Warning for Mindaika: 2. All opinions should be stated as succinctly as possible.

• Aldriana · 04/29/09, 5:48 PM

Post: There Will Be Blood (DPS) - Meet the new thread/Same as the old thread
User: Mindaika
Infraction: 2. All opinions should be stated as succinctly as possible.
Points: 0

Administrative Note:

Message to User:

Please don't quote-split in this fashion.

Original Post:

Originally Posted by Valtiel

We're absolutely running in circles here.

You keep insisting that you can integrate random events in a mathematical model. I'm telling you that yes, it's possible, but it makes the model innacurate. You're creating "average", but average doesn't just happen.
If I'm capped, I know that my performance under the conditions outlined within the model will result according to the model.

This is absolutely untrue. "Average" ALWAYS happens. It is a mathematical function. If event A is 100, and event B is 200, and each occurs 1000 times, the average is still 150, even though there was no instance of that actual event. Whether or not you are hit capped, performance under parameters of a model will have results similar to the model's prediction. You cannot model a truly random event. However, we are not dealing with random events. The RNG does not generate random numbers, it generates pseudorandom numbers, and more importantly, it generates them within fixed bounds. If it did actually generate random numbers, you could have one Death Strike hit for 50, and the next one hit for 2.4 million.

If I'm not, I can walk into a fight expecting to get a performance as close as possible to the statistical average I've reached with modeling, but I have possibly no guarantee it will happen. My model shifted from a mathematical model to a statistical model, which means that odds are that it will be accurate over the course of several parse and will become more accurate the more parses I stack (again, theorycally, because I can still be unlucky all of my life). But the "instant" accurace of the model is next to none. Walking into a fight with 2% miss chance and expecting to miss 2% of the time has no actual accuracy.

The only guarantee you have with being over the hit cap versus being under is that you've eliminated one variable from your model. I'm not sure you understand how models work. Being under the hit cap simply expands the range of values (i.e., damage) that you can get. Also, you cannot move from a mathematical model to a statistical model, as statistics is a mathematical science.

Does this mean that it's impossible to model if giving up 2% hit chance for 10% ArP is beneficial? Absolutely not. My point is that RNG is incredibly complex to calculate for DKs (because of the strictness of gcd handling) and makes the model ultimately unaccurate, so eliminate RNG is a benefit in itself because it allows you to move from an inaccurate model to an accurate one.

Again, untrue. The RNG is not at all difficult to calculate. 2% miss is an average of 2% miss. No, that doesn't mean you will always have exactly 2% miss in every fight. However, is does mean that across the literally millions of players, the average miss will be 2%. Using a 2% miss rate in a model and expecting it to reduce damage by 2% does not make the model inaccurate simply because it is an average of 2%. If it was, it would make the ArP model inaccurate as well, since actual armor ignored is not constant.

However, I find it silly to keep arguing this point which is as Method pointed out mainly semantic, while apparently avoiding the real question, which is: how do you convert 60 hit rating in 3-4% extra damage?

True, it is semantic, but it's also important that people understand how models work. The purpose of a model is to make predictions of outcomes. You CAN use a model to predict the damage done by a DK with [some variables], you CANNOT use a model to predict the mana gained by a warlock with a fixed spirit value. Life Tap uses constants in it's calculation, so there is no need to model what will always be a constant outcome.

The mentality of "1% miss rate means that you MIGHT ACTUALLY miss EVERY SINGLE ATTACK" is the same mentality of gamblers. Shivand's argument of "What happens when RNG screws you on an add and it takes too long to get him down and the raid wipes?" to justify hit capping goes another way as well: What happens if you're hit capped and the RNG screws you and every single raid member gets the lowest possible damage for every single hit? Taking out one variable will not guarantee that you will have higher damage, as you are still dealing with randomness.

The general models for damage are:

[0% miss]: Random number * constants
[1% miss]: 0.99 * Random number * constants

The final answer though is: 60HR is around a 2% increased chance to hit, which equates to an average of around 2% more damage, assuming you are at least 2% below the hit cap.