Randomness (r_ness) wrote,
Randomness
r_ness

I've been too busy to post about finance in a while...

...but this story is just a bit too good to let pass by. Apparently J.P. Morgan has somehow managed to lose at least a couple of billion dollars through some extremely questionable trades in the last few weeks.

From the FT's Alphaville blog:
In a conference call on Thursday night, CEO Jamie Dimon put mark-to-market losses at $2bn for the unit. Dimon has thus gone from calling the CIO’s trades a “a complete tempest in a teapot” to referring to the losses “plays right into the hands of a whole bunch of pundits out there”.
Other things Dimon said during the conference call:
“…Errors, sloppiness, and bad judgement.”

“Bad strategy, badly executed and poorly monitored”

“It could get worse. This could go on for a little bit.”

“Badly executed, badly monitored. I’m not going to repeat it 800 times”

“I know it was done with the intention to hedge tail risk… it was unbelievably ineffective”
Lisa Pollack of the FT has been covering the story for a few weeks now, ever since it became apparent that something funny was going on with JPMorgan's trades.

Today she appeared on Alphaville's daily Markets Live online chat, and gave this summary in answer to her colleague Bryce Elder's question, "Are we any clearer on what actually happened?"
Well... yes and no
There was never a huge amount of disclosure around the trades
There are, as best I know, three bits of evidence
Only one of which came from JPM
First two are in the market data
1. The index CDX.NA.IG.9 had a huge increase in the net notional amounts outstanding over the last half year
2. The skew blew out
Meaning that the index was way cheaper than the constituents.
From this we knew
That someone, or several someones, are pressuring the market in a particular way
Lots of selling pressure on the index, i.e. selling protection
3. JPM said the CIO had downside hedges on high grade credit
The CDX.NA.IG.9 is high grade credit
(for the most part)
But we just said that there was selling pressure, i.e. selling protection
These things don't compute
UNLESS it was a curve trade.
Where to are buying at one point and selling at another
The selling pressure at the long end of the curve can create the skew technical
And the bigger amount of buying you have to do at the short end of the curve can lead to the notional increase
It's a basic flattener trade
The thing about flatteners though
Is that you have to manage them quite actively
To get the ratio of buying to selling right
It looks like they were managing that...
... in fact they were managing that too much.
Hence the skew issue
The trades were too big for the market
They may have stopped managing the ratio even
Once they realised what they were doing
That the market didn't have enough depth
(And that pundits were having a ball)
(That it didn't look good)
But once you stop getting the ratio right
The trade starts to look a lot more risky
WAY more risky
That may be what we are seeing now
With their losses
They are finally having to admit
That they did this too big
And the market can't take it.
Bryce Elder's summing up for the audience:
I understood a lot of that. Thanks.
To be honest, the intricacies of synthetic flatteners and suchlike give me the yips.
But I do know about gambling.
I know that, when your system stops working, few resist the temptation to double down.
Because, when it starts working again, you cover your losses.
But systems can't do that.

I do like these stories.
They seem to prove that it's a mug's game applying statistical probability to a weighted selection of events.
It's not a faulty VAR model, it's a faulty assumption that you could ever quantify the actions of a crowd.
And a faulty belief that your trajectory of returns is anything other than random.
Tags: money
Subscribe
  • Post a new comment

    Error

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.
  • 12 comments