@flipsidecrypto What are some ingredients that goes into constructing a retention table. ✅ You can control who gets counted inside.✅ You can control what to measure.✅ You can control the time period. Sorry I had to reuse this picture lol. Needed a catchy thumbnail

There hasn't really been a fundamental change since 2021/2 in how NFT value accrual works, or am I totally out of the loop? Maybe except the igloo universe, but the rest haven't really changed.

TIL if you do a full copy from google docs, including images, and paste into Apple Notes, everything gets copied over. Including pics.

Wholesome story in 2 pics starring OpenRouter and CTO of cloudflare.

I know it sounds "duh" but the liquidations can be more violent that usual IMO.

I'm not sure how much the ETH treasury meta impacts ETH, but prices may be increasingly reflexive. Not a good or bad thing, just that:. 1) Be aware that prices may surge or drop sharply.2) If ETH drops in prices, instead of buying dips, keep an eye out for liquidations first.

Interesting that virtually every Flipside AI post I've made, people are talking about data issues 🥲. Issa oke, we need more experimentation on how we interact with data. Let's see where this goes 🙌.

Talos strangely feels familiar in its mechanics that I can't put a handle on. Some quick thoughts . 1) Why use Talos to expose yourself to onchain yield when you can do it directly? You expose 100% to the yield and the yield isn't shared with others. If you know the underlying.

My new "find me some pessimism" workflow is that you take a claim that is too good to be true, input it into an LLM of your choice, and then ask why it doesn't work or scale. Here is one: What if a trusted manager recommends a promising individual into a private recruiter group.

Ofc the exception to this is that morpho can provide superior yields on Arb and I definitely think this will happen (ofc risk profile is different) . Then the bet becomes if you believe that users are indifferent to these risks and would just churn from aave to morpho.

afaik if you want to make a poor man's db on the blockchain, use EAS to help you form a schema and then just append only all the way. Then you can use Dune to draw the data out. Not the right use case btw.

The difference between an AAVE and Morpho on Base is that Morpho users tend to heavily borrow alt. tokens other than ETH and stables and then deposit more stables. Aave users usually deal with stables and ETH. This means that Morpho's main advantage on Base is in alternative

In any case, if you are projecting and modelling the number of users or trades, you should model the "retained users" like decile 1 and 2. So, if 100 users come in, only 10 people would generate 88% of the volume. In other words, models projecting user-volume outwards are.

Ranking and placing equal number of customer per decile, we see that the top decile trades for a disproportionate amount of volume (they represent 88% vol). Then once you reach decile 2, the number drops of very fast (showing large skew) to something more sane. I reckon

Most pump traders trade 1-2 times a day, 2-3 tokens at a time. This means the token mindshare is extremely constrained. Don't expect like, 5 narratives, to run at the same time lol. Just the top 3 is enough

I think another cool feature that Flipside may be unintentionally doing, is to help users reason about the data as seen by the extra unrequested metrics. If data is a part of a reasoning flow, then your JTBD is not just providing data but help in reasoning to achieve goals too.

With token diversity as predictor of long term user engagement for lending protocols, this means that there is a lot more work to do for lending protocol to segment, understand, and cater to their users better. You can see that multi asset enjoooyers have a highest retention

Flipside AI is pretty cool to go from analytics idea -> show me the data. Making retention chart on aave and morpho in a few minutes is fun.

Following the book from Tech Revolutions and Financial Capital, when the bubble bursts, a few things happen:. 1) As expected, apps are likely to fail because they chase hype and narratives (money generating money). Financial capital detaches from productive capital, and.

Courtyard io has a powerful business model actually. The rest depends on whether the card collecting community are willing to break their own user flows for it. Need more customer acq. Good moat but TAM is capped geographically and by the size of the collectables industry.

Lending protocols are quite interesting from a social mechanic because. 1) As a borrower, you are trusting that other borrowers do not make big borrows that would spike your rates. 2) As a borrower, you are trusting that depositors not to make big withdrawals to spike your rates.