having sat on crypto dealing desks for many years, I know that offering liquidity to savvy traders is a painful but necessary part of the game
GLP holders learned that same lesson last night!
2/ last night, trader X successfully extracted profits from GMX's AVAX/USD market by opening large positions at 0 slippage, then moving AVAX/USD on other venues in their favor
this created a sinusoidal pattern for over an hour as X switched from long to short 5 times
3/ let's take a look at the first cycle which took place from 01:15:31 to 01:28:11 UTC. X was able to extract roughly $158k in profit by trading clips of $4-5mm at a time
4/ X did this 5 times (with less impact each time), so let's say they extracted ~$500-700k profit. Ofc X was paying spread to market-makers on the other venues they were trading on to move the price of AVAX, so the net collected is less
5/ this isn't an exploit as much as GMX working as designed! X executed large trades in against GLP holders with 0 slippage: at the oracle price without factoring any price impact
in the real world, putting on risk requires you to pay liquidity providers on the opposite side
6/ why use GMX and not FTX perps? because you can't trade at an oracle price on FTX, you pay some slippage as you execute up the orderbook
you'd move price from $17.95 to $20.25 to buy 200k units of AVAX-PERP. So you'd lose on the FTX leg AND on the other venues you're moving
7/ so the real issue is GMX doesn't reflect the true cost of liquidity like other venues do, it offers unlimited liquidity at a mid-market oracle price
how can GMX fix this?
8/ first, GMX can look at orderbook depth on CEXes, reflect the cost of liquidity in the executable price, and pay any spread collected to GLP holders
this removes the incentive to do this type of exploit, though it runs contrary to GMX's marketing as a zero price impact product
9/ second, GMX can try to identify toxic customer flow and cut it off or widen their spreads specifically -- similar to how forex brokers segment customer flow into a-book vs b-book
10/ third, GMX can cap max position size on any particular asset to a small fraction of liquidity available on CEXes so that any extractable profits are small relative to the cost of moving the asset price on CEX
11/ the last point is important: this problem is one that comes up more and more as GMX grows larger and attracts savvy traders. it's a good problem to have! defi liquidity venues are going to have to converge on cefi best practices over time as market sophistication increases
• • •
Missing some Tweet in this thread? You can try to
force a refresh
1/ sorry for the long thread: I’ve had many convos around the SEC approval of IBIT options. my smart colleagues in the industry (h/t @dgt10011) have written great commentary around it...
my own views are more nuanced and tempered, but as always expect the unexpected in crypto:
2/ first things first, we have to remember crypto already has a well-known and extremely liquid (by crypto standards) options venue called Deribit. it trades ~$40 bn notional monthly in BTC options. this compares to CME at ~$3bn
3/ yes, it’s an “offshore” and “crypto-native” venue. yes, it’s where crypto retail traders who have some derivatives savvy go to trade. But a lot of tradfi firms are market-making there too…
1/ the CME futures roll traded at dizzying, near-historic wides during last week’s Nov to Dec roll window, touching 23%+ annualized at times, making it briefly one of the most attractive risk-adjusted trades in crypto
2/ we know tradfi guys / macro tourists are already long crypto ahead of ETF news, they’ve built the position over the last few months and are now paying handsomely to roll it
commitment of traders data showing asset managers increased length by about $1bn since end of Sep
3/ $BITO, the rolling CME futures ETF showed an increase of $560mm in AUM since end of Sep, of which around $230mm came from inflows and the rest from spot appreciation (shares out went from 64mm to 77mm)
1/ today's massive BTC options print on CME demonstrates a larger trend happening in crypto options
2/ at 11:11 AM ET, customer bought 600x of the May 32k / 38k call spread, each contract has a 5 BTC multiplier, so the notional of this trade is $82mm. customer paid $1085 per unit = $3.26mm in premium for a max payout of $18mm at maturity
3/ but if we dig a little deeper, we see another print happened on Deribit at 11:22 AM ET for the same size and this one was a seller of the call spread at ~$940 per unit
1/ BTC has lagged the performance of other crypto assets through the last cycle
most traders intuitively feel that BTC trades “heavy” – why is this happening?
there’s two simple metrics we can look at to validate what we intuitively feel:
2/ first metric is the ETH/BTC ratio, ETH price divided by BTC price, which is 0.0733. this is still near multi-year highs of 0.0880 which we touched prior to the Fed-induced meltdown in risk assets in Dec 2021
3/ the “flippening” when ETH mkt cap = BTC mkt cap occurs at ETH/BTC ratio of 0.0159. sizable positioning in ETH calls reflects mkt consensus of continued ETH outperformance
charts below show ETH put/call ratio is only 0.24, substantially lower than BTC’s at 0.53