Amsterdam: Wow. The Dutch know their crowdsourcing. I spoke yesterday to an audience of 350. I wanted to know who I was dealing with. I'm pretty accustomed to speaking to people who've never even heard of the term before. "Who's heard of Threadless.com?" I ask. Almost every hand goes up. That's okay, I figure, Threadless is an international sensation at this point. "Who's heard of InnoCentive?" Now I've got them. Outside of the sciences, almost no one I talk to has heard of InnoCentive. Two hundred hands shoot up. I was impressed, and a little intimidated. That said, the program was a smash, from my perspective anyway. The audience asked detailed, provocative questions and my hosts were gracious, witty and incredibly generous, treating me to a dinner both wonderful and wonderfully leisurely, in the proper European manner. But now I'm back, and so, I hope, are you. Below please find the continuation of Chapter Seven:
Marketocracy: Collective Intelligence Improvised
Think of TJ White as the argument for diversity, personified. In 1999 not many investors would have picked White as a candidate to manage their portfolios. Up to that point White had spent his life accumulating a lot of experience doing very little. He was an unremarkable student at his Midland, Texas high school. He signed up for an unremarkable six-year stint in the US Navy, followed by another six-year stint in Colorado pursuing what he calls his “second childhood,” working a series of jobs meant only to “support a lifestyle” that revolved around skiing and gold prospecting. He excelled at neither. One morning late that year White woke up, looked out at the parking lot outside his one-bedroom apartment and had a moment of clarity. “I was a loser. I was 30, and I had nothing. No skills, no college, no career.”
A few days after New Year’s Day 2000 White moved to Dallas, and the puzzle pieces started falling into place. He quickly found a job at a Home Depot not far from his house. “The manager was ex-Navy, and we hit it off,” White recalls. A few weeks after that he met the woman who would become his wife, at a get together of the Tall Texans of Dallas. “I stumbled across the Website when I first moved to Dallas,” explains White. “I’m six-foot-two, and Cheri’s six-foot-two. We got serious pretty quickly.”
And then White found what he believed to be his calling: Day trading stocks over the Internet. “Everyone was talking about millionaires working at regular jobs like Home Depot. Why not me?” (Many Home Depot employees minted a fortune in company stock during the late ‘90s.) White convinced Cheri to let him invest her life savings, combined it with his own and began investing. He put $6,000 in shares of a technology company he read about in the Dallas Morning News, and another $4,000 in other tech stocks. By the end of the year the bottom had fallen out of the technology sector, and White’s money had evaporated.
TJ and Cheri went to a local hamburger joint for a talk. TJ wanted her to give him the rest of her life savings—$2,000 to invest. “She looked at me across the table, took my hands in hers and said, ‘You’re not good at this. You tried, and you lost.’” White emerged from the experience poorer, but the recipient of two valuable lessons. First, he’d discovered that he loved the process of investing—pouring over prospectuses, studying profit-earnings ratios and separating inflated claims from real potential for growth. More importantly, he realized that he should never invest in a company whose business he couldn’t understand. “I don’t know anything about bio-technology or computers. But I can wrap my head around someone digging a hole in the ground to look for some oil.”
By the next year White had discovered a safer outlet for his passion. An investment firm called Marketocracy allowed people to create “model portfolios” on their Website. In other words, Marketocracy is a prediction market for the stock market. After a free registration, anyone could open up to 10 accounts, each of which would start with $1 million in Monopoly money. White stocked his first fund with broad investments in what he calls the “blue collar industries.” At first White continued to lose money, but soon he stopped trying to day trade and started looking for long-term values. He put an elementary formula to work, restricting his positions to companies in which growth rate exceeded their price-to-earnings ratio. This is the recipe advocated by such investors as Warren Buffet and the Fidelity fund manager and author Peter Lynch, who advocated an “invest in what you know” strategy, but White says he didn’t know that at the time. “I thought I’d invented it.”
White’s instincts have proven remarkably effective. He has now been trading on Marketocracy for seven years, and his track record has beat even the best funds on Wall Street. If you had given White $1 million dollars back in 2001, you would have $4,176,000 right now. Not all of White’s investments have been in phony dollars. By slowly funneling his meager savings into brokerage accounts, White was able to quit his job at Home Depot soon after he and Cheri wed in 2005. “I have $166,000 invested right now, and we’ve paid off two cars.” Cheri quit her job at a software company and now helps run a doggie day care. “She hated her job, to put it politely,” says White. “That’s been the best part of all this.”
Discovering such unlikely investment geniuses as White is what Marketocracy does best. Over 100,000 people have created what the company calls “model portfolios,” and roughly 20,000 of those are considered “active traders,” in that they regularly—even compulsively—monitor their portfolios. Marketocracy watches the performance of these ersatz fund managers and uses a selection of the top-performing 100 portfolios to guide the investment decisions of their “Masters 100” fund, which has about $35 million in very real assets under management. That’s not a lot of money for a mutual fund, but it’s a considerable vote of confidence for a such an unorthodox approach to investing.
On its face, Marketocracy would seem to be standard-issue collective intelligence crowdsourcing. While a large chunk of Marketocracy’s so-called “masters” hail from stock-related industries, a surprising number are a lot like TJ White—attorneys, chefs, geologists and others who simply have a special insight into one sector of the market and a knack for sniffing out a bargain, or knowing when to take a profit on a high flyer. It’s what Page calls a “crowd of models”—which means Marketocracy is basing its decisions not on the crowd, but on a smaller crowd of the best performers. “It’s like a crowd of experts,” says Page. In this case, the experts include a mix of Mensa types with a heavy peppering of Brown Socks. It would seem to be a winning formula: Since its inception in late 2001 the Marketocracy “Masters 100” has outperformed that stock market benchmark, the Standard & Poor’s 500, by an average of nearly 40 percent. Such a track record would seem to make a clear-cut case for the virtues of a diverse crowd.
But the reality behind Marketocracy’s investment management approach is both more interesting and more complicated. It also reveals a lot about the tricky ways in which collective intelligence manifests, and the conditions that must be maintained to facilitate it. Over the years Marketocracy has created a fine-tuned hybrid that boasts the best qualities of both a crowdcasting network and a prediction market. It relies on a straight portfolio performance to identify the diamonds in the rough like TJ White, but following a few disastrous quarters, the firm has also learned to exercise a great deal of discretion, sometimes following their 100 elite investors and occasionally breaking from them.
Kam and Taguchi weren’t thinking about group intelligence at all when they created the company; they were simply looking for a better system to identify trading talent. From 1994 to 2000 the pair helped run Firsthand Funds, which outperformed every other mutual fund in that period, averaging a 56 percent return on investor’s money in its first five years. When they left to start their own fund, they were inundated with resumés. In order to identify the best trading talent, they asked applicants—and anyone else with an Internet connection—to create a mock portfolio. The hope was that some 5,000 people would set up accounts with Marketocracy, and that after a year or two the pair would have enough data to hire the best traders of the bunch to help them manage their new fund.
“That goal changed pretty quickly,” Taguchi laughs, exchanging looks with Kam. Some 50,000 would-be investors signed up for mock portfolios in the first year. “We’ve always embraced a team idea,” says Taguchi. “But that idea evolved into just using a very large team”—in other words, the 100 best investors at any given time. In November 2001, Marketocracy launched the Masters 100. A lot of eyes were trained on Marketocracy. Kam and Taguchi had been rock stars of the market’s last bull cycle, and there was a lot of interest in—and skepticism about—their unusual new approach to managing a mutual fund.
At first it seemed Kam, Taguchi and the crowd would all be vindicated. During the first year of its operation the market was in a full-scale retreat, but the Masters 100 outperformed the market from the start. By the end of the first year, Marketocracy had beat the market by some 14 percent. By late 2002 the S&P 500 hit its nadir, and stocks began climbing as the economy started heating up. At the time Kam and Taguchi were running the fund using the simplest of formulas. The company simply allocated its assets in almost precisely the same ratios as their masters. “At first we equally weighted the fund’s positions to reflect those of the Top 100,” says Taguchi. “So if the Masters took a three percent position in Apple, we did too.”
The fund continued to outperform the market. Marketocracy’s crowd-powered investment strategy worked great in the bear market of 2002 and the bull market of 2003. “In 2002 the defensive people took control,” says Kam. These investors were well-suited for making decisions in a down market, but as stocks began trending northward Kam and Taguchi wanted a different breed of trader at the helm. “In 2003 we started subbing in the more aggressive investors.” It looked like a foolproof model. In 2003 the fund returned a remarkable 42.82 percent, and a flood of investment money started pouring in.
But then in 2004 the market entered a new, more complicated phase. “It was a choppy market. Up and then down, and not always the same sector.” The M100 underperformed the market, and investors began fleeing the fund, which dropped from nearly $100 million assets under management—the key metric for an investment fund—to $50 million in just over a year. Clearly there were bugs in Marketocracy’s algorithm. For one, Kam and Taguchi realized that as the top investors got to know each other they started conferring on their positions. (Marketocracy hosts events where members can meet and talk shop.) This had an upside—members, for instance, could learn about industries they didn’t currently invest in—but a big downside: Deliberation is the enemy of collective intelligence, because it reduces diversity. As individuals confer, they also reach consensus. One of the chief conditions that allows the crowd to make smart predictions or come up with novel approaches to a problem is autonomy: They make their choice independently. “We started seeing a herd mentality emerge even among our best traders,” says Kam with a sigh. So Kam and his team instituted a number of changes to the site, one of which made it impossible for members to watch each other’s trades. “It helped immediately.”
But Marketocracy’s real breakthrough was to realize that their pool of top performers—the top 100—was too small, and that using an algorithm to guide investments was too limited a methodology. The company wasn’t benefiting from the talents and abilities of other traders who, while not performing as well as the very best, still possessed some unique expertise that might allow for a handsome profit on a trade. So Marketocracy reached out, and began engaging the entire community on its decisions.
Not long after Marketocracy crashed in 2004, Kam and Taguchi put their hybrid model to work. “We noticed that a certain subset of our traders were buying loads of shares in this oil shipping firm called Knightsbridge Tankers,” says Kam. No one in the Masters 100 had touched it or likely even heard of it. But a bunch of the less elite traders with accounts on Marketocracy were loading up on it. “The stock was trading at historic lows, and all these people were swimming upstream by investing in it. We wanted to know why.” So they sent out some emails to the traders. “We received this incredibly detailed information.”
It turned out that those traders had access to information that neither the Masters nor Wall Street investors had access to. “The company had all these tankers that were about to be scrapped. “How would you know that? Who would know that?” asks Kam, incredulously. “It turns out the tankers were registered in Singapore, and there were these guys that went to Singapore to look at the registrations. Incredible. The conventional wisdom is that when a tanker reaches the end of its life, it’s worth zero. But the price of steel started to go through the roof in the interim, and all that was about to be returned to investors as dividends.” Marketocracy made a killing.
The company effectively became a prediction market with important components of a crowdcasting network. “The reason diversity trumps ability in a problem-solving scenario is because you can always throw the idiots off the bus,” says Page. In the case of Marketocracy, by following the Master’s 100 too closely, they couldn’t exercise that option. Page explains, “Picking stocks is part prediction and part problem-solving, so Marketocracy’s approach makes a lot of sense.”
Tapping people’s collective intelligence involves trafficking in what the crowd already knows. Such crowdsourcing applications generally require small investments of time and energy on the part of individual contributors. They are also what we might think of as additive—they help people do better at their jobs, but don’t threaten to replace existing employees. As we’ll see, other forms of crowdsourcing promise to create much more disruption. In a few cases, that disruption is already occurring.