Talk:Battle Arena/@comment-33757834-20181027003205/@comment-27957947-20181030030803

Thank you, Julie.😘 I think many people don't realize the algorithms that dictate battling (and breeding) are basically customized gambling algorithms. Having a basic understanding of how odds, probability, and statistics work (which is explained in the link) can really help those having troubles understanding the Arena. That site is the best one I've found so far that includes info on most of the basics on the math/theory relevant to the battling algorithm. The last thing I am is a troll.😂

Further explanation, in case Starship returns:

To start supporting the hypothesis "metal is weak against red" you would need to collect lots of very specific data. You are trying to disprove the game data and that is a tall order. But, the way the game presents battle opponents, collecting the needed data is not possible. Here's why:

Certain dragons do appear in the Arena frequently, but it's almost always at different levels- level 11 in one tourney, level 14 in another, level 19 in another. Those fight results cannot be put into the same data group. How many times have have you fought the *exact* same dragon? Not many. None of us have.

You'd need to face the exact same dragon hundreds of times (and fight it with the exact same opponent each time) to even begin to think about drawing any conclusions. There are many, many, *many* possible paths to the final result of a battle. Even if you have fought 50 identical battles, that's still not a significant amount of data. That's what I meant when I said you are thinking too small.

Similarly, you also can't lump 30 instances where you fought a random red hybrid vs. a random metal hybrid into one data group and call the results conclusive. There are too many variables present. Was it really the red in Cobra that won a battle against Metal, or was it the Cobra's yellow? There is absolutely no way of knowing. Claiming it was definitely the red crosses into the land of speculation and conjecture. Not good.

If there was a way to track the results of 10Metal vs. 10Fire for hundreds of battles, that would just be the start of collecting the data needed to prove your theory.

So, a couple dozen wins by various dragons containing red or metal along with various additional types, with dragon levels and rarities all over the place is inconclusive. Not only is it a tiny and very flawed amount of information, the results fall under the umbrella of what is possible, not what is probable. There is a big difference, and that difference is very important when considering the outcomes of gambling algorithms.

Just because a roulette wheel lands on red 15 times in a row doesn't change the odds of the wheel landing on red. The odds are constant, unchanging. Spin the wheel 20,000 times and the odds will bear out (that's where the Law of Averages and the Law of Big Numbers, covered in the link, come in). The outcome of those 15 spins are an insignificant amount of data to draw a accurate conclusion.

In battling, you always want to use the dragon most likely to win. You can tell which dragon has the highest probability of winning against an opponant by looking at and comparing the odds given in the calculator. Will the dragon with the best odds always win? No way. He could lose 20 times in a row (see: roulette wheel). But, when looked at over hundreds or thousands of identical battles, that dragon *will* win more often than any other dragon in the game.

Dragons with lower odds can win against that opponent because it is possible (every single dragon has at least a 10% chance of beating every single opponent), but it's not probable.

We want probable, or likely, and not merely possible.

Example: If you could pit 15Air v. 18Battlesteed 1000 times, and 15Diamond v. 18Battlesteed 1000 times, the Diamond is going to come away with more wins. Sure, the Air will have a nice pile of wins, but Diamond will have more because he is more likely to win against Battlesteed. Winning with the Air 10 times doesn't mean he's a better choice for fighting Battlesteed, even though someone presenting their very small piece of anecdotal evidence ("I won 10 times!") may make it appear that way. The reality math and the battling algorithm is: 10 battle results aren't diddly squat when compared to the big picture. (see also: law of averages, law of big numbers- the algorithm is BIG).

I hope that all makes sense. It's a lot to take in. That's why the calc (and the folks who maintain it) is so freakin' awesome. All the variables are considered and applied, and all we have to do is tap a couple buttons to see the results. You won't realize Sobek is the best dragon to use against Wardrum, or which of the hybrid Champs have weaknesses (which aren't 10%ers) by looking at the color chart. It's just not sophisticated enough to give that information.