《The Man Who Solved The Market》中文版翻译连载56
he Man Who Solved The Market (56)
As a teenager, Peter Brown watched his father deal with a serious of daunting business challenges. In 1972, when Peter was seventeen, Henry Brown and a partner came up with the idea of cobbling together investments from individual investors to buy relatively safe, yet higher-yielding debt, introducing the world’s first money-market mutual fund. Henry’s fund offered higher rates than those available in bank saving accounts, but few investors had even a passing interest. Peter would help his father stuff envelopes and mail letters to hundreds of potencial customers, hoping to elicit interest in the new fund. Henry worked every day that year except Christmas, resorting to eating peanut-butter sandwiches and taking out a second mortgage to fund his business, as his wife, Betsey, worked as a family therapist.
“A combination of starvation and pure greed drove us,” Henry explained to the Wall Street Journal.
His lucky break came the next year in the form of a New York Times article about the fledgling fund. Clients began calling, and soon Henry and his partner were managing $100 million in their Reserve Primary Fund. The fund grew, reaching billions of dollars, but Henry resigned, in 1985, to move with Betsey to the Brown family’s farm in a Virginia hamlet, where he raised cattle on five hundred acres. Henry also competed in trebuchet, a kind of mechanical catapult, wining competitions with a contraption that sent an eight-pound pumpkin over one thousand feet. In their new neighborhood, Betsey became a civic activist and local Democratic politician.
第二年，《纽约时报》刊登了一篇关于这只羽翼未丰的基金的文章，让亨利迎来了好运。客户们开始打电话，很快亨利和合伙人的这只Reserve Primary Fund的管理规模就超过了1亿美元。基金规模不断增长，达到数亿美元，但是亨利在1985年辞职了，和贝齐一起搬到布朗家在弗吉尼亚州的一个小村庄的农场，在那里他有500英亩土地养牛。亨利还参加了投石机比赛，这是一种机械弹射器，他凭借一个精巧的装置赢得了比赛，这个装置能把一个8磅重的南瓜发射到1000英尺以外。在新社区，贝齐成为一名公民活动家和当地的民主党政治家。
Henry’s business still dominated his thoughts, though. For more than a decade, he squabbled with his former partner, Bruce Bent, whom Henry accused of reneging on an agreement to buy his half-interest in the company. Henry eventually filed a lawsuit, claiming Bent was rewarding himself excessively while running the fund, before the men finally worked out a deal for Brown to sell his half-ownership to Brent in 1999. ( In 2008, the fund would lose so much money from the debt of investment bank Lehman Brothers, among other things, that its troubles would sow fear throughout the financial system.)
While his family had wealth, friends say Peter sometimes expressed anxiety about his finances, perhaps due to his father’s early challenges or his extended battle with his partner. Peter reserved his own ambitious for science and math. After graduating from Harvard University with an undergraduate degree in mathematics, Brown joined a unit of Exxon that was developing ways to translate spoken language into computer text, an early form of speech-recognition technology. Later, he’d earn a PhD incomputer science from Carnegie Mellon University in Pittsburgh.
In 1984, at the age of twenty-nine, Brown joined IBM’s speech group, where Mercer and others had been working to develop computer software to transcribe spoken text. Conventional wisdom in the decades-old field was that only linguist and phoneticians, teaching computers rules of syntax and grammar, and a chance at getting computers to recognize language.
Brown, Mercer, and their fellow mathematicians and scientists, including the group’s hard-driving leader, Fred Jelinek, viewed language very differently from the traditionalists. To them, language could be modeled like a game of chance. At any point in a sentence, there exists a certain probability of what might come next, which can be estimated based on past, common usage. “Pie” is more likely to follow the word “apple” in a sentence than words like “him” or “the”, for example. Similar probabilities also exist for pronunciation, the IBM crew argued.
Their goal was to feed their computers with enough data of recorded speech and written text to develop a probabilistic, statistical model capable of predicting likely would sequences based on sequences of sounds. Their computer code wouldn’t necessarily understand what it was transcribing, but it would learn to transcribe language, nonetheless.
In mathematical terms, Brown, Mercer, and the rest of Jelinek’s team viewed sounds as the output of a sequence in which each step along the way is random, yet dependent on the previous step — a hidden Markov model. A speech-recognition system’s job was to take a set of observed sounds, crunch the probabilities, and make the best possible guess about the “hidden” sequences of words that could have generated those sounds. To do that, the IBM researchers employed the Baum-Welch algorithm — codeveloped by Jim Simons’s early trading partner Lenny Baum — to zero in on the various language probabilities. Rather than manually programming in static knowledge about how language worked, they created a program that learned from data.
Brown, Mercer, and the others relied upon Bayesian mathematics, which had emerged from the statistical rule proposed by Reverend Thomas Bayes in the eighteenth-century. Bayesians will attach a degree of probability to every guess and update their best estimates as they receive new information. The genius of Bayesian statistics is that it continuously narrows a range of possibilities. Think, for example, of a spam filter, which doesn’t know with certainly if an email is malicious, but can be effective by assigning odds to each one received by constantly learning from emails previously classified as “junk”. ( This approach wasn’t as strange as it might seem. According to linguists, people in conversation unconsciously guess the next words that will be spoken, updating their expectations along the way.)