Rarity Methadology

Rarity is a subjective matter. Some of the platforms use similar regular algoritms while some others use have totally different calculation methods.

We will first start with grouping the rarity platforms in terms of similarity to each other:

Mainstream websites that have a similar ranking algoritm ("a" represents that they are very close to each other and "b" represents they might somewhat differ from others)

• Trait Sniper (a)

• NFTinit (a)

• NFTNerds (a)

• Blur (b)

• Rarity Tools (b)

Mainstream websites that have rarities which can greatly differ from others above and each other.

• Rarity Sniper

• Openrarity (Opensea Built-in)

Trait Count: On vs Trait Count Off

Trait count stands for the number of traits (properties) an NFT has on itself. Making the rarity calculation with the "trait count: on" selected will assign a rarity value on the number of traits the NFT has.

In an imaginary collection that consists of 10 NFTs if 9 of the NFTs have 3 traits and only one them has 1 trait. The NFT with 1 trait would gain more from the trait count as in terms of trait count it is the rarest one with only one trait.

In another example, if in a collection 10 NFTs, 8 of them have 3 traits, one NFTs has 1 trait and another one has 15 traits. The NFT with 1 trait and the NFT with 15 traits would get the same score from the Trait Count value. Keep in mind that we are not talking about the overall rarity of the NFTs. This is only about the score each NFT is getting from the Trait Count attribute.

Real Example of Trait Count: On vs Trait Count Off

You can see the rarity calculation of CloneX below, with the Trait Count: On, rank 2 and rank 3 of the collection are trait count dependent. As a side note, NFTinit as default shows the most valuable traits on the left-most, this shows that these two NFTs are getting the most rarity points from their trait counts. There are only 2 NFTs which have 11 traits in CloneX thus trait count makes these two NFTs rarer.

On the top right corner you can switch between the calculation methods Trait Count: On and Off. When the calculation is switched to Trait Count: Off, which were previously rank 2 and rank 3 will lose rankings. As a matter of fact, when the "Trait Count: Off" is selected, previous rank 2 drops to rank 314.

And previous rank 3 becomes rank 344:

So is trait count:on a bad calculation method?

The short answer is kind of yes, but not always. It depends on the trait structure of the collection. If there are no special traits, it would matter more if an NFT has significantly a higher or lower number of traits, but this is something we don't see too often anymore. Back in the day people used to care more about rarity, it is not that it doesn't matter now, however, now traits have individual floors and they usually matter more than the rarity. We will talk more on that later.

Let's look at an example where the trait count actually is important. As you see in the below picture Genuine Undead has a floor of 0.34 as of 03-Feb-2023.

However, if the NFTs with which only have 3 Traits are chosen, the cheapest NFT is being sold for 0.5 ETH. Which makes the floor price of NFTs with 3 Traits 50% more expensive than the floor price of the collection itself:

If you go to Opensea you won't see Trait Count as an attribute for Genuine Undead:

But still, people value trait count as only 260 NFTs have 3 Traits:

As you see above, having more traits is not neccasarily a better thing, it could've been if for instance there was fewer NFTs which had 4 Traits, but not in this case.I personally find Trait Count: Off a more reliable method as it focuses more on the rarity of the traits.

Calculation Deep Dive for Nerds:

1 / (Trait Count/Collection Asset Count*100) / Trait Type Count * 1000

Trait Count: The number of traits NFT has

Collection Asset Count: Total number NFTs in the collection

Trait Type Count: The number of properties under a trait (for instance in the above screenshot Trait Count has a tyTrait Type Count:pe count of 5).

It is important to understand why trait type count is a divisive. As the trait type count increases the trait becomes less significant. Let's go over two cases to demonstrate this statement:

• a) Every single NFT has Trait X in a collection of 10.000

• b) Only 100 NFTs have Trait Y in a collection of 10.000

In this setting, since only 100 NFTs have Trait Y, they get more rarity points. Because 9.900 NFTs don't have any Trait Y, on the other hand, all NFTs in the colleciton has a Trait X.

We are going to make the calculation together for the Back: Blue trait for Moonbird Oddities collectin and see how much score an NFT would get from having that particular trait. We are going to fill the numbers for the formula:

1 / [(Trait Count/Collection Asset Count*100) / Trait Type Count * 1000]

Background: Blue exists in 1636 NFTs, so the trait count of the particular Background: Blue trait is 1636 as you see in the screenshot blow.

Asset count of Moonbird Oddities is 10.000

Trait Type Count: (which we talked in detail above) is the number of properties under a trait. The Background trait has 13 sub properties:

Finally if we put put the number in the formula we will get:

= 1 / [1636/10000*100) / 13 * 1000

And the result will be: 4.7 and this is how the score for each trait is calculated. The sum of all the scores creates the Rarity Score which is 411 for the below example and 4.7 of 411 is coming from the Backgroun: Blue trait (decimals not shown in the interface for user experience).