About the benefits of harnessing DeFI data in inefficient markets

Transcript

hello everyone i'm anastasia i'm working at kaiko kaiko is a crypto it's a leading crypto market data provider so we provide data on centralized exchanges and decentralized exchanges and today i wanted to present you a bit what i'm doing at geico so at kaiko i'm a product manager on d5 market data and um so it's been actually three years that i'm working there and through this experience i've discovered the importance of d5 data and more specifically taxes data to understand market events and potentially to analyze and identify signals of potentially

bad market events that happened uh in the past and that actually could happen in the future um so my job consists of studying and analyzing and reverse engineering many d5 protocols in order to build user-friendly data products and so the the data that we're going to look at today it's dex's data and i'll go through a data-driven analysis of usc's d-peg and so the terra collapse to actually show you how

by harnessing defy data and specifically what we have been building for a year now at kyco we we we actually can find many interesting things uh so i'm sure all of you are aware of what happened uh recently on terra and by reading a lot about this in the in the past weeks and actually months from now i successfully i mean i tried to remake the story from a to z on what happened on the markets and how like how the markets had a key role on

ust jpeg so everything began uh with terraform labs withdrawing 150 million dollars of ust from the key the main curve pool which was a pool made of ust and three crv three crv being a token representing another pool the three pool and inside of three cervi we have usdt usdc and die after this we we as expected the price of ust with respect to three crv went above one since usc is supposed to be equal to one dollar but

a very smart trader bridged 85 million usd from terra to ethereum and then swapped those ust to take back usdc which created a snowball effect and a massive sell-off with other traders doing the exact same like selling many usd on the same pool and which caused the [Music] the increase of usd reserves in this specific pool and the beginning of the drop of a usc's price all of those events happened between the 7th and the 8th of may

on the 9th and the 10th we observed a huge decrease in liquidity on taxes what i've called here the liquidity crisis on taxes and so since there were no more liquidity there was no more liquidity on taxes what happened is that other traders that actually heard about what was happening on ust and and heard and saw the the small d-peg and were trading on centralized exchanges actually sold uh ust on centralized exchanges and so the phenomenon migrated from decentralized to centralized exchanges and so kind of the same thing happened but way later uh on

liquidity uh once we had a liquidity crisis on centralized exchanges uh the the last solution for for holders of ust was to um to switch their usc and get back luna benefiting from the redeem mechanism so what they could do directly on terra was to redeem one dollar of luna by burning one ust whatever its price but this was only if luna's market cap was above usd's market cap however uh the only the first commerce the

the first people who actually did that could benefit from this because a very very fast lunar supply increased and since there are more and more luna in the markets one luna was worth less and so luna's price decreased which in the end made luna's market cap being lower than usd's market cap and so game over in the end it was the terror collapse so this was a story and uh by looking at the data that we have built at geico especially on the texas part since we in my team we have we have built this specific those

specific data sets um i wanted to see if there is really if there was something in the data very early before the 18th of may where the price really really dropped that we could have used to um to really understand and anticipate what was going to happen and actually there was the first thing that i that i i discovered uh was uh by looking at the very generic data on the liquidity pulse level so on the curve liquidity level so i've checked the the token reserves of the ust3 crv pool

and uh so on the one side and orange we have ust and on blue we have the other stable coins and normally for the price to be one they have to be it has to be a balanced pool however we can see that just on may 7 may 8 the reserves begin to be imbalanced across those two tokens which means that something is happening on the price which means that ust is worth less and less less than one dollar and this well i'll show you after that it was actually because of the massive sell-off that was happening

but this continued uh till the point where usc was representing in this specific pool of curve was representing more than 90 percent of the reserves um one last thing about this chart reserves are extremely important on taxes when we analyze the markets because dexes have a different microstructure of than centralized exchanges this microstructure is called imm's automatic automated market makers and on imms token reserves have a direct impact on the price and after looking at very aggregated

level data on the on the pulse level i have decided to look at more granule data so to check swaps so swap by swap what happened on this pool and on the on other dex pools on other decks like unisport p3 but also liquidity events that is the activity of liquidity providers and that's where i actually noticed and noticed that this thing the the increase in um in in the reserves of usc inside the bull on curve was actually due to a massive sell-off of usd with respect to usdc

and in total i've i've computed the sum uh on may 9 which was the all-time high in volume uh on cell volume of usd uh we observed more than 20 20 million ust that have been sold on only on this on this curve pool and only with respect to us usdc which means that they're actually more than those sales same thing for liquidity events when we look at more granular data we can actually determine also what is happening on the reserves if we go back to the first chart we see

that there is a huge decrease in um in liquidity uh in this pool on may 14th and this totally corresponds to may 13th actually this totally corresponds to a burn event a burn is a liquidity provider withdrawing liquidity from a curve pool and we see that the the amount actually totally corresponds to what due to the decrease in liquidity in the pool um and well the next question i asked myself was okay if there is if there is such uh such an event happening on a curve and another taxes since we saw that the same

thing was happening on unisrp3 this actually has an impact on the price and you all know that price of ust decreased tremendously even on may since may 10 and i asked myself is there uh are there any huge price discrepancies between decentralized exchanges and centralized exchanges uh on the on the price and actually i've noticed that not really so in blue in dark blue we see the average price on of ust with respect to stable coin pairs uh on centralized exchanges and the other the other lines on pink and blue are

curve and unisov3 and actually we see that the price kind of corresponds however by looking at the volumes i have discovered another story i have discovered that there was really a information asymmetry between traders on taxes and traders on centralized exchanges because when we look on the left at um the total volume traded daily on ust pools on curve and uni swap we see that there is a peak on may 9 just before the huge drop of the price however when we look at the same chart but on centralized exchanges and i've chosen here kraken and kucoin we see that this

peak only happens after which means that the information gets afterwards to traders who trade on centralized exchanges and actually when we look at the average trade size on centralized exchanges we see that it's way lower than the one on decentralized exchanges on centralized exchanges the average trade size is way below 100 000 while on decentralized exchanges it's above one million dollars a trade which uh means that uh either centralized exchanges trader most of them are retailers or retail traders or less experienced traders

and that on taxes we have mainly whales and maybe institutionals so as i told you this in the beginning in the storyline everything began on on decentralized exchanges but this had an impact on centralized exchanges and um i really like this chart that have been made by keiko's research team which shows that on may 11th we had the buy side of binance order book on usc usdt that was totally empty and there was also a tweet uh which was pretty fun about that and uh this reminded me of actually a

piece that i've read a play that i've read uh like 10 years ago it's a french play about uh it's actually the story of vladimir and estragon waiting for godot who actually never comes and like we can do a parallel between godot and usc buyers on may 11th on centralized exchanges and vladimir and s regal who are actually the usd sellers on may 11th to actually conclude about this data analysis we had actually days to identify what happened on um on ust uh to identify what happened and what

caused in the markets usd jpeg we had more than four days just by looking at the trades like the second chart here just by looking at the massive self we could have anticipated this and even centralized exchanges traders could have traded faster uh four days before so what we can say is that on chain on on on chain markets latency is not really a thing what really matters is just having access to the data just having the information being shared just uh having um market efficiency across all the markets uh of crypto

uh whether they're centralized or decentralized since decentralized exchanges markets are becoming more and more important when it comes to volume or just when it comes to some tokens since some tokens are only traded on decentralized exchanges and actually i remember uh when i was actually looking at those charts and realized that signals were there in the data and there there was a huge information symmetry uh on those markets i remember like my studies in finance and this theory that i've studied back then um it was like a theory uh

it's a quantitative model which shows the importance of informational symmetry on traditional financial markets and measures the risk that it can represent uh and actually i i i concluded that not having access to decentralized exchanges data which actually ensures having the right information to monitor to analyze those markets is taking and considerable risk uh and so well at caico we have built the the tools the toolkit to avoid such predictable market events and uh avoid taking uh unconsidered risk and

so those tools are tools that are standardized so we provide data which is uh available for centralized exchanges for decentralized exchanges through apis sdks so it's through traditional and reliable tools and infrastructure that we build by ourselves we take when it comes to decentralized exchanges data we take it directly on ethereum uh and on other blockchains um we provide this data real time in order to enable big traders and big investors as institutionals and enterprises to monitor and analyze in real time this data and so yeah to anticipate those market events that can be very very easily

anticipated thank you very much [Applause] any questions all right then enjoy the rest of the day and the amazing side events tonight [Applause] you