR in baseball. A 1.0 total in WAR means that player is worth 1 win by himself. In our match value concept, a 1.0 total means 1 win. It works in reverse as well.
Tournament value does not work in the same manner. I’ll use Punk being in match point loss as an example.
Punk has faced a match point loss 52 times in CPT Top 8 matches since 2017. He’s incredibly 34-18 in those specific rounds, and his match value is 2.23, good for fourth best all time.
This is the elevator pitch in terms of this concept — if every round Punk played was where he was on match point loss, this is what he would get out of it.
Taking the same concept and compiling the tournament win expectancy totals, his supposed tournament value in those 52 rounds is 0.58. Straight up, those rounds would only get him halfway to a tournament win.
But that’s where the concept doesn’t work. Since it’s a double elimination bracket, we have to account for both sides of the bracket, which means splitting the total into two. That doesn’t make sense.
Then there’s the idea of if Punk were to be dropped into the losers bracket, where does the alteration begin in that stat? Starting a Top 8 in winners means winning as few as three matches to win the title. Starting in losers means winning six matches in a row to win the title.
If tournaments were single elimination, there’s a possibility of tournament value being used. Right now, it cannot.
Is it the actual number or the idea?
When Kusanagi won his event earlier this year, I directly pointed to match value in terms of when he goes back to character select following a loss. He’s 10-2 in the games after he takes a timeout.
In 2021, his match value in all of the rounds of the games immediately after taking a timeout is 1.91. The next highest is Infiltration at 0.95.
Straight up, all of those rounds in the games have been worth nearly 2 whole wins to Kusanagi.
But is that how we should view match value, as the number or where it is compared to everyone else? Or is it a hybrid of both?
I’ve been teetering between going both ways, and I’m going to talk myself toward going one way.
In the case of WAR in baseball, yes it can be said that 1.0 means a player is worth 1 win to their team by themselves. But I don’t believe a lot of analysts use it in that term. They think of that number compared to the whole of the spectrum. There’s more to it than it being just 1 win.
Kusanagi’s 1.91 is insane when compared to the rest of the field. He in the history of the pro tour has gotten more out of taking a timeout than anyone else not named NuckleDu. Can we view it as 1 win and almost 2?
I think if match value were to become a thing, the actual number should be secondary to the idea of the number. Yes, we can say that Kusanagi taking all of those timeouts has been worth nearly 2 wins. We can also say that Punk in match-point-loss mode has been worth about 2 wins.
If we think about those 2 wins for each player in the scope of their situations, then we can truly see how outstanding they have been in those situations compared to the rest of the field.
This is probably the way to go with match value moving forward. Anyone can take the number as is. Straight up, positive is positive and negative is negative. Line the number up with the rest of the field and we really get to see what it’s all about.
This newsletter is supported by you the awesome stat freaks, fighting game fans and curious readers. Thanks for being here, and I hope you consider upgrading your subscription today!
Does volume hurt or help the stat?
This could be grouped into the previous question when we talk about the number or the idea. I’ve decided to separate this part of the topic.
Realistically, a player only needs to win six rounds to win a match. A player can win more if the stars don’t align, but it’s generally a race to six.
So if we look at the 34-18 record by Punk in rounds where he’s facing in elimination, just taking the wins means he would be past five wins if six equals victory.
Does that mean the 2.23 match value is under counted? Not really. For starters, there are six different scenarios where a player faces elimination. There are six different values that are added in this equation.
There could be a player who has to deal with all six — to date, Julio Fuentes is the only player to overcome six match point losses in a single CPT Top 8 match, done five years ago. Players do better in certain scores than others.
In these differential totals, it takes a lot to get to 1 win in this specific setting.
There are other settings where it takes a lot fewer rounds to get to 1 match value win, and we see that with the Kusanagi stat mentioned earlier.
After a few months of looking at the stat and how it’s formed discussion, volume helps tell the story. But it leads me toward viewing the number as the idea.
It’s not that Punk has, at face value, gained two match wins if every round is considered elimination although that technically can be said. It’s not that Kusanagi has gained two match wins if he took a timeout after every game where he lost although that technically can be said as well. It is that they are the best or among the best in those scenarios, and players have to either be aware of it or account for it. Those wins have to be placed in the spectrum among everyone else who has dealt with those scenarios. There has to be more stat diving in order to get these players off of their mantles.
I know I’m going to be leaning toward the idea rather than the number, but I won’t mind mentioning the number as is. What match value has done is force me (us? stat freaks?) to do more stat work. To truly understand things, I/we who analyze these matches statistically have to do more than just say the number and then hope for 500 retweets and twitter verification.
That probably means mentioning the match value number less often, picking situations where it can matter. That’s a bummer because I like mentioning it, but less is more as many people say.