I've been involved in the Theorycrafting thread over in the official forums and thought I might share some of my findings. Please be gentle.

First, HtL's uptime is easily modeled by

H = 1-(1-p)^(10/a)

where

**p** is your parry chance,

**a** is your attacker's attack interval (i.e. number of seconds between attacks), and

**H** is your time-averaged uptime for HtL as a percentage of the fight. Simulation bears this equation out, though it will obviously be much more erratic for shorter fights than for long ones.

Second, thinking about block/dodge/parry in an abstract sense, they all contribute to whether or not you can be hit by a full attack or not.

Dodge and parry result in no damage, while block results in 40-70% of the damage you would have taken without mitigation, before armor and other absorptions are factored in. It is always better to take a lesser hit than to take a full one. Mastery increases Block at a faster rate than dodge/parry rating increase dodge/parry. Consequently, until you attain sufficient block+dodge+parry+miss to avoid

*all* full hits, mastery is substantially better than dodge or parry per percent avoidance provided. Mastery can not (currently) exceed 40% of your unreforged, combined dodge+parry rating, so that provides a hard cap on how much mastery you can have, given your current gear. The remainder should be split

*almost* evenly between dodge and parry, with slight (<10%) favoritism given to parry due to HtL.

Once you have sufficient avoidance to not be hit by any full hits, mastery immediately loses all value: it is always better to take

*no* hit than to take even a mitigated hit (no, I'm not talking about the value of EH here; that's an entirely different matter). As such, mastery should only be used at this point to ensure one remains unhittable, and all available points should then go toward dodge/parry, maintaining the same ratio mentioned before.

Here are some example cases (also posted on the WoW forums):

Using 1750 rating points...
Best avoidance: 15.8%, 15.32%, 48.79% (84.92%)
DR: 520
PR: 530
MR: 690
Using 2000 rating points...
Best avoidance: 16.92%, 16.61%, 51.51% (90.04%)
DR: 590
PR: 610
MR: 790
Using 2500 rating points...
Best avoidance: 19.18%, 18.86%, 56.96% (99.99%)
DR: 740
PR: 760
MR: 990
Using 3000 rating points...
Found an unhittable configuration!
Avoidance: 24.7%, 25.17%, 45.25% (100.12%)
DR: 1170
PR: 1260
MR: 560
Damage taken: 23.13%

A few caveats:

- None of those cases actually add up to the number of rating points they claim to use, because I'm simulating in hops of 10 rather than hops of 1 to speed things along. The actual values will be within 10 rating points of those.
- I'm also not 100% certain about my damage taken calculation, which will ultimately influence nothing in terms of weighting dodge/parry/mastery against one another.
- Parry
*rating* may not be as important as parry itself, because of the contribution green strength now provides to parry rating. You can reforge gear-derived parry rating, but not gear-derived strength, so the amount of parry you can actually shunt around is likely going to be smaller than the amount of dodge you can shunt around.

I hope this is useful.