Modelling is all well and good, but I would prefer some logs from PTR that show what % it actually does. So far the only non-theoretical data on the subject is "never surpassed 2.5% of my healing". We seem to be locked into a cyclic argument of "it does less than 3% healing" "but it SHOULD do 3%" "but it doesn't" "but it SHOULD".See that is why I get annoyed. I will repost it for you.

As you can see the diminising returns on spellpower ramps up quite slowly because the denominator is 1+K*spellpower and K<<1. This is why the hc amp trinket with 9% on everything is not really any better compared to same ilvl trinket than the normal or LFR versions are.

To show you how slowly lets redo the calculation with 55k spellpower

1000*0.000104157691/(1+0.000104157691*55000) = 0.015479677928. An increase of 10% in spellpower is only an 8% drop in the conversion.

1800 intellect at 520 ilvl (~40k spellpower) is a 4.75% increase. At 580 ilvl (~60k spellpower) it is a 3.4%. This is though with zero overheal. The diminishing returns on spellpower just don't work like you think at all. Also the value of intellect decreases with overheal, which needs to be taken into account (see below). The 1350 odd intellect you need for a 3% increase at 50k spellpower becomes 3.6% at 40k spellpowerl and 2.6% at 60k spellpower. It would take 110k intellect before it is busted down to 1.5%. I.e. you need your spellpower to nearly tripple in order to cut its value by half.

Notice how you need MORE intellect at 60k spellpower to produce a given increase in healing, but the cleave and amp trinkets pretty much produce the same % increase no matter how much spellpower you have.

Thus the more spellpower you have the better the trinkets are compared with a flat intellect value, because they scale as % points. I took a low value of spellpower to calculate a minimum ceiling. This is the best case scenario effectively.

1000 intellect increases crit by 0.44%, which is ~0.4% healing with 30% aegis and 30% crit. This is why I am ignoring it, an extra thing to calculate that only changes things by +/- 15% or so.

1350 intellect produces a 3% increase at 50k spellpower AT ZERO OVERHEAL. At 30% overheal and assuming linear losses, which is not necessarily valid, you would need 1930 intellect to produce a 3% increase. Based on a more complex modeling of overheal I estimate that you would need ~1800 intellect (not spellpower) for a 3% increase when your overheal is 30%. This modelling includes the 0.8% extra crit you get from 1800 intellect.

When I say multistrike above I mean cleave. The multistrike trinket looks ok on paper, but suffers greatly from overhealing, so it may not work well for holy. The cleave trinket is less good on.

I am always happy to discuss, but with numbers. If you just want to tell me what your gut instinct says, without bothering to do even rudimentary calculations or at least reading through mine, is not a discussion.

If you feel that my model is not valid, feel free to provide your own estimate and lets see if its different.