I was working at Morgan Stanley on that day (Black Monday - https://en.wikipedia.org/wiki/Black_Monday_(1987) - October 19th, 1987) in the pairs-trading group run by Nunzio Tartaglia where we ran a highly-automated strategy using Sharp APL on the IBM mainframe. Our strategy was so efficiently implemented that we typically accounted for more than one percent of the volume of the New York Stock Exchange using only 5 traders.
As I recall, we made about $10 million on that day because of our pairs-trading strategy - https://en.wikipedia.org/wiki/Pairs_trade - which means that we had an offsetting short position for every long position. This means that the money we lost on the decline of our longs was more than offset by the gain in our shorts. (I just noticed an error on that Wikipedia page where they claim that pairs trading was invented by Gerry Bamberger; I know from a conversation with a former boss that the strategy goes back at least to the start of the XX century but I do not have any written reference for that.) I recall the trading room as virtually silent that afternoon as the exchanges backed up and trades were extremely slow to execute. Fortunately, because of our matched book, sitting on our hands during that chaos - the largest stock market crash in history (still) - was not a bad thing to do. The slowdown was at least partially volume-related as the exchange cleared almost 600 million shares that day, well over the typical average of 100-200 million at that time. Today, total daily share volumes appear to be in the billions. On Sat, Aug 10, 2019 at 1:45 AM Donna Y <[email protected]> wrote: > Hi Bob > > Thanks for all those details. > > I heard this third hand—it was not about a system crash-- at the time > Morgan Stanley trading systems were probably the best in the industry—I > think what I heard was there was surprise that the automated systems had > parameters that created a spiral of rapid trading into falling prices—also > I think Morgan Stanley offset equity losses with gains in currency trades. > They subsequently modified their software to include a circuit breaker-- a > mechanism to shut down trading when the market falls too fast or individual > securities trade dramatically outside the normal range. > > What I meant by a ride was not that it necessarily ended in a loss but > that there was tension over the course of that day. My personal > recollection is only what I heard from colleagues in our investment > department. > > Computers now account for 50 to 90 percent of stock market trades on a > given day. > > > Stock exchanges can now execute trades in less than a half a millionth > of a second—more than a million times faster than the human mind can make a > decision. > > > Donna Y > [email protected] > > > > On Aug 10, 2019, at 1:00 AM, Robert Bernecky <[email protected]> > wrote: > > > > Hi, Donna, > > > > As I recall, on Black Monday, Morgan Stanley was still one of > > our (I.P. Sharp) major customers running SHARP APL on a suitable > > number of their mainframes, which were likely IBM 308x complexes. > > > > The story I heard from an insider ran along the following lines, > > about the sorts of failures that arose that day: > > > > 1. Trading volumes and currency volumes were so high that some > > "traditional" trading companies, with their automated systems > > written in C-likelanguages, suffered from integer overflow, > > which did NOT crash their systems. Rather, they merely > > collected a (2*32) or similar residue on the volumes and > > other numeric data of interest, which resulted in traders > > being given extremely misleading data (clearly no longer > > information at this point...), which resulted > > in the traders working themselves into very deep holes. > > > > 2. Other companies, running COBOL-like languages, crashed when > > their code encountered integer overflows. Being sensible firms, > > they had written transaction logs, which faithfully tracked all > > work that day. When the systems came back up, they carefully > > replayed all the transactions, hit the integer overflows, and > crashed. > > > > 3. Even other companies, running some other languages, were > > unable to keep up with trade rates, and although they did not crash, > > fell far behind, giving traders misleading data, which became > > more misleading as the day went on. > > > > 4. Morgan, and presumably others running APLish systems, > > observed two effects: First, the systems did slow down, due to > > the high trading volumes. They did not crash, because integer > > overflow merely meant that ongoing computations were in > > floating point, rather than in integer. > > This caused further slowdown, but not disasterously so. > > > > 5. Morgan made money that day. Not a lot, but certainly more than > > their competitors. > > > > I no longer have contacts with people who are/were at Morgan, > > but as the story I heard is not in line with the "Morgan Stanley > > was taken for a ride" claim, it would nice to learn more about > > what really happened. > > > > Bob > > > > On 2019-08-09 7:59 p.m., Donna Y wrote: > >>> Investigations after the October 19, 1987 crash revealed that what > would have been a normal down day in a correction that had begun in August > was turned into the heart-stopping, portfolio destroying 1987 crash by > uncontrolled automated waves of sell-programs that flooded in from > program-trading firms and overwhelmed the market. As their ‘portfolio > insurance’ protective stops were successively hit the automated sell orders > came so fast on top of each other at ever lower prices that market-makers > could not match them up with buyers. Very quickly there were no buyers > anyway, and the decline just plunged into a dark bottomless hole. > >> This was the beginning of programmed trading. > >> > >> At the insurance company where I worked the investment department was > not even affected but I heard Morgan Stanley was taken on a ride by their > new program trading software. > >> > >> Natural disasters like what happened with the earthquake in Japan you > mention are not predicted but even a tweet from Trump can perturb the > market or as what happened in 1981: > >> > >>> The selling spree was set off by Joe Granville's January 1981 > newsletter, which advised investors to "sell everything". It was later > described by Business Week magazine as "a mindless wave of selling that > destroyed billions of dollars in stock value from a forecaster who drops > his pants in public to get attention." > >> > >> Donna Y > >> [email protected] > >> > >> > >>> On Aug 9, 2019, at 7:09 PM, Jose Mario Quintana < > [email protected]> wrote: > >>> > >>> Regarding Black Monday, apparently, the ones who knew better did not > have > >>> enough conviction to short the S&P500 in any considerable amount > before the > >>> event occurred. > >> ---------------------------------------------------------------------- > >> For information about J forums see http://www.jsoftware.com/forums.htm > > > > -- > > Robert Bernecky > > Snake Island Research Inc > > 18 Fifth Street > > Ward's Island > > Toronto, Ontario M5J 2B9 > > > > [email protected] > > tel: +1 416 203 0854 > > text/cell: +1 416 996 4286 > > > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > -- Devon McCormick, CFA Quantitative Consultant ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
