Kitty Kad NFT Rarities (Part 2 — Rankings)

After a few more weeks of counting kitties one by one, we’ve finally got another update.

As always, here’s a summary of the article

1. Overview of rare features
2. Different ways rarities were calculated
3. Community Rarity Ranking Lists

Overview of rare features

In the last article we shared the high level features and their rarities. Some attributes (like dot eyes) and items (like rice field hats) clearly stood out as rare.

#2401 with a rice field hat

There are also a bunch of sub-features. These are things like colored ear-tips, not having dimples, bracelets when wearing a jacket etc.

#7 with colored ear tips and dimples

There are a lot of these sub-features and you will soon be able to see them when looking at the kitties on kitties.kittykadcom . You can also see the features your kitty has in this excel.

The question is, how do we calculate how rare a kitty is 🤔

Different ways of calculating rarities

There is no right way to calculate rarities. But there are many good ways to approach it.

Basic Approach

In general — they all involve some way of calculating the rarity of individual features and then combining them together (multiplying, averaging etc).

Everyone has their own preferences, but to me multiplying the rarities seems like the best approach.

Advanced Approaches

There are a few tricky things for Kitty Kad Gen 0s which need some more enhancements to a basic rarity multiplication formula

1. Some traits have “conditional” probabilities. For example — different bracelets are only visible on kitties with jackets.
   This requires the probability of the feature + subfeature to be considered together.
   For all the math fans out there it means P(Feature) = P(Template=active, Sub feature=active)
2. Some features can be hidden (i.e. glasses cover up eyes) and sometimes both can be seen (a monocle covers only one eye).
   This requires assigning some values to the hidden features by giving a default score/probability if the feature isn’t seen.
3. The raw data of kitties reflects how the code works, more so then how the probabilities should be calculated.
   This requires standardizing empty fields in the data to be the same as the default values etc.

Community Input

Getting feedback from the community on the perceived rarity is also a big factor. For example “gold crown” may not be a trait, but the community could decide that a crown that looks gold deserves special recognition.

This can be done via people’s subjective opinions, or even having scores submitted by the community to be added to the formulas

For example — if the community feels like the kitty below has a cool “ghost theme” due to the eyes and color scheme, but ghost theme is not an official feature. The community can decide to rank this kitty higher (which will be reflected in the kitties price).

#6682 — Ghost looking kitty

Rarity Ranking Lists

Every Gen 0 kitty is unique — Kitty Kad

First of all, I want to say that it is up to the community to determine rarities and rankings for each kitty.
Math can only get us so far and some formulas even can take community input as part of the calculations.

For this reason, there is no “official” Kitty Kad ranking list (this is also similar to what other projects like Crypto Kitties went for).

However, I have worked closely with members of the community to help them prepare rarity / ranking lists to get things started. The two are below

1. Advanced ranking list — (old list , see update)
   This one is the most advanced one I’ve seen. It considers a lot of factors like conditional probabilities, hidden features, advanced cleaning up of data etc.
   More details on how this method works can be found here
   (credit — this was prepared by Kitty #8631)

#942 — Rarest “advanced ranking” kitty

2. Basic ranking list (note — this is now based on out-dated data, since its not used it hasn’t been updated, see update)
This one computes rankings in a way that is common for most NFT projects, but misses out on some of the nuances specific to kitties.
This was prepared by an anonymous member of the community.
Full calculated probabilities can be found here

#3944 — Rarest “basic ranking” kitty

For anyone interested — raw feature data can be found here.

As a reminder — at the end of the day it is up to how the community wants to judge kitties, the math above is just a starting point.

P.S. you might be wondering what is my personal favorite kitty? Well that’s obviously kitty #1, wouldn’t sell it for anything in the world ;)

Me!