User:HOFMANN58D/702 Research

The official GemStone IV encyclopedia.
Jump to navigation Jump to search

I cast 702 on hobgoblins and recorded 1404 successes that generated crit messages, entirely on unwounded hobgoblins who were not stunned. Raw data here User:HOFMANN58D/702_Research_Data

You gesture at a hobgoblin.
  CS: +81 - TD: +9 + CvA: +20 + d100: +21 == +113
  Warding failed!
  Target staggered by strong strike for 25 points of damage!
   ... 10 points of damage!
   Tendons in the hobgoblin's shield arm snap.
Cast Roundtime 3 Seconds.

In the following I will refer to "13" as the success margin, "Target staggered by strong strike" as the damage message, "25" as the raw damage, "...10" as the crit damage, and "Tendons in the hobgoblin's shield arm snap" as the crit message.

Success Margin Raw Damage
Damage Message Min Max Min Max
A weak blow 1 5 4 4
A light strike 1 10 5 9
A nice hit 1 39 10 14
A strong strike 6 60 15 19
A painful blow 6 90 20 24
Target staggered by strong strike 11 91 25 29
A powerful hit 11 92 30 34
A very painful blow 21 88 35 39
Target staggers in pain 22 87 40 44
A masterful strike 41 86 45 49
A most painful blow 41 87 50 54
Target reels from brutal strike 62 90 55 59

This is not a complete table: since I only kept hits that caused a critical hit and hobgoblins only have 60 HP, any higher damage message would necessarily not be kept.

Unlike crit messages, whose damage is defined entirely by the message, the raw damage is limited by the success margin as follows:

Success Margin Min Max
1-5 4 13
6-10 7 22
11-20 10 31
21-40 13 40
41-60 17 51
61-?? 20 59

For example, "A powerful hit" can occur on success margin 11 and above, but can only do raw damages of 32 through 34 if the success margin is 21 or above:

  • of 18 observations on success margins 11-20, 8 did 30 raw damage and 10 did 31
  • of 5 observations on success margin 21, 1 did 30, 1 did 31, 1 did 32, and 2 did 33
  • of 142 observations on success margin 21+, 23 did 30, 32 did 31, 28 did 32, 26 did 33, and 32 did 34

This effect also truncates the lower range, for example "A strong strike" can occur on success margin 6 through 60, but can only do raw damages of 17 through 19 when the success margin is 41 or above:

  • of 137 observations on success margins 6-40, 22 did 15 raw damage, 32 did 16, 30 did 17, 29 did 18, and 24 did 19
  • of 30 observations on success margins 41-60, 6 did 17, 8 did 18, and 16 did 19

Raw damages appeared to be distributed symmetrically within any damage messages when taking the constraint of success margin into account.

.

Three successes generated results outside of these bounds.

  • 35 success margin with 42 raw damage
  • 1 success margin with "A painful blow" and 22 raw damage
  • 3 success margin with "A strong strike" and 19 raw damage

In all three cases the hobgoblin was lying down, which is something they do ambiently. Per Warding, "If the target is incapacitated (for example stunned or immobilized), most spells receive +15 to effective warding margin.", so it would appear prone gives some incapacitated bonus for 702, although only +6 would be required to make all three successes fall within the bounds. For future data collection I will ensure hobgoblins are not lying down or sitting, which they also do ambiently sometimes.

.

Damage messages are not distributed symmetrically within a success margin range.

Damage Message 1-5 6-10 11-20 21-40 41-60 61-??
A weak blow 7% - - - - -
A light strike 55% 12% - - - -
A nice hit 38% 33% 28% 7% - -
A strong strike - 34% 24% 18% 8% -
A painful blow - 21% 16% 17% 16% 12%
Target staggered by strong strike - - 23% 21% 16% 12%
A powerful hit - - 10% 17% 11% 14%
A very painful blow - - - 17% 16% 12%
Target staggers in pain - - - 3% 14% 12%
A masterful strike - - - - 13% 12%
A most painful blow - - - - 6% 13%
Target reels from brutal strike - - - - - 13%

If we assume raw damage values are distributed symmetrically within a success margin range, the true values are:

Damage Message 1-5 6-10 11-20 21-40 41-60 61-??
A weak blow 10.0% - - - - -
A light strike 50.0% 18.8% - - - -
A nice hit 40.0% 31.3% 22.7% 7.1% - -
A strong strike - 31.3% 22.7% 17.9% 8.6% -
A painful blow - 18.8% 22.7% 17.9% 14.3% 12.5%
Target staggered by strong strike - - 22.7% 17.9% 14.3% 12.5%
A powerful hit - - 9.1% 17.9% 14.3% 12.5%
A very painful blow - - - 17.9% 14.3% 12.5%
Target staggers in pain - - - 3.6% 14.3% 12.5%
A masterful strike - - - - 14.3% 12.5%
A most painful blow - - - - 5.7% 12.5%
Target reels from brutal strike - - - - - 12.5%

This suggests raw damage is distributed symmetrically and the damage message is assigned after, so we can generate an average expected raw damage per range by simply averaging the minimum and maximum. In the following table "observed" uses the observed rate of damage messages and "model" uses these proposed values.

Success Margin Observed Model
1-5 8.5 8.5
6-10 15.1 14.5
11-20 20.0 20.5
21-40 26.4 26.5
41-60 33.7 34
61-?? 39.7 40

Using the model gives an average raw damage per cast for a character with 80 endroll before d100 of 23.3.

.

Crits follow Disruption_critical_table in all but three cases:

  • "Strike to the hobgoblin's left/right arm sprains biceps." can cause 3 or 5 critical damage
  • "Throat strike causes the hobgoblin to cough." can cause 2 or 5 critical damage

Crit rank is governed entirely by success margin, with no effect from damage message.

Success Margin 1 2 3 4 5
1-10 65% 35% - - -
11-20 38% 31% 31% - -
21-60 - 37% 33% 30% -
61-?? - - 30% 33% 37%

It seems probable the splits are 67%/33% for the first tier and 33%/33%/33% for all others. This also lets us prove that the apparent duplicate crits are actually rank 0 crits, and the 1-10 range is actually 33%/33%/33% with rank 0 rank 1 and rank 2: between the three crits there are 20 cases where the success margin is 11+, and the lower crit damage appears in 0 of the 20. The rank 0 of the disruption table must be otherwise identical to the rank 1 except for neck and arms, which is not entirely academic since it slightly changes the expected crit damage, which for rank 0 requires us to adjust for frequency of body part hit, deflecting from 5 to 4.7 for rank 0.

Body Part Rate
Abdomen 13.1%
Back 11.1%
Chest 12.6%
Head 06.9%
Left Arm 11.3%
Left Eye 01.1%
Left Hand 02.3%
Left Leg 05.6%
Neck 06.0%
Right Arm 10.1%
Right Eye 02.7%
Right Hand 06.0%
Right Leg 11.1%

Notably, the values for head+neck+eyes sum to 16.8% versus a dev stated value of "a little over 16%" from Bolt_spell/saved_posts so it's probably just whatever the random body part percentages are. This gives total expected damage of:

Success Margin Raw Crit Total
1-5 8.5 6.5 15
6-10 14.5 6.5 21
11-20 20.5 10 30.5
21-40 26.5 15 41.5
41-60 34 15 49
61-?? 40 20 60

.

.

Areas left to research:

  • max crit rank
  • full damage messages

.

.

.

Updates with lesser orc and lesser burrow orc data:

Success Margin Raw Damage
Damage Message Min Max Min Max
A weak blow 1 5 4 4
A light strike 1 10 5 9
A nice hit 1 39 10 14
A strong strike 6 60 15 19
A painful blow 6 91 20 24
Target staggered by strong strike 11 94 25 29
A powerful hit 11 92 30 34
A very painful blow 21 92 35 39
Target staggers in pain 21 96 40 44
A masterful strike 41 95 45 49
A most painful blow 41 92 50 54
Target reels from brutal strike 62 96 55 59
A devastating hit 61 96 60 64
Massive internal disruption 81 97 65 70
Success Margin Min Max
1-5 4 13
6-10 7 22
11-20 10 31
21-40 13 40
41-60 17 51
61-80 20 61
81-?? 23 70
Body Part Rate
Abdomen 12.7%
Back 11.4%
Chest 14.6%
Head 06.1%
Left Arm 10.7%
Left Eye 01.1%
Left Hand 02.2%
Left Leg 06.3%
Neck 06.1%
Right Arm 10.1%
Right Eye 02.6%
Right Hand 06.0%
Right Leg 10.1%
Success Margin Raw Crit Total
1-5 8.5 6.5 15
6-10 14.5 6.5 21
11-20 20.5 10 30.5
21-40 26.5 15 41.5
41-60 34 15 49
61-80 40 20 60
81-?? 46.5 20 66.5

The crit ranks generated in the 81-?? range show no evidence of being different than the crit ranks generated in the 61-80 range, and the additional body part locations do not change the first tier's (rounded) expected crit damage of 4.7. I have added the additional data to the raw data page, and updated the main 702 page.