It is said that a credit-driven recessions are the worst of the recessions. They happen extremely infrequently, but when they happen they tend to be far worse than the usual recessions. Recessions tend to happen every 3-5 years. Usually they last about a year and wipe out maybe 10-20% of equity gains (for lack of better yard-stick to measure the "severity" of a recession).
Credit recessions tend to impact the sytem much more heavily. They tend not only to give the equity markets a whallop, but they also leave lasting marks on the financial landscape. During a typical recession a few companies go out of business, usually those with operating cash-flow issues, and the rest move on the create record profits a few years later. Credit recessions are a different beast all together though. Major companies get wiped out, banks go under in droves, and the way in which credit is extended as well as managed changes. It is said that the California banking industry is still recovering from the last credit recession, which was probably 1990-1991. Too bad for them because a lot of signs are saying we're hitting another one.
So what's so bad about a credit recession? Unlike a normal recession, they are led by banks. When credit blows up, banks tighten up their purse-strings. That means banks stop lending as freely, and they do so by making their interest rates (credit spreads) prohibitively high. Then companies stop spending because they can't finance new investments with debt any more. The equity markets tighten up because earnings suck. Suddenly the economy stops cold. Companies go under in droves because they can't get adequate funding and aren't selling their goods/services as robustly, which further affects banks and makes them widen credit spreads even more. It's a pretty disasterous cycle. At the end of the day, usually a few dozen banks go under, sometimes insurers and other financials go with them.
It looks like this time around the subprime mortgage hit may have started an unfortunate chain reaction. Losing money on subprime has made banks wary of their balance sheet issues. If they lose money on the subprime mortgage stuff, then suddenly they have to defend the amount of balance sheet they are keeping. They do so by raising interest rates and not lending out as freely as they did before. Thus we start down the cycle that I described above.
We are already seeing the effects on our financial landscape. The behemoth Citigroup has all but died, their stock price back to levels seen in the last recession. They moved from #1 in market cap to #3 in the last year or so (they were passed by Bank of America somewhat before this credit blow-out and were passed by JP Morgan just a few days ago). It looks like SIVs (Structured Investment Vehicles--I'll do a Learn the Lingo on these at some point, I may have mentioned them in another Learn the Lingo post) may "go the way of the dinosaur" as they say. Banks will have to deal with those assets on their balance sheets instead of structuring them and spinning them into their own entities. Monoline insurers are looking pretty precarious right now as the first one (Ambac) just got downgraded from AAA. I'll also try to do a Learn the Lingo on monoline insurers, but the main issue is that they insure municipal bonds. They are absolutely useless if they don't have that AAA rating because why would you want anything other than a AAA rated company insuring a state/municipality? Ambac is the second biggest insurer, and people are already talking about the biggest insurer, MBIA, also being in trouble. Bad times.
So that's where we stand now. Credit recession or not, it looks like our financial landscape has changed once again. It'll be interesting to see what happens to CDOs, subprime mortgages and other financial instruments that were hit so hard this time around. Knowing Wall Street though, they'll always find a good way to get in trouble again. In fact, those same instruments may well find their place back in the markets in due time.
Saturday, January 19, 2008
Saturday, January 12, 2008
Learn the Lingo (Optionality and Greeks)
Derivatives are a booming business these days, and I figured I've yet to talk about options and the ways we describe them. It's been a while since I've done a real "learn the lingo" post.
People talk about the greeks all the time (not referring to their frat/sorority buddies), but it seems precious few outside the derivatives world understand what they're talking about. Actually, precious few IN the derivatives world understand what they're talking about. Try to avoid being one of those people. Do note that I use the term "derivative" both in referring to derivative contracts and in referring to the mathematical derivative. I'm sure you'll figure out when I'm using it in each way, but do ask if my writing is confusing.
Call Option - a contract to enter into the right to buy something at a set price.
Put Option - a contract to enter into the right to sell something at a set price.
Future - a contract, usually on an exchange, allowing one to buy X units of something at the price the contract was transacted.
OTC - Over-the-Counter, refering to transactions done as deals between entities as opposed to exchange traded, thus making the market for such transactions more opaque.
Forward - a contract, usually OTC, to buy X units of something at a set price.
Delta - the sensitivity of a derivative to a small move in the underlying reference material. Sometimes delta is used to refer to the trading of the reference material itself. The reference material can be an underlying security, a commodity, a rate, even some economic variable. Delta is also the first derivative of the price of a contract with respect to the underlying reference material (remember from Calculus that the first derivative is the slope of a curve? Notice how a the slope approximates a small move in the y coordinate due to a small move in the x coordinate?).
Gamma - the second derivative of a contract with respect to the underlying reference. This shows the sensitivity of the price of a contract to large moves in the underlying. It is the curvature of the price sensitivity graph. Do you remember doing Taylor approximations in Calculus? Delta and gamma (the first and second derivative of the price of a contract with respect to it's underlying) are often used to find the price of a contract after a move in the underlying via Taylor approximation. The important concept, however, is that gamma refers to the sensitivity of a contract to large moves in the underlying.
Theta - option decay. Options are worth more when there is a lot of time before they expire. It shouldn't be surprising that the longer you have the option, the more it's worth.
Vega - the sensitivity of a contract to the volatility of the underlying reference. This is often a parameter in pricing models, but conceptually it refers to how much more a contract is worth when the underlying reference material becomes more volatile. Most options are worth more with volatility. In fact, option traders are often referred to as vega or vol traders.
Rho - the sensitivity of a contract to the underlying interest rate assumption. This is important because the interest rate changes the discount factor of value over time.
Black-Scholes - The Black-Scholes model has been the benchmark option pricing model since it's inception. It incorporates all the greeks above. In fact, it pretty much gave birth to the greeks.
People talk about the greeks all the time (not referring to their frat/sorority buddies), but it seems precious few outside the derivatives world understand what they're talking about. Actually, precious few IN the derivatives world understand what they're talking about. Try to avoid being one of those people. Do note that I use the term "derivative" both in referring to derivative contracts and in referring to the mathematical derivative. I'm sure you'll figure out when I'm using it in each way, but do ask if my writing is confusing.
Call Option - a contract to enter into the right to buy something at a set price.
Put Option - a contract to enter into the right to sell something at a set price.
Future - a contract, usually on an exchange, allowing one to buy X units of something at the price the contract was transacted.
OTC - Over-the-Counter, refering to transactions done as deals between entities as opposed to exchange traded, thus making the market for such transactions more opaque.
Forward - a contract, usually OTC, to buy X units of something at a set price.
Delta - the sensitivity of a derivative to a small move in the underlying reference material. Sometimes delta is used to refer to the trading of the reference material itself. The reference material can be an underlying security, a commodity, a rate, even some economic variable. Delta is also the first derivative of the price of a contract with respect to the underlying reference material (remember from Calculus that the first derivative is the slope of a curve? Notice how a the slope approximates a small move in the y coordinate due to a small move in the x coordinate?).
Gamma - the second derivative of a contract with respect to the underlying reference. This shows the sensitivity of the price of a contract to large moves in the underlying. It is the curvature of the price sensitivity graph. Do you remember doing Taylor approximations in Calculus? Delta and gamma (the first and second derivative of the price of a contract with respect to it's underlying) are often used to find the price of a contract after a move in the underlying via Taylor approximation. The important concept, however, is that gamma refers to the sensitivity of a contract to large moves in the underlying.
Theta - option decay. Options are worth more when there is a lot of time before they expire. It shouldn't be surprising that the longer you have the option, the more it's worth.
Vega - the sensitivity of a contract to the volatility of the underlying reference. This is often a parameter in pricing models, but conceptually it refers to how much more a contract is worth when the underlying reference material becomes more volatile. Most options are worth more with volatility. In fact, option traders are often referred to as vega or vol traders.
Rho - the sensitivity of a contract to the underlying interest rate assumption. This is important because the interest rate changes the discount factor of value over time.
Black-Scholes - The Black-Scholes model has been the benchmark option pricing model since it's inception. It incorporates all the greeks above. In fact, it pretty much gave birth to the greeks.
Sunday, January 6, 2008
Day in the life of a Quant
A couple people have asked for something along these lines. I'm going to do two versions of this. "Quants" are an overly generalized term, so there really are multiple types of quants. There are quants that support derivatives desks, there are quants that do statistics for proprietary trading, and there are quants that build apps for use by an institution. I really only know anything about the first two.
Exotic Derivatives Quant:
5:45am - get up, shower
6:30am - be on train
7:00am - get to office
7:10am - deal with first trader complaint. A pricing module for knockout options on fx broke.
7:30am - morning call with traders, quants and sales people
7:45am - work on forward discount curve for dollar denominated assets.
8:20am - first trade of the day comes in. A European bank wants to price a equity index basket option. You start on the code immediately (VBA or C++, depending on the difficulty of the basket/option being priced and also depending on the bank from what I've heard).
9:00am - finished writing and debugging code. Go to your validating quant to check work. (Usually two quants will work on a given model at the same time. If both models get to the same answer within a certian tolerance, then the code is considered good).
9:20am - your code and validation code match up, code is sent to the model validation quants (usually yet another group that compares the model used to several other models). You go check your work for a while and then go back to the project you were working on before the trade came in.
11:30am - lunch at the desk
12:30pm - new trade comes in, a hedge fund wants to do a reverse repo on a stock.
12:35pm - trader whines about his sheet freezing up while pricing the particular stock repo vs the counterparty. you tell him it's because he has too much stuff open and running.
12:40pm - trade is priced and your desk wins the trade.
12:45pm - trader complains that the hedge ratio on the repo can't be right. You wonder why he thinks there's a ratio on the repo. . .
1:30pm - model validation comes back with the 'go-ahead' on the basket option trade (sometimes they come back within the day, for more complicated stuff it can take days).
1:40pm - the counterparty is notified of the trade price, but you know they won't get back to you till the next day.
2:00pm - another equity repo trade
3:00pm - bond markets close. This is important to you because some of the exotics you price can cross fx, fixed income, equity and credit all at once.
4:00pm - equity market closes
4:10pm - make sure all the marks are correct in the derivatives book and double check hedge assumptions
4:45pm - work on correlation matrix for one of the trades in the book
5:30pm - you're an equity guy, time to go home.
note: this is probably a fairly easy day as far as leaving at 5:30, but generally speaking the trade volume tends to be pretty light. There's an occasional "truly" exotic trade that takes a couple days to price. Lots of repo type deals or total return swaps with hedge funds. Technically as the quant you're not reponsible for the risk, but you are responsible for giving good fast prices.
Stats Quant:
6:30am - get up, shower
7:00am - get to the office, log onto relevant machines
7:10am - Check overnight diagnostics on machine trading algorithms being run overnight. Since you didn't get a phone-call waking you up (automated, of course) and telling you of an error, you don't expect to see anything abnormal. You don't need to check overnight pnl because it was e-mailed to you at 6am.
7:30am - open project being worked on. Usually a program in C++ or some research in R/S-plus. Begin programming.
8:20am - fixed income markets open, make sure machine run trading systems are not failing. You really don't have to do much here because you're supposed to get an e-mail and some bells/whistles go off when stuff breaks. Not to mention the electronic systems were running overnight. Some of the trading algorithms only run during liquid hours though.
8:30am - go back to your programming.
9:30am - equity markets open, you do the same drill you did with the fixed income markets.
9:40am - back to programming
10:20am - one of the trading systems throws an error. It froze itself, but you need to manually take it out of it's risk. You delve into the code to see what happened.
10:50am - you found the bug and fixed it. You would test it on the test servers, but the bug happens so infrequently that it wouldn't matter. You did test it internally. You plug it back in and go back to your original project.
11:20am - you go out to grab lunch with a couple colleagues. You decide to stay out for lunch today as opposed to bringing it back to the desk.
12:10pm - you return to your desk and return to coding.
2:50pm - the daily manual trading algorithms produce their results. You and your colleagues discuss the results and start making phone calls to put on the trades given by the daily algorithms. These often involve OTC derivatives, so you actually need to place the phone call to trade.
3:20pm - trades finalized.
3:30pm - daily quant round-table meeting. You and the other quants discuss any issues you've had with coding, any ideas you have for new trading ideas and papers you read overnight. The discussion flows like a classroom. The meeting ends with the new trading ideas that seem worth following being moved forward for a full write-up. Those that have already been written up and fully discussed today move into development phase and are assigned programmers. A paper is handed out for reading overnight.
5:00pm - after the meeting you and a colleague decide to play a quick game of chess.
5:20pm - you lost, again. . . that guy used to be a championship player.
5:30pm - you go back to your workstation to save everything down and make sure things are running.
6:00pm - the fixed income electronic markets open back up (they are 23 hour markets) to check that all the diagnostics are running. You go home as soon as everything checks out.
note: as a prop-side quant your day is somewhat less structured. That means you really need to be driven to do your research and develop new ideas on your own. It's like being a really well paid academic. You don't have others pushing you to do something right as much as you do on a sell-side desk. Typically you need more background for this sort of work too. Usually you see PhDs who did their dissertations in a relevant or exotic fields (exotic because they tend to bring ideas that are unique and methods that are different from the usual tools known to the finance world). You probably will need to be able to program your own stuff (most PhDs in technical fields learn sufficient programming).
Hope this helped. Feel free to e-mail with more questions.
Exotic Derivatives Quant:
5:45am - get up, shower
6:30am - be on train
7:00am - get to office
7:10am - deal with first trader complaint. A pricing module for knockout options on fx broke.
7:30am - morning call with traders, quants and sales people
7:45am - work on forward discount curve for dollar denominated assets.
8:20am - first trade of the day comes in. A European bank wants to price a equity index basket option. You start on the code immediately (VBA or C++, depending on the difficulty of the basket/option being priced and also depending on the bank from what I've heard).
9:00am - finished writing and debugging code. Go to your validating quant to check work. (Usually two quants will work on a given model at the same time. If both models get to the same answer within a certian tolerance, then the code is considered good).
9:20am - your code and validation code match up, code is sent to the model validation quants (usually yet another group that compares the model used to several other models). You go check your work for a while and then go back to the project you were working on before the trade came in.
11:30am - lunch at the desk
12:30pm - new trade comes in, a hedge fund wants to do a reverse repo on a stock.
12:35pm - trader whines about his sheet freezing up while pricing the particular stock repo vs the counterparty. you tell him it's because he has too much stuff open and running.
12:40pm - trade is priced and your desk wins the trade.
12:45pm - trader complains that the hedge ratio on the repo can't be right. You wonder why he thinks there's a ratio on the repo. . .
1:30pm - model validation comes back with the 'go-ahead' on the basket option trade (sometimes they come back within the day, for more complicated stuff it can take days).
1:40pm - the counterparty is notified of the trade price, but you know they won't get back to you till the next day.
2:00pm - another equity repo trade
3:00pm - bond markets close. This is important to you because some of the exotics you price can cross fx, fixed income, equity and credit all at once.
4:00pm - equity market closes
4:10pm - make sure all the marks are correct in the derivatives book and double check hedge assumptions
4:45pm - work on correlation matrix for one of the trades in the book
5:30pm - you're an equity guy, time to go home.
note: this is probably a fairly easy day as far as leaving at 5:30, but generally speaking the trade volume tends to be pretty light. There's an occasional "truly" exotic trade that takes a couple days to price. Lots of repo type deals or total return swaps with hedge funds. Technically as the quant you're not reponsible for the risk, but you are responsible for giving good fast prices.
Stats Quant:
6:30am - get up, shower
7:00am - get to the office, log onto relevant machines
7:10am - Check overnight diagnostics on machine trading algorithms being run overnight. Since you didn't get a phone-call waking you up (automated, of course) and telling you of an error, you don't expect to see anything abnormal. You don't need to check overnight pnl because it was e-mailed to you at 6am.
7:30am - open project being worked on. Usually a program in C++ or some research in R/S-plus. Begin programming.
8:20am - fixed income markets open, make sure machine run trading systems are not failing. You really don't have to do much here because you're supposed to get an e-mail and some bells/whistles go off when stuff breaks. Not to mention the electronic systems were running overnight. Some of the trading algorithms only run during liquid hours though.
8:30am - go back to your programming.
9:30am - equity markets open, you do the same drill you did with the fixed income markets.
9:40am - back to programming
10:20am - one of the trading systems throws an error. It froze itself, but you need to manually take it out of it's risk. You delve into the code to see what happened.
10:50am - you found the bug and fixed it. You would test it on the test servers, but the bug happens so infrequently that it wouldn't matter. You did test it internally. You plug it back in and go back to your original project.
11:20am - you go out to grab lunch with a couple colleagues. You decide to stay out for lunch today as opposed to bringing it back to the desk.
12:10pm - you return to your desk and return to coding.
2:50pm - the daily manual trading algorithms produce their results. You and your colleagues discuss the results and start making phone calls to put on the trades given by the daily algorithms. These often involve OTC derivatives, so you actually need to place the phone call to trade.
3:20pm - trades finalized.
3:30pm - daily quant round-table meeting. You and the other quants discuss any issues you've had with coding, any ideas you have for new trading ideas and papers you read overnight. The discussion flows like a classroom. The meeting ends with the new trading ideas that seem worth following being moved forward for a full write-up. Those that have already been written up and fully discussed today move into development phase and are assigned programmers. A paper is handed out for reading overnight.
5:00pm - after the meeting you and a colleague decide to play a quick game of chess.
5:20pm - you lost, again. . . that guy used to be a championship player.
5:30pm - you go back to your workstation to save everything down and make sure things are running.
6:00pm - the fixed income electronic markets open back up (they are 23 hour markets) to check that all the diagnostics are running. You go home as soon as everything checks out.
note: as a prop-side quant your day is somewhat less structured. That means you really need to be driven to do your research and develop new ideas on your own. It's like being a really well paid academic. You don't have others pushing you to do something right as much as you do on a sell-side desk. Typically you need more background for this sort of work too. Usually you see PhDs who did their dissertations in a relevant or exotic fields (exotic because they tend to bring ideas that are unique and methods that are different from the usual tools known to the finance world). You probably will need to be able to program your own stuff (most PhDs in technical fields learn sufficient programming).
Hope this helped. Feel free to e-mail with more questions.
Thursday, January 3, 2008
TAF? What?
The Fed introduced a new toy the night after last month's Fed meeting. The new toy has an acronym "TAF" as everything in finance has to have a stupid acronym. TAF stands for Term Auction Facility. In the scheme of things, it's the Fed's way of being able to mess with libor.
Up till now the Fed had three major tools: the discount rate, the fed funds rate (via open market operations) and the reserve ratio. See my post about the Fed (another "learn the lingo" posting) if you need details about these. Now the Fed has introduced a new tool that lets it have a more direct impact on libor.
But wait! Isn't the fed funds rate directly tied to libor? Well, sorta, not really. There's certainly a relationship, but generally libor trades at a spread over fed funds. That spread historically sticks to 1% or less. Right now it has blown way out, so the Fed needs something to deal with that spread issue. Enter the TAF.
Here's how it works. Memeber banks (note: member banks need to be commercial banks or savings and loan instutions--specifically, this EXCLUDES the investment banks) can borrow up to 10% of a $20Bn sum being auctioned off by the Fed. The auction works as a dutch auction so the money is lent out at whatever rate banks bid for the amounts they specify. The Fed announcement set two auction dates in December (already passed, clearly -- they were fairly successful and didn't show ridiculous demand for cash, which is good) and two dates for the roll in January. In fixed income a "roll" indicates when one needs to go from one security to another, in this case cash borrowed over one term to cash to be borrowed over another term.
By lending directly to banks via this auction, the Fed is creating cash on the balance sheets of banks. Then banks don't need to borrow as much in the libor markets so libor can set down. Brilliant. They're still flooding the cash markets via open market operations for year-end, but we should expect a lot of the cash-hoarding issues to be settled.
Up till now the Fed had three major tools: the discount rate, the fed funds rate (via open market operations) and the reserve ratio. See my post about the Fed (another "learn the lingo" posting) if you need details about these. Now the Fed has introduced a new tool that lets it have a more direct impact on libor.
But wait! Isn't the fed funds rate directly tied to libor? Well, sorta, not really. There's certainly a relationship, but generally libor trades at a spread over fed funds. That spread historically sticks to 1% or less. Right now it has blown way out, so the Fed needs something to deal with that spread issue. Enter the TAF.
Here's how it works. Memeber banks (note: member banks need to be commercial banks or savings and loan instutions--specifically, this EXCLUDES the investment banks) can borrow up to 10% of a $20Bn sum being auctioned off by the Fed. The auction works as a dutch auction so the money is lent out at whatever rate banks bid for the amounts they specify. The Fed announcement set two auction dates in December (already passed, clearly -- they were fairly successful and didn't show ridiculous demand for cash, which is good) and two dates for the roll in January. In fixed income a "roll" indicates when one needs to go from one security to another, in this case cash borrowed over one term to cash to be borrowed over another term.
By lending directly to banks via this auction, the Fed is creating cash on the balance sheets of banks. Then banks don't need to borrow as much in the libor markets so libor can set down. Brilliant. They're still flooding the cash markets via open market operations for year-end, but we should expect a lot of the cash-hoarding issues to be settled.
Tuesday, January 1, 2008
New Year
It's a new year. And what's that mean for Wall Street? Well, those who haven't reset their annual numbers in December now reset their numbers. Everyone's on a fresh slate. PnL, sales credits and deal closings are all zero or close to it. From today on, anything that happened last year for deals, trades, etc all are forgotten and you get a new chance to shine. Last year's rock-stars are no longer and need to prove themselves once again. Market liquidity comes back and bankers are back on the phones with new-found zest. There are few industries where each year brings a completely new set of risks and rewards. If you sucked last year, you could be the star player this year. If you were the star last year, you could blow up this year. It's a whole new game.
So what's this mean for you? It means you need to stop thinking about last year and start thinking about this year. Letting last year's victories and follies mess with your mind is the best way to screw up a new year. If you got on bad terms with someone, forget about it. If you screwed up a trade, don't think about it. If you lost a deal, it's history. There are new winners to be made.
What if you're still looking for that job? Well the new year brings opportunities for you too. Payouts happen between now and February, which just happens to be when people start moving around too. People quiting means openings in various positions. Firms have to hire to replace those who leave, and there will be a lot of people leaving their firms this year. February to April is the biggest hiring season for the Street. Look forward to it and pursue aggressively.
Good luck to all in the new year.
So what's this mean for you? It means you need to stop thinking about last year and start thinking about this year. Letting last year's victories and follies mess with your mind is the best way to screw up a new year. If you got on bad terms with someone, forget about it. If you screwed up a trade, don't think about it. If you lost a deal, it's history. There are new winners to be made.
What if you're still looking for that job? Well the new year brings opportunities for you too. Payouts happen between now and February, which just happens to be when people start moving around too. People quiting means openings in various positions. Firms have to hire to replace those who leave, and there will be a lot of people leaving their firms this year. February to April is the biggest hiring season for the Street. Look forward to it and pursue aggressively.
Good luck to all in the new year.
Subscribe to:
Posts (Atom)
