Dark Pools

1. The Market's Digital Transformation

Levine foresaw a market in which all information streamed seamlessly through microprocessors.

An antiquated system. In the 1980s, the U.S. stock market, particularly the New York Stock Exchange (NYSE) and Nasdaq, was dominated by human middlemen—specialists and market makers. These individuals controlled trading, often through shouted orders and hand signals on physical floors or over the phone, charging hefty fees (spreads) and operating with limited transparency. Joshua Levine, a teenage computer programmer, viewed this system as inefficient and ripe for disruption.

Vision of automation. Levine envisioned a future where computers would automate the matching of buy and sell orders, eliminating the need for human intermediaries. This computerized market would offer:

• Efficient order tracking
• Elimination of paperwork
• Significant cost savings
• Real-time price visibility for all investors

Challenging the status quo. Levine also challenged the archaic practice of quoting stock prices in fractions (e.g., $2.50) rather than pennies (e.g., $2.45), arguing that fractions artificially widened spreads and benefited insiders. He believed that technology could force greater honesty and transparency, ultimately democratizing access to market information and trading.

2. The Rise of the SOES Bandits and ECNs

Houtkin knew he’d discovered a gold mine.

Exploiting loopholes. The Small Order Execution System (SOES), initially designed by Nasdaq for small investor orders, became a "gold mine" for traders like Harvey Houtkin and Shelly Maschler. They discovered that by rapidly buying and selling stocks from slow-moving market makers caught with stale quotes, they could pocket instant profits. These traders, dubbed "SOES bandits," used early computer systems to exploit fleeting price discrepancies.

Datek's technological edge. Maschler's firm, Datek Securities, became a formidable SOES bandit operation, largely due to Joshua Levine's programming genius. Levine developed:

• The Watcher: An automated system that tracked orders, positions, and profits, allowing traders to monitor hundreds of stocks simultaneously and place orders rapidly using keyboard shortcuts.
• The Monster Key: An algorithm within the Watcher that instantly calculated prices 20% away from the best bid/offer, allowing traders to jump to the front of the queue and secure trades at the best available price.

Regulatory backlash and growth. Nasdaq and its market makers fiercely resisted the SOES bandits, imposing fines and restrictions. However, the bandits' success, fueled by Levine's innovations, demonstrated the power of electronic trading. Datek's volume soared, and the Watcher became a precursor to the day-trading explosion of the 1990s, showing how technology could empower individual traders against established institutions.

3. Island's Revolutionary Transparency

Island was the first fully lit pool.

A new trading paradigm. Frustrated by Nasdaq's resistance and Instinet's exclusivity, Levine created Island in 1996. Island was a groundbreaking Electronic Communications Network (ECN) that directly matched buy and sell orders, bypassing human market makers entirely. It was designed to be:

• Lightning-fast
• Inexpensive ($1 per trade initially)
• Supremely transparent, publishing all orders and trades on its website (BookViewer) for free.

Democratizing information. Island's free, real-time data feed (ITCH) was revolutionary, offering unprecedented visibility into market depth and order flow. This challenged the traditional model where market data was expensive and controlled by exchanges, embodying Levine's "information wants to be free" ethos. This transparency attracted a new breed of automated traders.

The maker-taker model. To further boost liquidity, Levine introduced the "maker-taker" model in 1998, paying firms a small rebate for "making" liquidity (posting bids/offers) and charging those who "took" liquidity (executing against existing orders). This incentive structure, later adopted by most exchanges, became a cornerstone of high-frequency trading, though Levine would later regret its unintended consequences.

4. The Emergence of High-Frequency Trading

The Bots were growing, flexing their muscles, and becoming increasingly powerful.

Automated market making. Island's speed and low costs attracted pioneering high-frequency trading (HFT) firms like Automated Trading Desk (ATD), Renaissance Technologies, Tradebot, and Getco. These firms deployed sophisticated algorithms and artificial intelligence (AI) to:

• Automatically post bids and offers
• Execute trades in milliseconds or seconds
• Profit from tiny price differences (scalping)

The need for speed and proximity. HFT firms' strategies demanded extreme speed. Dave Cummings of Tradebot, for instance, realized that physical proximity to Island's servers (colocation) was crucial to overcome latency (the time delay in data transmission). This led to exchanges offering colocation services, where HFT firms paid to place their computers directly next to the exchange's matching engines, creating a significant speed advantage.

Maker-taker's impact. The maker-taker model became a primary driver for HFT, allowing firms to profit from rebates even on razor-thin spreads. This led to:

• Massive order volumes, with thousands of orders per second.
• A focus on capturing fees rather than traditional market making.
• The proliferation of "phantom liquidity" from orders placed and canceled rapidly.

5. Exchanges Adapt or Die

The NYSE for years had boasted of huge investments in technology as it struggled to keep pace with competitors.

Regulatory pressure and ECN growth. The SEC's Order-Handling Rules (1997) and Regulation National Market System (Reg NMS, 2005) forced traditional exchanges to compete with ECNs. Reg NMS, in particular, mandated that orders be routed to the venue with the best price, and allowed "trade-throughs" of slower manual markets, effectively prioritizing speed over the NYSE's traditional floor.

Mergers and modernization. Faced with dwindling market share and the threat of obsolescence, the NYSE and Nasdaq were compelled to acquire their electronic rivals:

• NYSE-Archipelago merger (2006): The NYSE acquired Jerry Putnam's Archipelago, gaining its electronic trading technology and becoming a publicly traded company (NYSE Group). This signaled the end of the NYSE floor's dominance.
• Nasdaq-Island merger (2005): Nasdaq acquired Instinet's trading engine (the former Island system), integrating its superior speed and technology to compete with the NYSE.

The new infrastructure. These mergers led to massive investments in data centers (like NYSE's Project Alpha in Mahwah, NJ) and high-speed connectivity. The old guard of human specialists and market makers rapidly declined, replaced by HFT firms that became "designated market makers" (DMMs) on the NYSE floor, further cementing the dominance of machine trading.

6. The Flash Crash: A Systemic Warning

We saw a living breathing real-time example of the potential catastrophe that could take place.

Unprecedented market chaos. On May 6, 2010, the Dow Jones Industrial Average plunged nearly 1,000 points in minutes, only to rebound just as quickly. This "Flash Crash" exposed the extreme fragility of the HFT-dominated market. Even sophisticated traders like Thomas Peterffy (Timber Hill) and Peter Brown (Renaissance) were baffled by the speed and scale of the event.

Algorithmic feedback loops. The crash was triggered by a large institutional sell order for S&P 500 E-mini futures, which was met by high-frequency traders rapidly buying and selling, creating a vicious feedback loop. As prices plummeted, HFT algorithms, designed to pull out during extreme volatility ("panic ticks"), withdrew liquidity, exacerbating the decline.

Broken market mechanisms. The crash revealed critical flaws:

• Stub quotes: Wildly wide bids/offers posted by HFTs to stay in the market without trading, which became the only available prices for some stocks during the crash.
• Inter-market disconnections: Nasdaq temporarily cut off NYSE Arca due to execution delays, further fragmenting liquidity.
• Phantom liquidity: The realization that HFT's massive order flow could vanish instantly, leaving a vacuum.

7. A Rigged Game? The Debate Over Market Fairness

The financial markets of … the world’s developed countries are at a turning point. Technology, market structure, and new products have evolved more quickly than our capacity to understand or control them.

The cost of speed. The Flash Crash intensified concerns that the market was "rigged" to benefit high-speed traders. Critics like Haim Bodek (Trading Machines) argued that exchanges, in their pursuit of HFT volume, created "toxic order types" and provided speed advantages (colocation, faster data feeds) that allowed HFTs to front-run other investors.

Maker-taker's perverse incentives. Bodek believed that the maker-taker model, combined with intense HFT competition, forced exchanges to subtly favor certain firms to ensure they received rebates. This created a zero-sum game where sophisticated players won by exploiting market microstructure, while ordinary investors paid the "take" fees and suffered from less favorable executions.

Regulatory challenges. Senator Ted Kaufman and others highlighted the SEC's inability to keep pace with technological changes, leading to a "two-tiered market." The Trillium Brokerage Services case, involving "layering" (gunning fake orders to manipulate prices), provided concrete evidence of HFT manipulation, further fueling the debate over whether the market had become a "casino."

8. The Global Algo Wars and AI's Frontier

The end game for Wissner-Gross and Freer had nothing to do with AI Bots and picosecond arbitrage.

Relentless pursuit of speed. Post-Flash Crash, the HFT arms race accelerated globally. New fiber-optic cables were laid to shave milliseconds off transatlantic trades, and microwave technology emerged to achieve even faster speeds. Data centers proliferated worldwide, linking markets in a "push-button money grid" where trading occurred at nanosecond (billionths of a second) speeds.

AI's evolving role. While early AI in trading was limited, new breakthroughs in "Big Data" and machine learning promised to create truly intelligent trading machines. Firms like Kinetic Global Markets and Cerebellum Capital attempted to build AI systems that could:

• Scan vast amounts of data (shipping trends, social media, news) for predictive signals.
• Dynamically adapt strategies to changing market conditions.
• Potentially act as "digital Warren Buffets" for long-term investing.

The human-machine partnership. Despite the rise of autonomous AI, some, like Haim Bodek, explored "Advanced Chess" models, combining human intuition with machine power. However, the overarching trend pointed to a future where AI Bots, operating across a globally interconnected, ultra-fast network, would increasingly control financial markets, raising questions about systemic risk and human oversight.

Last updated: March 22, 2026