Cooking Tier Lists

Ranking changelog version 3.0.1

• Fixed bug with support Pokémon support impact
• Added Musharna
• Added Cresselia
• Added Darkrai
• Added Blissey

How to Use
Select your tier list from the dropdown.
Choose level 60 or 50 version.
Choose pot unlimited or limited version.
Hit the submit button.

When changing input settings you have to hit the submit button to updated the results again.

Deciding Tier List Placement
The tier lists are produced using simulation-based data.
The underlying algorithms are written by the Sleep API authors, the results are thus not hand-picked. The placements are simply what the data shows.

The delta rank showing for each Pokémon considers all ingredient lists for all Pokémon.
For example let's say previously Golem held rank 3 and 4 and Gengar held rank 5.
In the next version Golem holds rank 3 and 5 and Gengar goes up to rank 4.
Golem will show no rank change, since its best set was and still is rank 3, but Gengar will show +1 since it now beats one more ingredient set.

The Algorithm
In short; Sleep API simulates cooking every recipe for every ingredient list available for every Pokémon and then with every possible team composition to ensure we find the optimal ways to use each Pokémon.

For each Pokémon it does the following: find all available ingredient lists, calculate produced ingredients in a meal window for each ingredient list, check contribution to each recipe, find optimal team compositions to finish each recipe, adjust contribution power dependant on resulting team composition difficulty for finishing the recipe, add filler value.

When calculating a meal window of produce we assume optimal sub-skills and nature as well as max carry size bonus from evolutions, but to avoid locking certain team composition requirements or paid play we avoid good camp, helping bonus and energy skills from influencing the produce.

When finding optimal team compositions for each Pokémon/ingredient-list/recipe we use the produced ingredients and each recipe's required ingredients.
We find which ingredients the Pokémon couldn't cover for each recipe in the meal window. Then we simulate running each Pokémon with every possible team composition to find optimal ways for this Pokémon to complete each recipe.

The raw contribution power to each recipe is calculated as:
relevant ingredients of recipe covered * ingredient base value * recipe bonus% * recipe level max bonus%.

We calculate a filler tax for each ingredient based on their raw value compared to Slowpoke Tail percentage-wise.
The purpose of the tax is, to decrease the impact of fillers to the total score. Better fillers, like Slowpoke Tail, which are less abundant, get less (no) tax:
(ingredient base value / Slowpoke Tail base value) * ingredient base value

The total ingredient filler value is then calculated as:
surplus ingredients not used in recipe * taxed ingredient value.

The final contribution score it calculated as:
(raw contribution power / (1 - (0.2 * team size required for recipe))) + filler value + support skill value.

For the overall tier list we then select the 3 best performing recipes, we boost the value of the very best recipe by 50% and finally combine the contribution powers into a final score by which we rank the ingredient lists of all Pokémon.

For the meal type-specific tier lists we only calculate contribution value for recipes of that type. Since we're looking at a specific meal-type here, we only count the best 2 recipes. We still boost the best recipe by 50% and count a second recipe as fallback.

When deciding our tier list placements we divide all contributions scores in buckets.
Once a Pokémon no longer fits the threshold for a tier they are moved to the next tier and set the threshold for that tier.
The thresholds for the tiers are [S: 90%, A: 80%, B: 80%, C: 85%, D: 85%, E: 90%, F: 90%].
This leads to exclusivity on the outer ends, with upper-middle bracket being a little more lenient, upper-middle is usually where big jumps in score occur.