"[The trading loss] plays right into the hands of a whole bunch of pundits...."
There are a lot of moving parts in the Dismal tale of Dimon's demise. The starting point is that Bruno Iksil in the JPMorgan CIO Office, under the premise of hedging the bank's credit portfolio's tail risk had placed various tranche trades (levered credit positions with various risk profiles) in the only liquid tranche market that still exists - CDX Series 9 (an 'orrible portfolio of credits with an initial maturity at the end of 2012). These positions were low cost (steepeners or equity-mezz) but needed a certain amount of day to day care and maintenance (adjusting hedges and so on). As the market rallied, the positions required increasing amounts of protection be sold to maintain hedges (akin to buying into a rally more and more as it rises). His large size in the market left a mark however that hedge funds tried to fix - that was his index trading was making the index extremely rich (expensive) relative to intrinsics (fair-value).
This is the 10Y IG9 credit index (dark blue) and its fair-value (light blue) and the difference or skew (orange). What is clear is that the index remained massively rich to its fair-value through this period (red oval) and it was not until the last two months or so that the skew (red arrow) began to compress as perhaps Iksil got the nod and more and more people realized the arb...(or understood from where the technical pressure was coming in the index rallying)...
Hedge funds began to try to arb this position and got frustrated at the lack of convergence - and this is how we initially got to hear about Bruno Iksil - the London Whale - since those funds suggested someone was 'cornering' the index market in credit.
Critically - this is akin to looking at the 500 names in the S&P 500 - weighting them and seeing the S&P 500 index should trade at 1200 but it is trading at 1400 so you sell the index 'knowing' that the index is mispriced - (this never occurs in stocks since they are instantly and everywhere arbed between the index and its components - but can occur in credit because of illiquidity or in this case flow - what we call 'technicals').
This was very evident when one looks at the net notional being soaked up by the Whale and this 'hedge' position had clearly grown extremely large as it became a momentum trade not a hedge (at which time we suspect Iksil started to lose control). In early April, as news of this broke across the market, the credit and equity markets were beginning to quiver again at European contagion and US macro data and as a proxy for the volatility JPM must have been feeling we can see very significant (2-3 sigma) swings in the credit index they held. This would more than likely have triggered a risk manager to come along and look over the trader's shoulder - suggesting humbly that he exit/hedge/don't panic.
This is IG9 10Y spreads (upper pane) and their rate of change (lower pane) - (h/t @swaptions for idea) and as is clear the 3-sigma multiple day move likely scared a few risk managers (and Iksil) into fessing up...
Evidence from the HY market suggests that the trader used more liquid on-the-run indices to hedge as the spread of the HY18 credit index blew notably wider relative to intrinsics and net notionals dropped modestly. The market calmed down a little and it appeared from net notionals and the index skews that he tried again last week to unwind some more of the huge position that had clearly tripped various risk limits and VaR controls. This is where we find ourselves now - the net notionals remain huge (and implicitly on JPM's shooulders), his lack of selling has left the credit index maybe 20bps rich to where it might trade given its rough correlation with the S&P 500 and this would imply at least $3bn of losses already in addition at fair-value.
As is evident, IG9 credit index and the S&P 500 have moved in a very correlated manner - and IG9 net notionals (the amount outstanding in IG9 CDS) has risen alongside these moves as JPM built a bigger and bigger longer and longer credit position. The red vertical arrow shows the current dislocation if one assumes the cessation of Iksil's unwind efforts stalled IG9's selloff - which is the $3bn loss that remains to be seen and the black dotted line is an indication of the kind of notional unwind that would occur - which with a market moving as it is - would be highly disjointing.
Of course, the situation is far worse because 1) any efforts to unwind such a huge position will lead to the market yawning wide and swallowing him in illiquid bid-ask spreads; and 2) the rest of the world knows their position - so why would the hedge funds not push their position. Perhaps this explains why JPMorgan's CDS has remained relatively wide while its exuberant stock price shot up on stress-test ebullience - only to plummet back to CDS reality this evening. Critically, JPM will need to use whatever method they can to hedge this now over-hedged and over-long position - which likely means credit instruments such as JNK, HYG, HY18, and IG18 will all get their share of strange attraction as the trader mispriced not just the basis risk (the volatility between the hedge and its underlying) but the attraction of running with a trend when you have a bottomless pit of money to cover it - until now.
It is already evident in the on-the-run liquid indices - HY18 for instance has exploded wider twice now - in line with the net notional reduction and hedging moves from JPM's IG9 position...
This chart somewhat relates to the IG9 skew chart above in that it represents how far above 'fair' the spread of the index trades relative to the underlying names - the spikes show that there was huge technical demand for the index protection relative to the underlying risk of the portfolio.
and perhaps there was already concern in the market with regard JPM's counterparty risk or exposure from hedgies' trades as CDS has been far less exuberant than stocks...
Of course noone knows for sure what exact positions Iksil had on - though it is clear what hedging he needed to do to manage his hedges. As Peter Tchir ( @TFMkts ) noted this evening - perhaps this mark-to-model irregularity is what the Fed discovered and gathered all the banks last week to ascertain just who has what exposure to whom? As we tweeted earlier, perhaps Iksil just got carried away - and please understand that while CDS do indeed provide leverage, so do many other financial instruments - it is not the instrument that caused this - it is the trader as "you don't hedge risk when you bet on momentum continuing you idiot!"
Addendum - VaR is almost entirely useless as a risk statistic in regard to the kind of highly non-linear positions that we are talking about here and so the doubling of JPM's VaR suggests the tail-risk (or conditional VaR) is considerably larger.