To put the Curve pool back into balance 50,000 $ETH was sold, and another 20,000 $ETH was sent to Binance.
/2
What happened to $UST and $LUNA?
A long thread that looks at what happened in the last 4 days
/1
To put the Curve pool back into balance 50,000 $ETH was sold, and another 20,000 $ETH was sent to Binance.
/2
This leads to a $2B withdrawal of $UST on Anchor.
Now the peg is moving between 0.987-0.995.
However, the first defense is successful. But the peg never fully recovered.
/3
1) The attackers borrowed 100,000 $BTC from Gemini (which they short)
4/
3) LFG (Luna Foundation Guard) is buying $BTC in March and April as a defense mechanism to protect the peg
4) The announcement of the switch from the 3-pool to the 4-pool
5/
So LFG is removing $150M $UST from the 3-pool on Curve in preparations for the new 4-pool, then the attacker uses $350M of the $UST they've bought OTC to drain the Curve pool.
Next, there's now no liquidity on Curve.
/6
Anchor deposits now really start to tank.
The rumors are spreading quickly.
I remember the Anchor deposits were $14B at the top, but you could literally watch the deposits go down by $10M per minute. It was full panic.
/7
But the attack is not done.
/8
What do you think happened?
A massive de-peg.
LFG is stepping in and selling $BTC to buy $UST to try to restore the peg.
Now the death spiral is starting to get momentum.
/9
-Attackers dumping $UST
-LFG trying to restore the peg by buying $UST (selling $BTC)
-$BTC price goes down because of high sell pressure
-Increased panic because $UST de-pegs even more
/10
-This creates a new sell pressure on $UST which makes it de-peg more
- $LUNA price collapsing because of fear of how $LUNA / $UST works
/11
-Traders start to short $LUNA which further tanks the $LUNA-price, and also shorting $UST (possible on FTX)
-Then centralized exchanges started to ban $UST withdrawals, and now the panic is in full mode
/12
But the story doesn't end here.
Through all of this, Do Kwon has been hinting at a rescue plan.
/13
https://t.co/ltxICOn8RJ
/14
Close to announcing a recovery plan for $UST. Hang tight.
— Do Kwon \U0001f315 (@stablekwon) May 10, 2022
https://t.co/VBs71Rqu8s
/15
1/ Dear Terra Community:
— Do Kwon \U0001f315 (@stablekwon) May 11, 2022
This chart I made on Tradingview shows the peg activity between the two tweets on May 10th and May 11th.
/16
Everybody was waiting for the rescue plan, and people were talking about who really wanted to do this bailout.
At this time, the $LUNA price has been hovering around $30-35 for almost 24 hours (May 10th).
/17
The @lawmaster posted this tweet
https://t.co/yBNHTUUu4r
/18
The details I have heard but might not be final:
— Larry Cermak (@lawmaster) May 10, 2022
- Jump, Celsius and Jane St. committed already, Alameda not yet
- targeting $1B - $1.5B raise
- 50% discount to LUNA spot with 1yr lock + monthly linear vest over 1yr
- total of ~$700M in commitments as of morning https://t.co/BtqJrF0eqM
Why did it drop?
IMO the reason for this was that the plan didn't look solid.
Who would like to buy...
/19
Time goes by, and people are still waiting for the plan. It's now 8 hours since Do Kwon tweeted about the rescue.
/20
There was really no rescue plan.
At least not for $LUNAtics because the plan now is to slow-bleed out the $ LUNA token in order to save the peg.
/21
$The $LUNA price dumps from $5 to $0.80 within the next 5 hours.
/22
Check this by @0xHamz:
https://t.co/2FwYcERqmd
In other words, this could take some time.
/23
How many days will $LUNA hyperinflation last for?
— 0xHamZ (@0xHamz) May 11, 2022
Weeks to months
Assuming an orderbook flow limit of $40mm per hour it will take 10 days to clear $10bn of UST at $2.50
At $0.50 it'll take 50 days
Path dependency / reflexivity pic.twitter.com/a3Xz5bnZ0L
Absolutely not. The priority is to save the peg. The $LUNA price will be diluted. If you buy it, that's super high risk.
Seeing some people posting that the staking rewards for $LUNA are now 22% is awful. It's true.
But what's 22% APR on a token
/24
Zero.
(thread is not finished, more coming below in a couple of minutes)
/25
That remains unanswered. But there's a lot of good infrastructures that were built in almost 100 dApps
This is what was attractive with Terra in the first place
The smooth system with a pretty UX
And there are tons of gigabrains in the Terra system
/26
The saddest part in all of this was how many people had their $LUNA-tokens staked and due to a 24-day withdrawal period wouldn't unstake it.
/27
Do you sell your $UST into an asset on-chain that is dropping heavily ( $LUNA) and lose money?
Or do you send it to Binance, Kraken, or Coinbase where it's heavily depegged?
/28
And because of this, I guess a lot of people rather chose to do nothing in hope that everything would be all right.
/29
If the 4-pool had been live this would have been much harder.
Because clearing the 4-pool alone would require $3B instead of $350M.
The timing was not a coincidence.
/30
At this point, we don't know. Massive rumors about Citadel and Blackrock being behind this was FUD.
I think the biggest lesson we can make from the last 72 hours is that no coin is too big to fail and to never have your full net worth in one ecosystem.
/31
/32
https://t.co/sSuIHzTKRl
/33
How to make a >800 million dollars in crypto attacking the once 3rd largest stablecoin, Soros style:
— Onchain Wizard (@OnChainWizard) May 10, 2022
Everyone is talking about the $UST attack right now, including Janet Yellen. But no one is talking about how much money the attacker made (or how brilliant it was). Lets dig in\U0001f9f5 pic.twitter.com/nGVfqjpVJb
https://t.co/KIg5H0b0h8
/34
An $18 billion stablecoin is losing its dollar peg with all the magical chaos of algorithmic stables, with a dash of Bitcoin systemic risk drama.
— jonwu.eth (@jonwu_) May 9, 2022
Here's everything you need to know.
The $UST Depeg Thread:
\U0001f447
If you enjoy this kind of deep dive, make sure you subscribe to the newsletter below.
It's 100% free, and I only send out one issue per week.
↓
https://t.co/EJMf3tBEkj
/35
If you're interested in learning more about crypto/DeFi, then follow @route2fi
Would your followers enjoy this post? Make sure to retweet the tweet below with them
↓
https://t.co/8G6E9Wx7Hk
/36
What happened to $UST and $LUNA?
— Route 2 FI (@Route2FI) May 12, 2022
A long thread that looks at what happened in the last 4 days
/1
Gemini has given a statement that they didn't.
Thanks for correcting me.
Source:
https://t.co/aCcVEkh21y
/37
We are aware of a recent story that suggested Gemini made a 100K BTC loan to large institutional counter-parties that reportedly resulted in a selloff in $LUNA. Gemini made no such loan.
— Gemini (@Gemini) May 11, 2022
More from All
Margatha Natarajar murthi - Uthirakosamangai temple near Ramanathapuram,TN
#ArudraDarisanam
Unique Natarajar made of emerlad is abt 6 feet tall.
It is always covered with sandal paste.Only on Thriuvadhirai Star in month Margazhi-Nataraja can be worshipped without sandal paste.
After removing the sandal paste,day long rituals & various abhishekam will be https://t.co/e1Ye8DrNWb day Maragatha Nataraja sannandhi will be closed after anointing the murthi with fresh sandal paste.Maragatha Natarajar is covered with sandal paste throughout the year
as Emerald has scientific property of its molecules getting disturbed when exposed to light/water/sound.This is an ancient Shiva temple considered to be 3000 years old -believed to be where Bhagwan Shiva gave Veda gyaana to Parvati Devi.This temple has some stunning sculptures.
#ArudraDarisanam
Unique Natarajar made of emerlad is abt 6 feet tall.
It is always covered with sandal paste.Only on Thriuvadhirai Star in month Margazhi-Nataraja can be worshipped without sandal paste.
After removing the sandal paste,day long rituals & various abhishekam will be https://t.co/e1Ye8DrNWb day Maragatha Nataraja sannandhi will be closed after anointing the murthi with fresh sandal paste.Maragatha Natarajar is covered with sandal paste throughout the year
as Emerald has scientific property of its molecules getting disturbed when exposed to light/water/sound.This is an ancient Shiva temple considered to be 3000 years old -believed to be where Bhagwan Shiva gave Veda gyaana to Parvati Devi.This temple has some stunning sculptures.
How can we use language supervision to learn better visual representations for robotics?
Introducing Voltron: Language-Driven Representation Learning for Robotics!
Paper: https://t.co/gIsRPtSjKz
Models: https://t.co/NOB3cpATYG
Evaluation: https://t.co/aOzQu95J8z
🧵👇(1 / 12)
Videos of humans performing everyday tasks (Something-Something-v2, Ego4D) offer a rich and diverse resource for learning representations for robotic manipulation.
Yet, an underused part of these datasets are the rich, natural language annotations accompanying each video. (2/12)
The Voltron framework offers a simple way to use language supervision to shape representation learning, building off of prior work in representations for robotics like MVP (https://t.co/Pb0mk9hb4i) and R3M (https://t.co/o2Fkc3fP0e).
The secret is *balance* (3/12)
Starting with a masked autoencoder over frames from these video clips, make a choice:
1) Condition on language and improve our ability to reconstruct the scene.
2) Generate language given the visual representation and improve our ability to describe what's happening. (4/12)
By trading off *conditioning* and *generation* we show that we can learn 1) better representations than prior methods, and 2) explicitly shape the balance of low and high-level features captured.
Why is the ability to shape this balance important? (5/12)
Introducing Voltron: Language-Driven Representation Learning for Robotics!
Paper: https://t.co/gIsRPtSjKz
Models: https://t.co/NOB3cpATYG
Evaluation: https://t.co/aOzQu95J8z
🧵👇(1 / 12)
Videos of humans performing everyday tasks (Something-Something-v2, Ego4D) offer a rich and diverse resource for learning representations for robotic manipulation.
Yet, an underused part of these datasets are the rich, natural language annotations accompanying each video. (2/12)
The Voltron framework offers a simple way to use language supervision to shape representation learning, building off of prior work in representations for robotics like MVP (https://t.co/Pb0mk9hb4i) and R3M (https://t.co/o2Fkc3fP0e).
The secret is *balance* (3/12)
Starting with a masked autoencoder over frames from these video clips, make a choice:
1) Condition on language and improve our ability to reconstruct the scene.
2) Generate language given the visual representation and improve our ability to describe what's happening. (4/12)
By trading off *conditioning* and *generation* we show that we can learn 1) better representations than prior methods, and 2) explicitly shape the balance of low and high-level features captured.
Why is the ability to shape this balance important? (5/12)
