还记得那次闪电式大崩盘吗？2010年5月6日，道琼斯工业平均指数在短短几分钟内暴跌600点。市值达1,650亿美元的消费品巨头宝洁（Procter and Gamble）在几秒钟内就遭受了37%的损失。埃森哲（Accenture）和爱克斯龙（Exelon）跌至每股1美分。所有股票齐齐跳水。

    然而在接下来的一瞬间，一切又都恢复了正常。市场收复了几乎所有失地。问题是，没有人能解释刚刚发生了什么。就算你是摩根士丹利（Morgan Stanley）的大客户也没用，就是没有答案。

    记者最终还原了整个事件的原委。原来，堪萨斯州的一家共同基金公司在期货市场做了一份卖空，因此引发了抛售的连锁反应。在某些方面，这种解释比我们之前一无所知的状态更可怕。难道美国股市已沦落到这种地步吗？一只堪萨斯州的蝴蝶扇动了一下翅膀，就会导致整个股市崩盘吗？

    在《暗池交易》（Dark Pools）一书中，《华尔街日报》（Wall Street Journal）的记者斯科特•帕特森解释了我们陷入这场混乱的原委。有个问题已经在华尔街蔓延十多年了，而闪电崩盘只不过是它最主要的症状。高速交易员，也被戏称为高频交易员，因为他们每秒进行的股票买卖能达到数千次，远远超过股市。

    一方面，他们提供持续的流动性，使得普通投资者不用向纳斯达克或纽约证券交易所的中间商（所谓的做市商）支付惊人的价差。另一方面，他们的不透明技术已为我们上演了诸如闪电崩盘的惨剧。

    作为一名长于数学的财经记者，帕特森善于深挖有关高速交易——华尔街新宠——的生动报道。他的上一本书《宽客》（2010年）详尽描述了量化交易员的全新世界，他们利用计算机算法来处理交易中的苦差事。而在《暗池交易》中，他则瞄准了市场上的跳水现象，在这个过程中，宽客可以在几微秒（百万分之一秒）内进行股票买卖。

    这种跳水最初由纽约斯塔腾岛区的交易员和编程怪才乔希•莱文建立，他们形成的联盟看起来有些另类。最后，天体物理学家也加入进来。他们合力把美国股市改造成了超出前人想像的高速机器。

    Remember the flash crash? On May 6, 2010, the Dow Jones Industrial Average plunged by 600 points over a couple of minutes. Procter and Gamble (PG), the $165-billion consumer giant, lost 37% of its market cap within seconds. Accenture (ACN) and Exelon (EXC) dropped to a penny a share. All hell seemed to be raining down.

    In the next instant, everything was back to normal. The market regained almost all of its losses. Problem was, no one could explain what had just happened. It didn't matter if you were Morgan Stanley's favorite client, there just weren't answers.

    Journalists eventually pieced the story together. It turned out that a mutual fund company in Kansas had set off a chain reaction of selling by placing a single order in the futures market. In some ways, this explanation was more terrifying than our previous state of ignorance. Had the U.S. stock market come to this? Crashing because a Kansas butterfly flapped its wings?

    In Dark Pools, Wall Street Journal reporter Scott Patterson explains how we got into this mess. The Flash Crash was the first major symptom of a problem that has been spreading across Wall Street for more than a decade. High-speed traders, dubbed high-frequency traders because they trade in and out of stocks thousands of times per second, have overtaken the stock market.

    On the one hand, they provide constant liquidity so regular investors don't get charged egregious spreads by middlemen -- the so-called market makers -- at Nasdaq or the New York Stock Exchange. On the other, their opaque technology has given us episodes like the Flash Crash.

    Patterson, a wonderfully numerate financial journalist, is good at ferreting out vivid stories about high-speed trading, the Street's newest infatuation. His previous book, The Quants (2010), was a rich narrative about the new world of quantitative traders -- guys who let computer algorithms do the hard work of trading. In Dark Pools he zeros in on the market plumbing that has allowed quants to jump in and out of stocks in microseconds (millionths of a second).

    The plumbing was first built by an unlikely alliance between Staten Island traders and a programming nerd named Josh Levine. Eventually astrophysicists joined in. Together, they retooled the U.S. stock market into a speed machine unlike anything ever imagined.

    这一切是如何影响普通美国401(k)养老金计划投资者的？高频交易员每秒能做成千上万份交易。他们不断搜罗着那些小小的机会。让我们举例子说明。假设你的富达基金经理要买入50,000股埃克森（Exxon），股价75.20美元。他不会一下子就完成整个交易，而是1,000股、1,000股地分批吃进。

    他以75.20美元购买了1,000股后，高频算法察觉到某个大型投资者正在买入埃克森。因此，他们也开始买入埃克森，并把价格推到75.22美元甚至更高。那位富达的经理然后以75.25美元购入了另一批。高频交易员再次跟进，把价格拉高至75.30美元。当富达经理买入最后1,000股时，埃克森已一路涨到75.50美元。这意味着该经理的最后一次购买损失了250美元，因为他的买入价已不是75.25美元，而是75.50美元。而那250美元原本应成为你的投资，40年后的保守回报有望达到2,500美元。

    帕特森没有大谈特谈这种高频交易，而是引人入胜地叙述了这个新世界的疯狂增长及其危险后果。他遵照迈克尔•刘易斯公式，找到了那些鲜为人知的英雄，并解释了那些复杂的金融策略。其中一个就是莱文，从根本上说，这位温和的程序员发明了现代电子股票交易市场，他的工作地点是一间位于曼哈顿南区宽街（Broad Street）的办公室，里面到处是垃圾，还有几只宠物龟在游泳池里游泳，角落里矗立着以色列反坦克火箭筒。

    莱文在20世纪90年代，开创了称做“岛”（Island）的电子交易平台，以对抗他在纽约证券交易所和纳斯达克看到的不公平垄断，那是做市商在进行股票交易的地方。其中的问题是，中间商会勾结起来，牟取每笔交易中的巨大利差。

    具有讽刺意味的是，帕特森发现：为了建立没有中间商的电子交换，莱文需要高频交易员来为买家和卖家提供的流动性。最终，其他电子交易平台，也就是所谓的池，开始形成。结果，高频交易员成为了新的中间商，为池提供了生存所需的交易量。

    举个例子：有一天，位于芝加哥的高平交易公司、同时也是做市商Getco正在举行晨会。纽约股市开盘后五分钟，一个抓狂的电子池职员打来电话，质问为什么Getco还不进行交易。

    这种新兴的电子股票从内部蚕食着纳斯达克和纽约证券交易所。在过去的好日子里，纽约证券交易所把持着美国股市90%的交易。而今天，它处理的量只有四分之一。

    接下来会发生什么？帕特森唯一可以确定的是，高频趋势将会继续下去。他写道，一家名为Spread Networks的公司在芝加哥和纽约的交易中心之间建立起超高速连接，这个价值3亿美元的项目会将光缆直接铺到新泽西州的纳斯达克数据中心。其结果会如何呢？每轮交易的时间将减少3毫秒（千分之三秒）。

    How does all this impact everyday 401(k) investors? High-frequency traders place hundreds of thousands of orders each second. They are constantly on the prowl for small opportunities. So let's say your mutual fund manager at Fidelity is buying 50,000 shares of Exxon (XOM). We'll assume that the stock trades for $75.20. He won't place the whole order at once, instead buying piecemeal in 1,000-share blocks.

    After he buys 1,000 shares at $75.20, the high-frequency algorithms sense that some big investor is buying Exxon. So they also start buying Exxon, pushing the price up to $75.22 and higher. The Fidelity manager then buys another block of shares at $75.25. The high-frequency traders swoop in again and push the price up to $75.30. By the time the Fidelity manager buys his last batch of 1,000 shares, Exxon is all the way up to $75.50. That means the manager lost $250 on the last block by buying Exxon at $75.50 instead of $75.25. That $250 should have been invested for you and, estimating a conservative return over 40 years, grown to $2,500.

    Patterson skips across the high-frequency landscape in an engaging narrative that tracks this new world's blinding growth and its perilous consequences. He follows the Michael Lewis formula of finding little-known heroes to explain complex financial maneuvers. One is Levine, a meek programmer who basically invented modern day electronic stock markets from an office on Broad Street in Lower Manhattan stuffed with trash and pet turtles swimming in a pool, not to mention an Israeli bazooka standing in the corner.

    In the 1990s, Levine started an electronic exchange called Island to fight what he saw as unfair monopolies in the New York Stock Exchange and Nasdaq, which used market makers to execute stock orders. Problem was, the middlemen colluded to skim huge spreads off of each order.

    Patterson quickly gets to the irony: In order to build an electronic exchange without middlemen, Levine needed high-frequency traders to provide liquidity for buyers and sellers. Eventually other electronic exchanges -- called pools -- started forming. High-frequency traders became the new middlemen, providing the trading volumes the pools needed to survive.

    Case in point: One day a morning meeting went long at Getco, a high-frequency trading firm in Chicago. Five minutes after the start of trading in New York, a frantic Island official called asking why Getco wasn't trading yet.

    This new world of electronic pools of stocks eviscerated demand at the Nasdaq and the venerable New York Stock Exchange. In its halcyon days, NYSE hosted 90% of U.S. stock trading. Today, it handles a quarter.

    What's next? Patterson is only sure that the high-frequency trend will continue. He writes about a company called Spread Networks building a super-fast connection between the trading hubs in Chicago and New York, a $300 million project to lay fiber optic cable straight into a Nasdaq data center in New Jersey. The upshot? Cutting 3 milliseconds (three one thousandths of a second) off of the round trip of a trade.

    《暗池交易》介绍了股票交易的新世界，是一本颇具娱乐性、简单易懂的书，尽管帕特森的标题有些误导。他用了概括性术语描述了整个美国市场，而暗池从技术上讲只是一个细分市场，在其中进行的买卖交易都避开了公众视线。

    更棘手的问题是，帕特森没有解释如何防范高频机器人和具有学习能力的先进算法，避免它们引发新的股市崩溃。揭露了我们所面临的可怕现实，详细阐述了推动股市的各种力量是一回事。但要找到解决方案则是另外一回事。

    因为，这个问题根本就没有简单的解决办法。不幸的是，证券交易管理委员会（SEC）在管控方面已经滞后，但个别的高频交易员是不会兴风作浪的，这就像一家银行的倒闭不会使整个系统陷入瘫痪。看来，要想了解如何监管美国股市——这个新兴的狂野西部，我们或许不得不等待作者的下一本书了。

    译者：杜伟华

    Dark Pools is easily the most entertaining and accessible book to cover the new world of stock trading, even though Patterson's title is misleading. He uses the umbrella term to describe the entire U.S. market, although dark pools are technically sub-markets that mask buy and sell orders from public view.

    A bigger problem is that Patterson doesn't explain what will prevent high-frequency robots and advanced-learning algorithms from causing the next market collapse. It's one thing to detail the scary new reality in which these forces drive our stock market. It's another to offer solutions.

    That's because there are no easy fixes. The SEC is woefully behind in regulating, and individual high-frequency traders aren't problematic, just as one bad bank won't cripple the system. We may have to wait until his next book to learn how you police the new Wild West, more commonly known as the U.S. stock market.