by JR Cook - 9 years ago     show comments

This is the 3rd part of my series looking over at PVP Live’s attempt to provide advanced knowledge about Hearthstone using an statistical approach.  In part one of the series, I had an interview with their director of Data Science. During the second part I analyzed the Stats they decided to track and discussed the difficulties of incorporating this kind of analysis into Hearthstone. For this third and final part, I will propose some different metrics to be tracked, and I’ll explain while I believe that those new stats will both have a stronger appeal to the Hearthstone Fan Base, and actually provide better insight into the game itself.

Creating Relevant Stats

When people look into stats in sports, they are used to see metrics of success. Stats are often ways to evaluate a players contribution to the team, and they track moments that somehow are very relevant towards the Team’s chances of winning. Take for example Sacks in Gridiron Football, hitting the QB can make him slower and less accurate while he’s rattled by the hit, losing the down and yardage are also very relevant as they could prevent the enemy team from scoring points, and it can also mean that the team will get better field position on its next offensive possession. In Basketball, getting a rebound or a steal means that you have acquired (or retained) possession of the ball, you can’t score without ball possession, so again, these stats are very relevant.

Even in non-team sports you have some relevant stats, things like first serve % in Tennis, or well… basically the entire game of Poker. Even Golf has things like Shots to Hole, a website that claims can improve your Golf game based in advanced statistical analysis.

But how does this translate to Hearthstone? As we looked into during part 2 of this series, the Stats Currently being tracked by PVP Live are more descriptive of what’s happening on a match, rather that evaluating a player or deck performance.

Any player with experience can tell you that despite netting you 3 Damage to Enemy Hero, a turn 1 Sinister Strike isn’t really the best of Plays. You can probably figure out on your own that  despite both getting you +1 to Minions Killed, killing Ragnaros is very different than Hero Powering down a Wisp. In fact, Minions Killed per game is too easily influenced by length of game and the decks used, perhaps Minions Killed per Turn would be more interesting.

Finally, even when the decision to whether or not using your HP as a resource can be key moment of a Hearthstone game, Blowback in its current incarnation doesn’t really provide much insight on whether the players are making the correct calls or not.

So if this statistical approach isn’t that relevant, why is it being done? Well, if you’ve watched PVP Live broadcasts you should realize that they have a high commitment to growing the eSports scene, that the end game for them is eSports to gain mainstream acceptance and even compete with the Traditional Sport Powerhouses. Presenting Stats similar to those of traditional sports should be able contribute to the main goal of expanding the eSport audience, and in general I believe that social conditioning could help eSports evolve and eventually capture mainstream attention.

Reinventing the Wheel

PVP Live has presented a product that looks very different than anything we’ve seen in Hearthstone, and you can tell that they’ve put great effort into producing a quality product. In fact, I believe they might have gone too far, to the point where you could tell that they put so much effort into creating something new and shiny, that they forgot to look at was already there and successful. By looking past the existing advanced knowledge about Hearthstone, they might just have reinvented the wheel, and made it square.

If you are knee-deep into the Hearthstone scene you probably heard advanced terms such as Tempo, Value, Card Advantage, Overextending, Initiative, Setting-up Lethal, and many more. If you look around, there’s a myriad of information that professional players and teams have produced. There’s a deep understanding of how Hearthstone works, and how to determine the correct lines of play. Perhaps if the PVP Live Data Science Team would have looked into the existing Advance Knowledge of Hearthstone instead of trying to impose their own, they would have come up with metrics that wouldn’t just allow you to get a better statistical representation of the game, but would have also been accepted much better by the Hearthstone community.

Quantifying Value

But how do you take these advanced concepts and make him into parse-able data? How do we take the existing concepts and turn them into stats? How do we make sure that we come up with Relevant Stats? To be honest its a lot of work and somehow that’s better served as an entire article, but lets go over the more important ideas.

Tempo:

The easiest way I see to quantify Tempo is by looking at the total stats and effects currently on board and assigning them a converted Mana cost. At the end of each turn you can quantify how strong is each players board position, and perhaps provide the average advantage a player held through the game. The specifics on this need of course to be worked out with delicate care; for example, how do we consider the amount of mana our opponent will have to answer our board? How much converted mana is a Deathrattle effect like Sylvannas or the activated ability of a Grim Patron worth? How much converted Mana is each point of stats worth? All of these need to be answered with exhaustive research.

Card Advantage:

How valuable is having an extra card? Should we measure the amount of cards in hand? How should we count the minions that we already played and are still on board, how do we count extra minions like a Silver Hand Knight‘s Squire? Again, these are the things that need to be figured out, and most likely these questions have already been answered at least partially by very reputable sources.

Playing The Odds

Finally, there’s one little thing I want to talk before I wrap this up. One of the things that I like the most about watching a game like Poker on TV, is that you can see the odds players have of winning before the cards are revealed. I’m a bit surprised this hasn’t been implemented in Hearthstone yet. In Hearthstone, you often run into situations where you are playing around certain cards, but how likely is your opponent to have it? How likely is he to top deck it, if you are positive he didn’t have it before? Those odds can easily be calculated, and I believe they could add a lot to a Hearthstone broadcast. Sadly, the company that would be most inclined to do so is indeed PVP Live, and since they allow the players to play any deck freely, then they have no way of knowing for sure how many copies of that card is actually on a deck, if it all. Thus, they are unable to calculate the exact odds.

Still, this is something I’d like to see in the future. I guess that by being knowledgeable about the current meta, you can still make an educated guess of what cards are in someone’s deck. Would you like to see this too? Let me know using the comment section below!

There you go guys, this is my third and final look into what PVP Live has been trying to do to revolutionize Hearthstone via advanced statistical analysis. But this isn’t the end of my Journey into advance Hearthstone Knowledge, I guess now I have to amp-up my research and see how can we actually translate the existing knowledge of Hearthstone Pro’s into numbers and stats! Yay me, I guess…