Archive for August, 2008
Credit Delta
I’m not an expert in the greeks. I have studied them for the CFA, but that only means that I’m ready to take a test and not necessarily explain or apply anything. The CFA does go far to make sure that you understand the concepts in application and not just in theory, but still doing one thing and studying about it are 2 different things.
Again, back to the idiots on the flow desk. Typically, when a trading desk is estimating their cds pnl, they use what’s called Credit Delta to calculated it. Credit delta is the similar to what CFA calls DV01, which is the present value of the yield changing by one basis point or PVBP. CDS technically doesn’t have an interest rate component like bonds do so its better referred to as CS01, which is the present value of the credit spreads changing by one basis point. If you did your homework on the fixed income portion of the CFA, you know that bond yield has several components: liquidity spread, rates (discount), credit spread. On CDS, the credit spread enables you to trade in only the credit component of the bond yield. Also realize that for bonds all of these components change, affecting the bond price. So if your yield is 7%, you could say that you have 2% of that representing rates, 1% representing liquidity, and 4% representing credit spread. The CDS on that bond with similar maturity should trade at 400bps to match the credit spread. In practice, the liquidity spread is somewhat ignored for pricing CDS and the bond z-spread is used as a proxy, usually. When trading however, liquidity is embedded into the price so large lots can have different prices than smaller lots. For all intents and purposes, we’ll ignore the liquidity spread. So if you know how credit spreads changed on a given curve and you also know the CS01, then you can calculate the pnl by multiplying the two. This is how the desk estimates pnl for CDS.
For example, if you had a 30bps widening on the curve, which could be from 150bps to 180bps on the credit curve, and a CS01 of $1000, then you just lost $30,000. This is simplified. In actuality, the CS01 of any trade is broken down into buckets that correspond to the maturity points on the curve and then summed to get the overall DV01. See my little pic below:
This trade is a 10MM buy of protection with a spread of 295bps and is showing you the pnl of a 10bps change in credit spreads. Generally speaking, you can estimate the CS01 of a CDS trade by dividing the notional of the trade by 10,000 and multiplying that by the duration, which is usually always around 4 for a 5Y trade (magnitude of the spreads can have an effect but its small so we’ll ignore it for simplicity). The screenshot you see above shows more refined numbers that can be generated from a program that values and prices derivatives like FinCad. There are some residual numbers in the 4Y and lower buckets from using this program, but it doesn’t mean much materially when estimating pnl. You have to remember that until you have unwound the trade, you are estimated your current unrealized pnl. In reality, the LTD pnl, which is the same thing as the NPV, is an approximation of the cashflow that you should expect when unwinding the trade. The number you estimate is rarely exactly what the counterparty is going to pay you but its usually within about 1-2%. These products are traded over-the-counter and everyone is estimating NPV based on their own default models and probabilities so everyone is going to come up with a slightly different number. Bloomberg terminals have a screen called CDSW that is usually used as a proxy and generally accepted as the market standard but there are always slight differences. My point is to not sweat over the details and that you may get slightly different numbers when doing your own calculations, but they should be close to what I’m seeing and probably never exact.
Getting back to how the desk estimates pnl, you can now see how straightforward it is to figure out what you would lose/gain in your CDS position based on the CS01 and changes in the spreads. One thing that the guys in the flow desk don’t account for in this estimate is how changes in the discount rates will affect the NPV, or put another way how changes in rates can affect your pnl. As you should know by now, CDS is a purely credit instrument. On the other hand, bonds are credit and rates instruments, which is a fancy way of saying that both changes in credit and rates will affect the value of a bond. If you know anything about calculating NPV based on future cash flows, you know that adjusting the discount rate used in the calculation will change the NPV. Larger yields create smaller NPVs and vice-versa. This is the same mathematical relationship between bond yield and price as price is the same as NPV. So if you have a CDS contract where you are paying or receiving 295bps on 10MM in notional quarterly for 5 years, then you should expect changes in rates to effect your overall NPV. Changes in NPV equal your daily pnl. Our estimate of credit pnl using CS01 and changes in credit spreads doesn’t account for this change. I won’t get into estimating your DV01 for CDS right now because I need to understand better how to estimate it before I start saying things that are just plain wrong. The one thing to know is that the more in-the-money or out-of-the-money you are in a CDS contract, the more that changes in rates will affect your pnl. This is because there is a larger different between the market spread and the trade’s spread and this is the amt that is being discounted to come to a present value. The closer to mark that your trade is, the less that changes in rates will affect your overall pnl.
The reason why I give the flow guys such a hard time is that they miss this point. They estimate their pnl and end up being way off from actuals because they had a big day in rates and are way in-the-money in their position. I actually had a trader tell me that he never had anyone tell him that before about rates and asked me why nobody ever told him. I never went to business school or anything but studying for the CFA, you learn about this whole NPV phenomenon so I tried to hold my breath and not be too much of a smart-ass and just played dumb. Luckily, I never had issues with having to explain credit gamma, which bond folks will know better as convexity. Well, I have to finally get back to work and continue explaining pnl so I can go home and start drinking.
Reminiscing about “Reminiscences”
Reading the news ticker today on Bloomberg, I find that history continues to repeat itself. “Reminiscences of a Stock Operator”, authored by Edwin Lefevre, is a first-person story about a big swinging-dick stock speculator/trader/market-maker from about 1909 to around 1915. He is rediculed by the press and other speculators as being a insiduous bear. Throughout the story, he’s stating his claim about why he selling short and why anyone should sell short. The coined phrase that every successful trader knows is to sell high and buy low. So why is it that the there is such animosity to those that do sell short. The Bloomberg article read, “Short-sellers are the often-despised traders who place bets that stocks will go down.”
The article is really about naked short-selling, which is basically intentionally creating false negative rumors in order to make money. Its manipulation. The book has many anectdotes about this same thing, which is usually done by insiders but also by many other players. The public is always the last to know about anything, right or wrong. That is a fact.
The problem that I have is that these naked short sellers are in it for quick profits, not long-term damage. If you hold a position in something and the price moves down and then back up, you have lost nothing. The problem comes when the public starts trying to trade on this misinformation, not necessarily in the change in price. Granted, if there wasn’t a stir created, nobody would have sold out of their positions as a large discount, but on the flip-side if those long would just stay long and weather the storm, the price will bounce back to where it was and probably higher once the rumors were discredited. I don’t know how the SEC is going to crackdown on the rumormill, but good luck. There’s also not much of a story here in my opinion because this has been going on forever. For those of us that are speculating and trading as for more than investing purposes, this is more of an issue, but nothing new. Its definitely not robbing the “widows and orphans” as the douchebag Peter J Henning stated from Wayne State University Law School stated.