The Right Place at the Right Time

It all looked like it was going to hit the proverbial fan one afternoon during the financial crisis, however, when he was photographed outside of a pub in a crowd of bankers who had just been laid off from a nearby institution, pint and cigarette in hand, and said photograph was splashed on the front page of the city freesheet’s evening edition. One of the senior editors, a fearsome Scot with a hair-trigger temper came storming into the newsroom, the paper clenched tightly between his fist, his face an extraordinary shade of puce, demanding to know where said reporter was. Nobody wanted to say, and after a few moments of uncomfortable silence, someone volunteered that he wasn’t there. “Of course he isn’t,” thundered the senior editor. “He’s where you should all bloody be, getting the story from the lads who just got sacked, doing real journalism, God’s work!”

Thus, in one of the stranger turns of the newspaper industry, we were all soundly remonstrated for not being at the bar mid-afternoon, and ordered there sharpish if we valued our careers.

The point of this story is twofold—one, newspapers are a weird business. Two, sometimes the appearance of excellence isn’t all it’s cracked up to be. Take, for instance, the explosion of interest in sentiment analysis a few years ago, and questions about whether people could predict the stock market by running Twitter’s data firehose through a series of algorithms. I met people during that time who swore they could see the future through such wizardry, and others who clearly didn’t believe a word of it, yet were obliged to say it might have merit because their companies had invested thousands in software and staff with multiple PhDs.

Lo and behold, when market volatility began to dampen, and serious questions were asked about exactly who, or what, constituted the majority of tweets, the question of whether it should be used to inform trading decisions quietly began to be sidelined. Even more so now, when someone posting a status could just as likely be a paid-by-the-tweet teenager in Eastern Europe as they could the American housewife they claim to be. Welcome to the world of digital psyops, it’s not a fun place if you’re a “smart” sentiment algorithm.

The same thing happens now when people talk about artificial intelligence (AI), or distributed-ledger technology (DLT), or cognitive computing, or quantum, or any number of other new and shiny technologies. Some have proven through proof-of-concept trials and risk management exercises that they could have handled historic periods of market stress well. The problem with this is that the parameters of those events are known, and that colors the judgment of those examining the performance. 

Future events, such as what many experienced recently with volatility options, come out of nowhere and have the ability to T-bone a portfolio in minutes. Anyone claiming that technology holds the answers to market problems should be taken with a pinch of salt—as should anyone claiming unproven and immature technologies will fundamentally change the markets in the future.

Likewise, the fundamental theory or vehicle behind these technologies—as happened with Twitter—could be thrown into doubt through outside developments that nobody could see coming, like investigations into the integrity of US electoral processes that focus on social media.

Frankly, until real deployments bed in, and until real data is passed through real systems—indeed, until they handle real events and go through real black swans—nobody really knows how systems will react. Oftentimes, like my erstwhile, ale-loving colleague, it’s just a case of being in the right place at the right time, rather than being there by design. It’s important not to lose sight of that. W