Content from this post originally published on SplinterSkins.com
Skins - Special Items
Splinterlands skins are special because they provide no gameplay benefit, they're only cosmetic. Skins are also never earned by bots, they can only be obtained by purchasing them in the shop.
Splinterlands Skins were purchased by 1,939 unique players with a Hive account as of April 22, 2022 (and several more without a Hive account). Skins are the only purely cosmetic item to be listed for sale directly in the Splinterlands shop, with their own dedicated tab. This made skins the most accessible cosmetic item for the average player to see and purchase.
Looking at Splinterlands Skin purchases provides the best representation of real players, not bots. Players who enjoy the game and who play for fun, players with good aesthetic taste, and the most dedicated - these are the types of people who purchased skins. Skins are a special item for especially awesome players. There's no reason for bots to purchase skins, but we checked just in case, and they mostly haven't - keep reading.
All results are from a snapshot taken on April 22, 2022.
Splinterlands skins are popular - they've been purchased by 1,939 unique players with a Hive account! 1,610 players purchased individual skins, and 478 purchased skin sets. Those numbers add to more than 1,939 because 149 players purchased both a set and individual skin(s).
While the above chart shows that more players purchased individual skins than sets, sets still made up the majority of overall skin sales. This is because of the increased quantity of skins that a set contains, with a single set of common skins holding 24 skins, a rare set 23, an epic set 17, and a legendary set 10. Let's look at unique player purchases by rarity:
When you're buying an individual skin (not a set), the Splinterlands UI only lets you buy one of each. Sets had no such limit. Even so, the vast majority of set purchasers only purchased one set or fewer of each rarity. Out of 828 set purchases of a unique set rarity for the purchasing player, only 23, or 2.7%, purchased more than 1 set of the same rarity:
This suggests that most players bought skins for their own cosmetic use, not for later resale. Skin buyers are an exceptional group of committed people - real players, not bots. Splinterlands botnets follow a pattern that the username is most often a text username followed by a number, I.E., "Jarius1". The botnet then also controls "Jarius2", "Jarius3", and so on. To check if any skins were purchased by bots, we analyzed the list of purchasing players to see if some players had the same text username followed by different numbers:
*Out of 1,939 unique skin purchasers, there was only one group of 5 following the botnet pattern: "sl-proxy-1", "sl-proxy-2"... on through "sl-proxy-5". This is technically a botnet, but it's controlled by Splinterlands itself to purchase for players who haven't purchased the Spellbook. We included the transaction data from these proxy purchases in our analysis, but excluded them from the count of 1,939 unique skin buyers.
Data Accuracy and Variances
We checked that the data from HiveSQL matches up with transactions found on other explorers like Hive Block Explorer, so the data we used is accurate. While the skin purchase data was very close to what would be expected, there were a few slight anomalies (less than 1% of the total). It's possible a few transactions went through on the blockchain, but did not go through on Splinterlands.com (most likely), or that Splinterlands accidentally sold a few extra of certain skins and skin sets (less likely). Here's how our results match up with the expected distribution:
- Excluded Transactions that Explain Variances: We excluded one legendary set purchase, one common set purchase, and several individual skin transactions from our analysis because they had a blank "required_posting_auths" field. The "required_posting_auths" field is how we tied purchases back to the purchaser's account for our analysis, so we excluded them for consistency. These excluded transactions explain minor variances in the numbers for certain skins.
- Legendary Skins: Our data shows most legendary skins as having 99 total sold, instead of the expected 100. This is probably because we excluded one unverified legendary set purchase. We excluded it because it did not have the "required_posting_auths" field filled out with the purchasing account, but it's likely that this transaction was actually valid.
- Epic Skins: Most epic skins have around 203 sold in our data, instead of the expected 200. The likely explanation is that 3 set purchase transactions went through on the blockchain, but did not go through on Splinterlands.com. One method to know which 3 transactions failed would be to track DEC payments from every epic set purchaser's account to the smart contract, as well as any refunds. Then, whichever transactions don't have a corresponding DEC payment for the purchase, or received a refund, would be the 3 failed transactions. We decided that would take more time than it's worth just to find a few failed transactions.
- Rare Skins: Our data for rare skins matches up very closely with what's expected. Alric Stormbringer (ID 16) sold exactly 500 copies, combining individual skin purchases and set purchases, as expected. Other rare skins which had not sold out at the time of the April 22 snapshot had also sold the expected amount, or close to it.
- Common Skins: Our data for common skins matches up closely with what's expected. The only common skin to have sold out at the time of the snapshot, Divine Healer (ID 34), recorded 209 individual sales to Hive accounts. Added to the 791 common set purchases, this brings us to the exact 1,000 purchases expected for it to sell out. There was one additional purchase from a user without a Hive account for this skin, however, so either that purchase or another of the purchases likely didn't go through.
TL;DR of Accuracy
The data we used is 100% accurate to what's on the blockchain. We have also verified that this data from the blockchain is over 99% accurate to what Splinterlands themselves displays in their shop for skin sales.
Methodology - How to verify this data for yourself:
First, register an account on HiveSQL.io, which maintains a convenient database of Hive transactions. Next, open Microsoft Excel, and go to the "Get data" tab. Follow these instructions the folks at HiveSQL have so nicely provided.
Next, you'll want to pull the relevant data on skin purchases from the SQL connection you just set up. See HiveSQL's TxCustoms docs for more info.
This is the code we used to get the skin set purchases:
tid, json, required_posting_auths
This is the code we used to get the individual skin purchases:
tid, json, required_posting_auths
Now that you have the data in Excel, you can analyze it using Excel tools and formulas. Excel's SEARCH, COUNTIFS, COUNTIF, ISNUMBER, CELL, and SUMPRODUCT functions were especially useful for our analysis and data organization. We created the charts in Canva.
Our snapshot was Friday, April 22, with many common and rare skins still for sale. Your results will vary because of later purchases.
Special thanks to @arcange and everyone else behind HiveSQL for helping make this post possible.
Learn more at SplinterSkins.com
SplinterSkins does not purport itself to have any affiliation with Splinterlands, or any employees, developers or partners thereof. None of this content is financial advice, legal advice, tax advice, investment advice, or any other type of advice. This content exclusively represents the opinions of the author(s).
What you did requires great knowledge.
What could you conclude out of this, except that your data is accurate?
What will your next step be?
Our next step was to announce an airdrop for all skin buyers!
We plan to do custom NFTs for active community members and more in the future.
I personally appreciated your Splinterlands card power analysis @groentje. Messaged you on PeakD 🍻