Resistances

[magey]

• Do resistance breakpoints exist or does it scale linearly?
• How does the damage reduction work across the range of values?

[ohganot]

As far as I know, the breakpoints are a myth and it should scale linearly.

[magey]

As far as I know, the breakpoints are a myth and it should scale linearly.

Updated.

[vigo2]

It'd be interesting to test if and how negative resistances work.

[Blaticus]

There are no break points (based upon the data referenced) as far as average damage reduction is concerned. Interestingly, for non-binary spells, "100%" resist events (incoming damage reduced by 100%) are rolled down to "75%" resist events (incoming damage reduced by 75%). The effect being that the maximum average damage reduction from resistances caps at about 69-70% rather than the expected 75%. This makes the average damage reduction scaling ever-so-slightly non-linear (diminishing returns) towards the very high resistance score end of the graph.

As far as the chances for individual resist events go (incoming damage reduced by 75%, 50%, 25%, 0%), each curve is quite non-linear. The basis functions for each curve was hotly debated, the three theories being proposed were: piece-wise linear, quadratic/cubic, or Bernstein polynomials. There is apparently a "break point" for completely removing a 0% resist event from occurring. From the data referenced, it appears to be at about 69% of whatever the max effective resistance score is versus that mob (so versus a boss, with a maximum effective player resistance score of 315, it would be about 218 resistance score to drop the chance of a 0% resist event occurring to 0%. The next highest resist event (25%) chance bottoms out at about 4% at maximum effective resistance score and cannot be completely pushed off the roll table.

The original resistance score explanation image from the WoW website from 2004-2005 is inaccurate in two major ways:

1. It makes no mention of "100%" resist events for non-binary spells being rolled down to "75%" resist events, which reduces the maximum average damage reduction possible
2. It implies that a 0% resist event cannot be completely removed and instead asymptotes towards zero, when in reality they can indeed be pushed off the roll table (the data referenced shows nearly 14,000 hits received while over the resistance score threshold with not a single 0% resist event occurring).

Link to data: from the EJ forums, snap-shots on webarchive, posts were around April 2007.
http://web.archive.org/web/20110808083353/http://elitistjerks.com/f15/t10712-resisting_magical_damage_its_relation_resistance_levels/

[Cleavis]

Interestingly, for non-binary spells

From the data referenced

Just wanted to clarify the EJ data you referenced is specifically referring to elemental melee attacks.

If you would like to check non-binary spells, I would recommend using the Classic client and provide some data from there.

[Blaticus]

Elemental melee attacks (that connect) are handled the same as other non-binary spells. We are fortunate that the Sapphiron encounter provides thousands of data points, per attempt, per raid, per week. All data I've analyzed from Sapphiron's Frost Aura, as well as Huhuran's Poison Bolts, agrees with the EJ data.

EDIT: I'm including data extracted from my guild's logs for our Sapphiron kills for three weeks. I sit at 311 Frost Resist through the entirety of each encounter. The sample size is relatively small but you'll notice:
0 "100%" resist events (the full resist events seen are the 1% spell miss rolls)
0 "0%" resist events
Blat_Sapphiron_Resistance_Data.xlsx