Choosing Your Ladder Deck Like the Pros
It's frequently hard to know exactly which deck you should be using to climb the ladder. Obviously one you're good at is a prerequisite, but even then you've undoubtedly had a moment where you've questioned whether or not this is actually the deck you should be running. With variance there are wonderful win streaks and painful losing streaks that you can't let sway your confidence.
Professional players like Trump use math to decide which would theoretically be the best to run. This formula accounts for your win rate with decks, against all of the other decks, and the rate at which you've been encountering them.
F1W1 + F2W2+ F3W3 + FnWn = Predicted Win Rate.
This is the equation that we'll be using to calculate which deck we should be using. So let's take a look at the value for W, your actual win rate, first.
These are all of my values for W which is equal to the number of times you've beaten the deck divided by the number of times you've played against it. So in the highlighted case I've defeated Midrange Druid two of the four times I've encountered it when playing Grim Patron Warrior. You do this for each match over a long period of time and you begin gain a large sample size that will help your data become more accurate. A 0 means that you've never won a 1 means you've never lost. Most of the time the value will be a decimal between the two.
Then we have F which is the likelihood you'll encounter said deck. It's simply the number of times you've ever played against that deck archetype divided by the total number of games in the sample size.
In the spreadsheet to the right, I've highlighted my chance of running into a Midrange Paladin deck on the ladder. Over the past week or so I've faced it three times during my 41 matches. With that math there's about a seven percent chance that I'll play against it each time I go to start a new match.
You'll want to be updating both of these databases following each match as to increasingly make your data more accurate. F however, can be reset when you want. It's simply a gathering of what you've been facing lately. Have you noticed a shift in the meta and want to throw away some of the older records to make way for more relevant data? Go ahead and do so. I've gotten into the habit of resetting it every week, others like Trump base it simply on the last few days.
When the time comes that you're wondering which deck you should be using, it's time to do some math. Let's refer back to that equation: F1W1 + F2W2+ F3W3 + FnWn = Predicted Win Rate.
We now have F and W and can calculate approximately what my win rate would be if I took Grim Patron Warrior onto the ladder right now. (Note: the following chart includes data for decks that aren't shown in these images and therefore has a different result).
Grim Patron Warrior holds a value for .05 on F, but since I'm yet to take down that deck in a mirror match, it adds 0 percent to my projected win rate.
0 + F2W2+ F3W3 + FnWn = Predicted Win Rate.
I've yet to lose to an Aggro Paladin and therefore I get the full seven percent. Midrange on the other hand is a 50/50 tossup so we're going to round up and give myself a .04 on that one.
0 + .07 + .04 + FnWn = Predicted Win Rate.
Extrapolate this out through Midrange Druid, Midrange Hunter, Face Hunter, and Zoolock respectively and we'll get our final result.
0 + .07 + .04 + .10 + 0 + .05 + .09 = .35
Based on that information we're projected to have a 35% win rate with Grim Patron Warrior. As you can see to the left, with all of the data I've accumulated, none of my decks actually have more than a 50% chance of winning.
As you play more decks, get more data, and are therefore able to project better, you'll be able to more quickly adapt to changes in the meta.
If you decide to utilize this yourself, let us know your results! Have any questions? Let me know in the comments and I'll do my best to address them.
获取 Wowhead 高级会员
[As little as
less than $1 a month
to enjoy an ad-free experience, unlock premium features, and support the site!]
[Show 0 Comments]
[Hide 0 Comments]
[Sign In to Post a Comment]