Apologies I must have looked at the wrong parse or something. I see that now.

What are your iterations at? Should be 10k for best stat results.
Is this with T15 or your own gear? Single target Patchwerk? Post your setup or log to the results so everyone can get a full view of what you ran.

Might be silly, but are you hit and expertise capped?

I ran it with 50k iterations in a patchwerk style fight using "good" instead of "elite" as skill of the player.

Hit capped and expertise capped simming with my curret gear and reforge. No change to the default priority list , just simple import and simulate.

Edit: I did the sim again here are the Results. Note that if I do use these scales to reforge it is dps DECREASE when simulated.

I can't speak for changes between versions of SimCraft, but I can speak for getting results A, reforging to match them, then getting a different set of results. I can also propose that the Dancing Steel PPM change (not Paladin specific) should cause variance between 5.20-1 and 5.20-2.

Value of stats is based on current stats. The more you have of X, the better Y and Z. If I Crit more often, Haste and Mastery become better. Why? Because with more Haste I make more attacks, thus more Crits. More Crits and more high damage attacks so Mastery has larger numbers to multiply by 1.Y%.

If I reduce my Crit and increase my Haste/Mastery, suddenly Haste/Mastery may not be that awesome because I Crit less often. Lots of Haste? Suddenly it would be nice if more of those attacks Crit or had a harder hitting bonus from Mastery.

Stats do not stand alone. Each stat is based on the others. Change one, you change the relationship. Adding 1 more Stat X may be an upgrade compared to adding 1 more Stat Y, but changing Stat X directly into Stat Y may be a downgrade.

Specific example:
Using slipey's top chart. Adding 1 more Crit could add 1.61 DPS, while adding 1 more Haste would add 1.56 DPS. You'd think changing every 1 Haste for 1 Crit would grant 0.05 DPS, right? So you reforge to drop 100 Haste for 100 Crit expecting 5 DPS... and lose 5, 10, or 20. Why? Because with less Haste, Crit is now worth less than 1.61 DPS each, and with more Crit Haste is now worth more than 1.56 DPS each.

The whole is more than the sum of its parts.

Has anyone tested [Fabled Feather of Ji-Kun] ?
I got my hands on it yesterday. On first sight it seemed that the proc rate is low however, and when pulling boss ( pre-pot, using cd's ) , it's proc rate is noticable. Thats what i personally noticed but not sure though.

Edit: the linked item should be the normal version not LFR one, it's bugging for some reason.

I've got a Thunderforged Normal Mode Fabled Feather of Ji-Kun and its proc rate is all over the place thanks to RPPM. I've had it proc 4 times in a row, one straight after the other as soon as the initial proc wears off. I've also had it not proc for 5 minutes. These RPPM trinkets make it almost impossible for sims to calculate accurate dps potential and are even worse in actual raids.

Not being able to plan when to use cooldowns to match up against trinkets makes it entirely RNG if you top the metres or plummet to the bottom. All you can do is hope the trinkets decide to proc while cooldowns are up, theres absolutely nothing else you can do about it.

Blizzard is being inconsistent, or disingenuous, with RPPM mechanics.

"Every time the trinket fails to activate, there's an increasing chance that it will activate." Yes, thank you, that's precisely how RPPM was advertised to function, according to you, Blizzard. The time-since-last-proc is calculated. The longer this time, the greater chance of a new proc. If you change/increase how this functions, all you're doing is changing/increasing the overall RPPM value. Example: If I have a 5% chance of proc, which gets better by 2% every 3 seconds, and it's increased to 5.5% and gets better by 2.2%, you have directly changed RPPM by 10%. Oh, and this wouldn't prevent a dry spell.

It sounds like they're (also?) adding a max-time-between-procs as a shield against prolonged periods of no-proc. If they are indeed doing so, it's breaking the fundamentals of Real Proc-Per-Minute. You're now guaranteed at least X procs within a timeframe (X being determined by the limiter) and potentially get a lot more if you get lucky. This isn't 'Real' PPM, it's 'Minimum' PPM. You aren't going to average the RPPM number, but some value higher than it.

So if Str/Agi trinkets are getting a 10% boost and RPPM is getting a floor, we're getting something higher than 10% increase on RPPM mechanics. Sounds like 3+ times as much code to do what the old ICD + decent proc chance code did perfectly fine.

That's not how RPPM was previously advertised and worked. It did not use the "time-since-last-proc", but instead the "time-since-last-proc-chance". So as long as you try to proc the trinket continuously with a equal attack frequency, you're proc chance never ever wen't up ( unless there was a CD, during which attempts to proc weren't counted ). That's what made the rppm system so random.

So I personally welcome the change to introduce a more stable mechanism, which tries to guarantee some procs over a typical fight. It makes the new trinkets much more reliable, and should reduce DPS variance a bit. Of course Blizzard needs to find a good formula for the new system, and fitting values for each item. And hopefully they can complete remove the haste regression from it.

Maybe I'm missing something obvious, but what's the issue with haste influencing RPPM trinkets? The only drawback I've seen in practice is that it can sometimes overvalue older-content trinkets due to haste scaling from new gear, in comparison with new trinkets that are available.

Additional details on the RPPM changes are available: Real Proc Per Minute (RPPM) Trinket Changes - MMO-Champion BlueTracker

(Relevant excerpt for STR trinkets)
At first glance this looks very favorable to ret, since I believe one of the more common practices is to pop cooldowns upon entering combat to align with prepotting and maximize usage over the course of the fight. RPPM trinkets having a very high chance to proc on combat start is helpful for at least aligning with that initial burst.