Sunday, July 24, 2005
Old Edges
Well, spent about a month out of town but have been back trading for the last 2 weeks - figured it was time to bring some life back into this blog. I'll try to pick off around where I left off. I feel that the best way for one to gain an understanding of what is real edge is to simply study edges that have existed (or still exist) in the market but are saturated with participants and are therefore difficult or impossible to profit from - especially for a beginner.
We can start with the simplest edge, and the one carrying the least amount of risk (practically none), a true arbitrage. As we all know, market makers can't trade with/against each other. The days of the SOES bandit are long gone - large insitutions with ridiculously low costs run programs all day long arbing a crossed equities market the instant the opportunity arises. True arbs are practically limitless - I know quite a few people arbing the S&P mini on the screen against the big contract on the floor, although this edge is becoming quite saturated as well. Derivatives talk brings us to another edge with almost as little risk as a crossed market - arbing a derivative instrument against it's cash component (usually a combination of equities, or sometimes other derivatives). Take for example the thousands of programs arbing the index futures against their cash components all day long, usually to within a tick or two. These kinds of trades are very profitable but only for those with the technology and requisite low cost structure.
Old and well known edges range from these simple ones to some of the complex trades popularized by the original Salomon bond group and more prominently by Long Term Capital. These were no longer riskless trades - rather they were closer to statistical arbitrage usually relying, or betting on, a return to mean or normality for usually what was a spread between various derivates (such as an off-the-run and on-the-run treasury bond). While these trades were hedged, they were clearly not nearly as low risk as some of the arbs mentioned above. Statistical arbs in general are a very touchy category - one must consider the posibility that whatever is causing the normality can disappear (this applies to not only the kind of trades done at major institutional prop desks but also the common equities pair trade; this is especially evident in merger risk arbitrage). Although, it must be mentioned, what caused LTCM's spectacular meltdown was not so much ignornace of the fact that their trades carried risk but moreso the fact that they were severely overleveraged and were underestimating their risk and possible volatility inherit to the positions in their portfolio.
The edges mentioned here, and countless others, are quite well known and as a result are taken advantage of, and profited from, by those with cost structures and automated technologies advanced enough to compete in these cut-throat areas.
We can start with the simplest edge, and the one carrying the least amount of risk (practically none), a true arbitrage. As we all know, market makers can't trade with/against each other. The days of the SOES bandit are long gone - large insitutions with ridiculously low costs run programs all day long arbing a crossed equities market the instant the opportunity arises. True arbs are practically limitless - I know quite a few people arbing the S&P mini on the screen against the big contract on the floor, although this edge is becoming quite saturated as well. Derivatives talk brings us to another edge with almost as little risk as a crossed market - arbing a derivative instrument against it's cash component (usually a combination of equities, or sometimes other derivatives). Take for example the thousands of programs arbing the index futures against their cash components all day long, usually to within a tick or two. These kinds of trades are very profitable but only for those with the technology and requisite low cost structure.
Old and well known edges range from these simple ones to some of the complex trades popularized by the original Salomon bond group and more prominently by Long Term Capital. These were no longer riskless trades - rather they were closer to statistical arbitrage usually relying, or betting on, a return to mean or normality for usually what was a spread between various derivates (such as an off-the-run and on-the-run treasury bond). While these trades were hedged, they were clearly not nearly as low risk as some of the arbs mentioned above. Statistical arbs in general are a very touchy category - one must consider the posibility that whatever is causing the normality can disappear (this applies to not only the kind of trades done at major institutional prop desks but also the common equities pair trade; this is especially evident in merger risk arbitrage). Although, it must be mentioned, what caused LTCM's spectacular meltdown was not so much ignornace of the fact that their trades carried risk but moreso the fact that they were severely overleveraged and were underestimating their risk and possible volatility inherit to the positions in their portfolio.
The edges mentioned here, and countless others, are quite well known and as a result are taken advantage of, and profited from, by those with cost structures and automated technologies advanced enough to compete in these cut-throat areas.
