CryptoPunks "Black Thursday"

Picture of Wall Street Crash

CryptoPunks market cap dropped 29.7% in ETH (435,017 ETH to 306,145 ETH) and 38.5% in USD ($1.82B to $1.12B) from May 11, 2021 to May 15, 2021.  

Why?  The average price of a sold punk on Thursday was 14.37ETH down from 23.61ETH the prior day (down 40%).

The mini-bear market may be turning. There was a small uptick today (May 16, 2021) of almost 5%.

Some observations:

1) You could see the drop coming about a day earlier when asks for floor punks dropped across the board.   Our model does not yet incorporate bids/asks, but probably should in a future iteration.

2) The initial community view was that this crash was in 'floor punks'.   The model discounted all punks (ex the non-humans, see below). Interestingly, we then saw some mid-tier punks also trade at low prices so, perhaps, in this case the model was partially/substantially right?

3) But is it reasonable to be believe all CryptoPunks are worth 40% less than 3 days ago?  Well, in 'TradArt', this is not hugely plausible, but "This is Crypto".  

10-20% daily moves in value happen even in blue chip tokens like BTC and ETH that have 100x to 500x larger market caps than CryptoPunks.   Given this, we have no reason to believe that CryptoPunks couldn't exhibit similar volatility.  It is "normal" in this realm.

3) You may notice that our alien, ape and zombie prices have not repriced.

This is because we were in the middle of testing a modeling approach that involves model CPs by type and then aggregating.  

We left it like this for now.   I think it is probably more accurate this way (I doubt many/any alien, ape, zombie holders are selling into the human punk crash), but of course this is a qualitative decision.

4) We will keep repeating this - the model is in beta and we are trying many different approaches, including aggressive experiments.   Our goal in the next few weeks is to automate the experiments and the results of the experiments so we can report out to the community.   Right now, it is very manually intensive to do so which is why we need to automate.

5) Lastly, we found and solved a bug on the line chart of the last 30 days of prices.   The prior chart had a suspiciously smooth line up in ETH terms; we found the error in the line chart data and have corrected it.