Guys, With recent break down in the Equity/Mezz ..Tranches in CDX/ITRAXX., does that sound the deathknell of Base Correlation and market moves on to Compound Correlation. Can any prop guy here, give us the real deal on what exactly happened

i am no correlation trader but basically-lots of people did : sold equity tranche protection with mezzanine hedge -position loosing money twice because equity widened and mezz tightened.what happened is that eq tranche investors have single-name ( idiosyncratic risk) , and since the autos surge in spreads and then GM/Ford downgrade, they stop lossed their positions, demand to buy back eq tranche is a lot higher.on top of that, you have the " obvious" capital structure trade buy GM bonds/ Sell Stocks, which backfired as well.

I second chesscorner on that. Many hedge funds put on the equity - mezz position (lots of carry and long correlation) and now they are all running for the exit at the same time.

Sorry fro my ignorance, but why should this imply the death of base correlation? Because the method breaks down for such a steep curve ?

I agree with Scholar, I don't understand either. Admittedly the market was scary, but most movements seemed explicable with hindsight. What was the reason for 3-6 correlation decreasing? Was this just a decrease in demand from 3-6 investors?Also, I don't see why the events of this week mean that the quoting method (base correlation) must change. I thought that people had been able to quote successfully for most of this week in the base correlations framework. I thought that the difficulty was in predicting the correlation skew dynamics in a stressed market - something that a trader should have to do, and not a model or quoting method.

my two cents - too many correlated positions across different asset classes (credit correlation, cbs, etc) by the same players (hedgies and prop desks). low vol of the last year has demanded a large increase in leverage to generate the same level of absolute returns expected by investors. this meant that when there was a drop in realized spread correlation - from the GM, visteon, ford and delphi blow out - caused the typical rush for the door. the fact correlation market is all the same way round with limited participants just highlights this risk.also just wanted to point out the this is implied correlation - to ask why its gone done - its just like implied vol in so much as if people buy lots of options then iv goes up - if people buy lots of 03 ig4 protection i.c. goes down - that simple.think there is room for some more serious pain out there... across all strategies that multi strat hedgies have on.

Jaxx has the right idea. I run a HF and have traded tranches, but cut everything when the GM profit warning came out. As a practitioner, I have given invited talks at a number of credit conferences, which means that I have heard a lot of opinions and arguments!There is nothing wrong with using base correlation, if you understand what it is telling you. The entire method of transforming CDS spreads plus a single correlation figure into an equity tranche price (like 0-3, 0-6, 0-9) is as acceptable as it ever was. The fact is that it does not constitute any kind of "real model". There is no modelling of spread dynamics (they are assumed to be constant), nor does spread-spread correlation enter the calculation anywhere. The entire framework is like an actuarial model of the loss distribution, so it only ever was useful as a way of converting prices like "5Y iTraxx equity is 43 bid" into a correlation figure. What is needed is a real model that produces parameters that can do more than parametrize. The parameters extracted need to be testable physically (for example you can measure historical volatility but not "the correlation of default times"), and they need to be able to predict things, such as where other maturity tranches ought to trade, for a start. I am sure one of the clever quants out there will cook something up. Hopefully the work is done outside an investment bank or HF, or we won't hear anything about this "real model" until it is too late!The recent (and continuing) panic is due, as Jaxx said, to flow being bigger than the market wants, in particular because the equity tranche is a rusty, leaky old 55 gallon drum of financial toxic waste excreted by investment banks, and nobody really wants it, though they will risk holding it for enough positive carry. There are (to my knowledge) no real money investors who would describe their investment criteria as anything like the risk profile of a CDO equity tranche. The thing is, unlike nuclear waste, you can't really wrap the equity tranche in long lasting cement and dump it into the Mariana trench. Somebody has to hold it. That is because of the way it is created; real money investors (mistakenly in my opinion) buy BBB rated 3-6 tranches from investment banks, who initially delta hedge with the index, or something like it. Then they have got a nice big warehouse of smelly equity tranche and need to ship it out. They manage to ship it around the market to HFs and other investment banks, or recently cleverly sell a guaranteed product using CPPI (but reportedly not hedge the gap risk - duh!) to real money accounts. But the only way to make it all go away is to get the real money investor to unwind his position, and collapse all the pieces of the CDO back into nothingness from which it was created.At least the good news is that the half-life of equity tranches is not measured in thousands of years...

Erstwhile / Jaxx, what you say makes sense. It is clear why the 0-3 correlation fell.The key for many people will be to understand why the 3-6 hedge didn't work. I.e. why did the implied spread of 3-6 stay the same / go down. Put another way, why did the implied correlation of the 3-6 fall? What was going on with supply / demand in the 3-6 market to cause this move? Were there more protection sellers than buyers? If this was the case, can someone explain why - I would have expected most people to have bought 3-6 protection last week, not sold.Thanks

Herbie: hedges only work in liquid markets and this is what caused the problem. The reason the hedge "didn't work", is that the market was not liquid enough to absorb the unwind of an existing large 0-3/3-6 trade. One can come up with reasons like "the recent increase in idiosyncratic risk due to the downgrades of GM and Ford credit caused a repricing of correlation risk in the market". This is rubbish. The "correlation move" did not occur when GM and Ford were downgraded. It occurred well afterwards, after someone decided to hoist the white flag on his position!For example, imagine the following situation in equity derivatives. Say IBM is trading at $75 and the Jan 2007 $75 call is trading at $10. Imagine you had on a position in which you had sold the call option short and bought the delta hedge in stock. Suddenly, someone buys 600,000 of the calls you shorted up to $15 and sells the delta hedge (about 2 billion dollars of stock) down to $72. The call went up and the stock went down, regardless of what the binomial model told you the hedge should do! Why didn't the hedge work? Because the flow in the market overwhelmed the liquidity. Does this mean you should double up your losing position? There is game theoretic information missing, so you can't make a very good decision. You need to call around and find out what is going on. In this case the crucial knowledge missing to make the decision is whether or not hedgers are on both sides of the transaction. If market participants delta hedge both sides, there may be no lasting overall market impact or distortion, but if real money (non-delta hedger) has bought the option and a dealer is delta hedging himself, then someone somewhere is going to sell low and buy high, increasing market vol. In the case of the credit derivatives, almost everyone who touches an equity tranche is a hedger. In this situation the game theory is like a number of companies storing nuclear waste. If there are loads of people storing a moderate amount, then each perceives there to be a moderate risk and there is no panic. If one company dumps a huge amount of it around the market in a big hurry, perception is that this is not a good thing to hold. This would make the waste storage price go up, which is what has happened here. People in the future who look only at historical data and miss out all the game theory arguments such as the above will simply conclude that you need to hold a lot more capital against these positions than was originally thought (and maybe they are right!) The 3-6 tranche only moved because it was being used as a hedge for the 0-3 tranche.

Dr. Earstwhile,Another master explanation...good work ..chief....please keep it up...

All true, but back to the topic of the ""Base Correlation is Dead". I always thought of the Base Corr framework as more of a static tool to benchmark the custom bespoke tranches to the liquid quoted market, i.e. it's a "an equivalent price" here and now, but it doesn't address the dynamics. As already pointed out here, there is no spread dynamics model, no observable parameters to estimate in the Base Corr framework. Maybe the correlated Default Intensities (Duffie) will emerge as a good alternative, but until then it seems that using Base Corr based hedging strategies accross the capital structure is just a simple technical way of predicting the supply/demand dynamics - maybe not really a good way in a thin market (with real money players being on one side of only some trades) and is subject to shocks like we recently observed.

We have seen the correlation market move all over the place recently. The link between upfront + tranche premium and base correlation is needed by the market in order to communicate and calculate hedges.Is there an upfront + premium regime at which base correlation will break down? For example, consider the 0-3 equity tranche with 5% running premium. Are there levels of the market-quoted upfront for which the market standard models stop working?(Obviosuly the market can quote any upfront it likes, and is not constrained to only quote upfront levels that are possible outputs from the standard gaussian copula?)Or, is it the case that a base correlation exists for any possible level of upfront / premium?Many thanks

well, i guess one extreme limiting case is that in which the upfront on the tranche (quoted with no running fee) is greater than the sum of all the equivalent upfront values of the 125 underlyings CDSs. then you could buy all the CDSs in 100% delta size, short the equity tranche and have net riskless cash left over.

I think the theoretical floor for correlation on a gaussian coupling for n assets is -1/(n-1) where n is the number of assets, so:- for 2 asset, the floor is -100%- for 125 asset, the floor is -0.8%- for infinite assets, 0%A floor in the sense that an option (say, a call spread...) would lose value if vol went up.

That would correspond to the highest protection price reachable by the standard model, right?What arbitrage trade would you put on to extract value if the price of equity protection went above the level predicted by rho= -1/(n-1) ?Also, what happens if you have nonhomogeneous correlation in your matrix? Would that alllow you to reach higher protection prices (for same spread levels of course)?