### Author Topic: How does Mr. Robot estimate trinkets?  (Read 35138 times)

0 Members and 1 Guest are viewing this topic.

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### How does Mr. Robot estimate trinkets?
« on: January 14, 2011, 02:54:01 PM »
Trinkets are pretty simple in Cataclysm:

There is usually one "passive" stat and one "proc" or "use" effect. The item budget for these two is generally equal. Our estimates are easiest to show by example:

The passive stat is 363 crit. The equip effect is a chance to grant 2178 spell power for 20 seconds. It has been observed that the up time for trinket effects in cataclysm is about 16.67%, or 1/6 of the time. If we multiply 2178 by 1/6 we get: 363, as expected. So, for Mr. Robot's purposes, this trinket is:
363 crit
363 spell power

Now, lets look at an "on use" trinket:
This has 321 mastery and on use grants 1605 agility for 20 seconds with a 120 second cooldown. If we were to just take the duration of the buff divided by the cooldown, we'd get 1/6 again. Multiply 1605 by 1/6 and you get... 267.5. This number doesn't look equal to the passive stat... If we instead Multiply 1605 by 1/5, we would get 321. So, it seems that blizzard is expecting "on use" trinkets to be active 20% or 1/5 of the time during a fight. Well, if a fight were 5 minutes long, that would be true. You could use the trinket once near the pull, then once again near 2:00, again at 4:00, and when the fight ends you would have had the buff for 60seconds/300seconds = 20%. Now, as the fight lengths diverge from odd numbers... the up time gets closer to 16.67%, like you would get in a 6 minute fight. In a 6 minute and 15 second fight... you could actually get your up time to 80/375 = 21.3% up time. Also, these trinkets can be timed up with cooldowns for greater effect. So, we have decided that it is safe to average all of this out and assume the value of a use effect is equal to the value of the passive stat on a trinket. Sometimes it is better, sometimes it is a little worse. This trinket would be:
321 mastery
321 agility

"Stacking Trinkets"
These are easy to calculate, for example:
285 haste rating and 17 spell power multiplied by 20. This trinket is:
285 haste
340 spell power

Now, for this particular trinket, that 340 spell power is only going to be realized by a class with multiple DoTs or HoTs up constantly. An elemental shaman, for example, only has one DoT that ticks once every 3 seconds. It would take a full 60 seconds just to stack up. That is why we don't consider this trinket for that spec.

There are some anomalies:
This has 321 intellect and on use grants 1926 spirit for 20 seconds with a 120 second cooldown. It seems that some items have higher values on trinket use and procs. This item would be:
321 intellect
385 spirit

This item is just plain awesome for mana efficiency. We still estimate it simply as:
380 spirit
380 haste

We ignore the on use for this one and just use the spirit total.

These are unpredictable and very spec-dependent. Right now we consider it best in slot for:
Frost DKs - DW
Enhancement Shaman

Otherwise we have a low-ball estimate that makes it better than blues, but not better than other ilevel 359+ trinkets with desirable stats. Let us know if you think any other spec should be added to the list.

This is also a weird one. It appears to have a 45 second internal cooldown. So, assuming a 30% chance to proc, it will be up about 12 out of every 50 seconds, or, 24% of the time. That averages out to 384 intellect before kings and armor specialization. So, 423 intellect. Of course, depending on the fight length that up-time could vary. We are going to use 400 intellect for the proc estimate. The damage portion of the proc scales with spell power and is affected by specs with fire damage bonuses. This is difficult to estimate. Base that is 1200 damage every 50 seconds, or, 24 DPS. If the spell power coefficient is 0.10 as some have suggested, spell power is around 8k-9k buffed with raid gear. Multiply by the 8% spell damage debuff and you have something around 2200 damage. That would be 44 DPS. Pretty small, but not negligible. That is like adding another 50 spell power on top, or something like that. So, we will consider this trinket:
321 mastery
400 intellect
50 spell power

This item looks kinda cool, but a little weak. We don't really consider it.

This item varies so much from spec to spec that we don't consider it. If it is proven to be BiS at a particular gear level for some specs, we will hard code it in.

Another weird item. That shield will get chewed up in a raid, so we just have it estimated at:
285 spirit
« Last Edit: January 15, 2011, 10:57:49 AM by Revulva »

#### MagicSN

• Newbie
• Posts: 2
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #1 on: February 04, 2011, 04:28:15 AM »
>The passive stat is 363 crit. The equip effect is a chance to grant 2178 spell power for 20 seconds. It has been observed that >the up time for trinket effects in cataclysm is about 16.67%, or 1/6 of the time. If we multiply 2178 by 1/6 we get: 363, as >expected. So, for Mr. Robot's purposes, this trinket is:

This data is at best my knowledge (and I have the trinket in question) wrong. With the 16.67% uptime I guess you assume a
120 seconds ICD (20 sec. / 120 sec. = 16.67%). In fact the trinket has a 90 seconds
uptime (you can also - asides from practical tests - get this same number confirmed at various
discussions about it on the net.

Is it possible you fix the numbers for the trinket calculation on the website? Thanks in

MagicSN

#### faelkin

• Newbie
• Posts: 6
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #2 on: February 05, 2011, 04:01:05 PM »
Hi all,

great job with the reforging calc.
Regarding trinkets: I did an overview of caster trinkets in cataclysm that has been thoroughly investigated by the folks at EJ, so I think it is quite accurate. (assuming that the cast times for proccing the effects are similar)

Give it a try and see how it works for you.
« Last Edit: February 05, 2011, 04:05:04 PM by faelkin »

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #3 on: February 06, 2011, 07:57:35 AM »
Thanks much for the link. I have seen other lists similar to this also on MMO-Champion. I'll make use of this when I update the trinket estimates.

It looks like there is enough actual game data now to get a good estimate on the actual up-time of trinket procs. I am going to go through the whole trinket list and update the estimates based on this information. I'll update this original post with the information that I use for each trinket.

#### MagicSN

• Newbie
• Posts: 2
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #4 on: February 07, 2011, 02:25:34 AM »
Thanks much for the link. I have seen other lists similar to this also on MMO-Champion. I'll make use of this when I update the trinket estimates.

It looks like there is enough actual game data now to get a good estimate on the actual up-time of trinket procs. I am going to go through the whole trinket list and update the estimates based on this information. I'll update this original post with the information that I use for each trinket.

I checked my recent Logs again, and it seems the claim of "90 seconds ICD" I wrote in my
earlier post was wrong (but 120 seconds was wrong also). My logs conform with the
100 seconds ICD also listed in faelkin's Spreadsheet for the Bell.

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #5 on: February 07, 2011, 04:11:28 PM »
Hi all,

great job with the reforging calc.
Regarding trinkets: I did an overview of caster trinkets in cataclysm that has been thoroughly investigated by the folks at EJ, so I think it is quite accurate. (assuming that the cast times for proccing the effects are similar)

Give it a try and see how it works for you.

The one calculation in there that I took issue with was the Theralion's Mirror (Heroic). You used a different "Time between spell casts" between the heroic and non-heroic versions. The non-heroic version you estimate at 335 mastery. I think the heroic should be 379, not 342, for comparison purposes.

#### faelkin

• Newbie
• Posts: 6
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #6 on: February 10, 2011, 06:22:07 AM »
thanks for checking revulva. you were right, I fixed it.

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #7 on: February 10, 2011, 08:31:34 AM »
Just so you know, I have incorporated the values from your spreadsheet into the estimates that are live on the site right now.

I created a similar spreadsheet for melee trinkets, but I don't have information for every trinket yet. I will be converting my spreadsheet to one of those shared google docs when I can so that I can hopefully crowd-source it.

• Newbie
• Posts: 1
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #8 on: February 10, 2011, 11:32:43 AM »
How does Mr Robot evaluate Heart of Ignacious?

Unlike Jar of Ancient Remedies, the stacking buff doesn't have a 'cooldown' period after the on-use, so the click is actually quite valuable (1605 haste rating for 20s every 2m.)

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #9 on: February 10, 2011, 11:44:12 AM »

#### potatoboy

• Newbie
• Posts: 4
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #10 on: February 25, 2011, 12:34:58 PM »
There is much discussion on EJ and several hunter blogs (WHU comes to mind) about DMC:Hurricane being changed to use the Physical Hit.  Femaledwarf has updated their modelling to reflect this and it appears to be accurate with in game results.  This makes the trinket much more desirable and on gearset in Femaledwarf appears to be a better choice than Essence of the Cyclone.  Have the Hurricane trinkets been reevaluated to reflect this?

#### dl3mk3

• Newbie
• Posts: 3
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #11 on: March 14, 2011, 01:36:55 PM »
I had a question about the warlock (specifically affli) BIS trinket list.  I'm currently running Talisman of Sinister Order and normal version of Anhuur's Hymnal.  Mr. Robot lists Sorrowsong as an improvement however when I run SimC i find that it is a DPS loss to replace either of them with Sorrowsong.  I guess i'm wondering if there are some item interactions that make sorrowsong specific to my character or if perhaps the non-346 level nature of the trinkets has been overlooked by SimC (the dps list on EJ doesnt take into account trinkets under 346).

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #12 on: March 14, 2011, 01:57:29 PM »
I would have to see your entire gear set to answer this question. Provide us with a link to your character on Mr. Robot or a region/realm/name and I can probably explain.

#### dl3mk3

• Newbie
• Posts: 3
• Karma: 0
##### Re: How does Mr. Robot estimate trinkets?
« Reply #13 on: March 15, 2011, 06:57:53 AM »
Rainboraider on US-kultiras (i am not optimized per Mr. Robot yet).  I ran using the current SimC and reforging sorrowsong reforged for hit when replacing the hymnal, and haste when replacing talisman.

Still came up with dps loss of 54 and 22, which is close enough to call all three trinkets equivalent.

#### Swol

• Mr. Robot's Drinking Buddy
• Hero Member
• Posts: 8644
• Karma: 266
##### Re: How does Mr. Robot estimate trinkets?
« Reply #14 on: March 15, 2011, 07:28:10 AM »
The item lists rank items independently of other gear.

Sorrowsong gives an average of 292 spell power, according to calculations others have done. Simply looking at the item list and replacing your trinket with hit for a trinket that has mastery... results in you being unable to reach the hit cap with your current gear:

Every item that can be reforged to Hit is reforged!

Replacing the other trinket creates a setup that is for all practical purposes identical.

You always have to look at the entire gear setup to decide if an item will actually be an upgrade. I am a little doubtful of how SimC is modeling Sorrowsong... it seems to me that it is being a bit undervalued, but I can't be sure.
Sorrowsong with mastery reforged to hit gives:
171 mastery
114 hit
292 spell power (average)

Talisman of Sinister order gives:
234 intellect
160 mastery (average)

Sorrowsong really *ought* to be a better trinket.
« Last Edit: March 15, 2011, 07:35:48 AM by Revulva »