Blue wrote:After that every following test was done in PvP.

Do you have the excel data of those tests?

If every test that followed was made with an unbuffed attacker, using a weapon with only a 10% bonus, that could give you misleading data.

Recall that my testing on evade rates showed that an attacker with low weapon skill actually increased his target's chance to evade. Motaro's testing showed that block rate is effected by weapon item level vs. shield item level. That's why weapons with lower DPS were blocked more than weapons with higher DPS (against a level 50 shield). Both of these factors could have given you incorrect results on your block testing.

Also,

Celad wrote:Thanks Blue for your work.

Blue, a lot of people whined about block rates being too low before the change, so lets look at what was changed:

block rate is increased

guard rate is increased <--- this is the big one

The guard rate change alone represents a major increase in support defense, as long as that support is within guard range. From your test it is clear that the guard rate was much too low.

Based on what we now know about defense penetration, guard rates, and how block is now performing on Uthgard, here is what I suggest:

* Go back to the old Roclar formula for block, but keep the new guard rate; this will help real RvR block rate for support

* Reduce the defense penetration bonus for DW/CD/LA/2hand from 1/2 to 1/4; again, this will increase real RvR block rate against the primary DPS classes

* Remove the evade bonus from dex and qui; high evade rates are completely imbalanced on a server with no defense penetration formula

That's it. Very simple, and easy to implement.

We will do more tests and hopefully come up with something more correct than the current situation. Please be patient.

I'll stop procrastinating and do a real evade rate test on live that we can develop a formula from. See you in 50 hours.

Seyha wrote:I'll stop procrastinating and do a real evade rate test on live that we can develop a formula from. See you in 50 hours.

Don't forget to write down every detail of your test environment. Missing one variable could make trouble later.

Not sure though, but I think the problem with the old block rate wasnt excactly the formula, I have in mind it was mostly because almost everyone in an e.g. 8vs8 fight was counted as attacker and therefor heavily reduced your chance to block and not only those you were actually fighting.

Heho,

I think, one thing which is wrong is that you more block with a larger shield. Many Warriors use larger shields because they think they block more and it looks like clerics have the highs block rate from the supporters.

Greetz Luydor

Murax wrote:Not sure though, but I think the problem with the old block rate wasnt excactly the formula

Nah, blockrate was way to low in all situations.

A fully buffed mob3 tank had trouble blocking yellow mobs, when the blockrate would be 80%+ on live.

Seyha wrote:I'll stop procrastinating and do a real evade rate test on live that we can develop a formula from. See you in 50 hours.

We still have your evade test results. But if you got more data I will be happy to use it in some way or another.

Absolutely

Here's my method:

Target: Level 50 Shadowblade, 145 dex, 120 quick, merchant armor with 0 enchantment bonus

There will be two attackers; a Shadowblade, and a Berserk. Both level 50 with the same strength, weapon spec, and weapon (lvl 1 starter axe).

I will do 24 total tests (2000 swings each), 12 for each attacker, at different strength and weapon spec values. Both attackers will have the same strength and weapon spec for each test. I will start at 115 strength and 1 weapon spec, and the last test will be done at 51 composite weapon spec and >300 strength.

Of course this will take days. I just finished the first test. I will create a thread in the bug reports section when I am done.

Sounds like a LOT of work. Be sure to write down every detail of the test environment. Better more than something is missing.