My 2 cents for what they are worth:
1. I never understood the rarity system. I believe it was a procedural way employed to simulate randomness in order to expand grinding in a multiplayer environment. It requires neither skill nor gut, but mostly luck or playtime, both of which aren't actual interesting gameplay decisions.
2. I believe unpredictability doesn't naturally link to randomness. A rarity system feels like it is strongly relying upon randomness. The idea of dropping ingredients/components at set intervals seems a lot more interesting though it may encourage grinding as well.
3. Strangely enough, your rarity system matches exactly that of Magic the Gathering TCG. And they've been quite successful with it (Legendary is called Mythic Rares, but we get the point).
Ironically however, Magic the Gathering cards are not ranked by rarity to reflect their actual power factor, but their complexity. Simple cards (utilities) show up at common, and will appear in several pro-level decks.
Obviously, this could be applied to your game in this way as well, but I doubt it would make sense. The question you need to ask yourself is, do you really need the rarity system for something?
Overall, I feel like you've assumed that because every game had a rarity system, so should yours.
Game Design is the ability perform problem solving, aka, find solutions to problems that spur from the design. I feel like you're developing a solution for a problem that may not exist.
Also, overcompensating by 'categorization' will have the side effect of making people care less (not) about your actual items, and will rely upon the color coding. The idea here is that they no longer need to think for themselves what they need. Typically, that means any item lower on the scale will have a very temporary lifetime in your inventory (1-5% of the game time you spend, when you're just looking for something better) until 95+% of the items dropped are irrelevant.
I believe the idea here rather would be to find a way to make every item interesting and restricting the player's ability to carry all of them around (is this not why the idea of stash was born?). Specialization of item is one way to achieve this. In Castlevania series, there is a sword with very low power, but the random ability to stone the enemies, which turns out to be quite efficient in many scenarios. Most hardcore players keep a copy in their inventory along with their uber weapons.
Also, some enemies are weaker to spears or swords, etc.
Choosing what to keep needs not to be a decision based on 'the computer told me to keep this through a clever color coding that allows me to waste less time' but rather on 'what do I REALLY need to keep around? What's my general strategy here? What must I keep around just in case?'
And choosing is a risk-reward strategy that actually generates interesting decisions.
And that my friend, is what you should be focusing on
I want to thank you Orymus for giving some good input on where to go, but I don't think it works for what I am doing. Strategically picking and using a weapon only works when you can use more than 1 weapon with a specific class or job, etc. A game like Dark Souls (or Demon Souls) would be a good example. Players can choose through a large variety of weapons, spells, and miracles with no rarity or color coded system involved. Each one can be used for specific strategic purposes, but a game where you have a set class and can only use a specific set of weapons limits the players strategy to "I need to have the best weapon for my class and level". Your Castlevania example is the same. If I was creating a concept for a game like Skyrim or Dark Souls then I would probably not use the rarity system since I could use any weapons I wanted. Games like Skyrim or Dark Souls also do not have the thousands of weapons, materials, usable items, etc. that are available in something like an MMO.
Borderlands (diablo is too) is a good example where they use color coding for the rarity of the weapons. I wouldn't say it was a perfect system, but it worked. The rare weapons were not always the best, but it helped to sort through them easily to find out which ones to pick up. The rare guns would usually have better effects, damage, and special stats, but not always. The rare weapons would still sell for more than a common weapon did. Even if a common weapon had better damage it didn't have that cool special effect that the rare weapon had.
What if I could also have the rarity of weapons automatically change depending on how many are available at the time. Lets say there is a "Legendary" weapon. For some reason a lot of players have been able to acquire the materials in order to craft this weapon. Now there are many of these "Legendary" weapons in the market. If too many become available it won't be "Legendary" anymore, but would change to "Rare" status instead. With so many in the market it would decrease in value anyways.
I didn't see other similar games and just decide that my game concept had to have a rarity system too. I looked through my entire document with so many items, monsters, equipment and the way the gameplay works. Then I decided it would just be easier to have a rarity system to make everything easier to categorize and sort through. I came here to find a better way to categorize things, but all I am hearing is to get rid of it, change the color coding to show an items value, level, type or something else instead. I believe that most developers keep using the color coded rarity system because it works better than people think. There are other ways of doing it.
I have to to categorize all the items, weapons, monsters, and equipment with one categorization system (having more than 1 would be a pain and confuse people). Rarity is the best way I could think of categorizing them by.