Based on Michael Lewis, Flash Boys: A Wall Street Revolt (2014)
Michael Lewis's Flash Boys (2014) tells the story of how the American stock market — the supposed exemplar of free-market capitalism and fair competition — was systematically rigged by high-frequency trading (HFT) firms that exploited microsecond speed advantages to extract billions of dollars from ordinary investors, pension funds, and mutual funds.
The book centers on Brad Katsuyama, a Canadian trader at Royal Bank of Canada (RBC) who discovered that something strange was happening to his stock orders. Every time he tried to buy shares, the stock seemed to move away from him just before his order arrived. What he eventually uncovered was a vast, hidden infrastructure of speed-based predatory trading that fundamentally changed how markets worked — and not for the better.
Flash Boys is not primarily an investing book — it is an investigative work about market structure. But for any investor who buys or sells stocks, it contains essential knowledge about:
The core discovery is that the US stock market, fragmented across 13 public exchanges and dozens of dark pools, created an environment where firms with faster connections could see your order arrive at one exchange and race ahead to other exchanges to trade against you before your order got there. This was not hypothetical — it was happening millions of times per day, on virtually every trade executed in the US equity market.
| Character | Role |
|---|---|
| Brad Katsuyama | RBC trader who uncovered the HFT problem |
| Ronan Ryan | Telecom expert who mapped the physical infrastructure |
| Rob Park | Programmer who built the tools to understand HFT |
| Sergey Aleynikov | Goldman Sachs programmer prosecuted for code theft |
| Haim Bodek | Former HFT trader who exposed hidden order types |
| Rich Gates | Asset manager who refused to accept the rigged system |
The book opens with the story of Spread Networks, a company that spent $300 million to lay a single fiber optic cable in the straightest possible line between Chicago (where futures trade) and New Jersey (where stock exchanges have their data centers). The purpose: to shave three milliseconds off the round-trip communication time between the two financial centers.
Three milliseconds. $300 million. This investment was rational because high-frequency trading firms would pay enormous fees for a three-millisecond advantage over their competitors.
SPEED EVOLUTION IN FINANCIAL MARKETS:
1990s: Human traders on exchange floors
Execution speed: minutes
2000s: Electronic trading, co-location begins
Execution speed: seconds → milliseconds
2005-2010: HFT arms race intensifies
Execution speed: milliseconds → microseconds
2010-2014: Microwave towers, hollow-core fiber
Execution speed: microseconds → nanoseconds
Cost of being 1 millisecond slower than competitors:
Estimated at $100 million per year in lost trading profits
This arms race consumed enormous resources to produce
ZERO value for the broader economy or for investors.
Exchanges began offering "co-location" services — allowing trading firms to place their computers physically inside the exchange's data center. The closer your server was to the exchange's matching engine, the faster you could receive market data and send orders. Exchanges charged premium prices for cabinets closer to the matching engine — sometimes measuring distances in feet.
After Spread Networks' fiber optic cable, the next frontier was microwave transmission. Microwave signals travel through air at nearly the speed of light, which is faster than light through fiber optic cable (light slows down in glass). Firms built chains of microwave towers between Chicago and New Jersey, shaving additional microseconds off communication times.
Lewis identifies three primary strategies used by HFT firms:
Strategy 1: Electronic Front-Running
When an investor sends a large order to buy shares across multiple exchanges, the HFT firm detects the order arriving at the first exchange and races to buy shares at the other exchanges before the investor's order arrives. The HFT firm then sells those shares to the investor at a slightly higher price.
ELECTRONIC FRONT-RUNNING EXAMPLE:
Investor wants to buy 10,000 shares of XYZ at $50.00
XYZ is offered at $50.00 across five exchanges, 2,000 shares each
WITHOUT HFT:
Order goes to all 5 exchanges simultaneously
Investor buys 10,000 shares at $50.00
Total cost: $500,000
WITH HFT:
Order reaches Exchange A first (by microseconds)
HFT firm sees the buy order at Exchange A
HFT races to Exchanges B, C, D, E and buys 8,000 shares at $50.00
HFT offers those 8,000 shares back at $50.02
Investor buys:
2,000 at $50.00 (Exchange A)
8,000 at $50.02 (from HFT at other exchanges)
Total cost: $500,160
HFT profit: $160 (on this single trade)
Multiply by millions of trades per day = billions per year
Strategy 2: Rebate Arbitrage
Exchanges pay rebates to firms that provide liquidity (post limit orders) and charge fees to firms that take liquidity (send market orders). HFT firms exploit differences in rebate structures across exchanges, posting and canceling orders at enormous speed to capture rebates without taking meaningful risk.
Strategy 3: Slow-Market Arbitrage
When a stock's price changes on one exchange, there is a brief moment before the other exchanges reflect the new price. HFT firms exploit this latency by buying at the old (stale) price on the slower exchange and selling at the new price on the faster exchange, or vice versa.
SLOW-MARKET ARBITRAGE EXAMPLE:
Time 0: XYZ trades at $50.00 on all exchanges
Time +1 microsecond: Large buy order arrives at Exchange A
XYZ price moves to $50.05 on Exchange A
Time +1 to +100 microseconds:
Other exchanges still show $50.00 (stale price)
HFT firm sees the new price at Exchange A
HFT buys at $50.00 on Exchanges B, C, D, E
HFT sells at $50.05 on Exchange A (or waits for other exchanges to update)
Risk-free profit: $0.05 per share × thousands of shares = significant money
Duration of opportunity: less than 100 microseconds
This happens continuously, all day, every day.
By 2014, HFT firms accounted for approximately 50% of all US stock market trading volume. Major firms included Citadel Securities, Virtu Financial, Jump Trading, Tower Research Capital, and others. Virtu Financial famously disclosed that it had only one losing trading day in 1,238 trading days — a statistical near-impossibility that reflects the structural nature of their advantage rather than any traditional form of skill.
Dark pools are private trading venues operated by broker-dealers (usually large banks) where buy and sell orders are matched without publicly displaying the order book. They were originally created to allow large institutional investors to trade big blocks of stock without revealing their intentions to the broader market.
Lewis reveals that by the early 2010s, dark pools had become the opposite of their original purpose. Instead of protecting institutional investors, many dark pools had become hunting grounds for HFT firms:
THE DARK POOL PROBLEM:
ORIGINAL INTENT:
Institutional investor wants to sell 1 million shares
Posts order in dark pool where it is hidden from the market
Another institution's buy order matches in the dark pool
Trade executed at fair price, without market impact
WHAT ACTUALLY HAPPENED:
Institutional investor posts sell order in bank's dark pool
Bank's dark pool is filled with HFT firms (who pay the bank for access)
HFT firm detects the institutional sell order
HFT front-runs the order, trading against the institution
The institution gets a WORSE price than if they had traded on public exchanges
The bank profits by selling access to HFT firms
The HFT firms profit by trading against the institutions
The institutions lose on every trade
The banks operating dark pools had massive conflicts of interest:
Dark pools reported almost no meaningful data about their operations. Investors could not determine:
Lewis describes front-running in the HFT context as fundamentally different from traditional front-running (where a broker trades ahead of a client's order). In the HFT version:
Whether this constitutes front-running in the legal sense is debated. Lewis argues that the economic effect is identical: the HFT firm profits at the investor's expense by trading ahead of an order it knows is coming.
HFT firms sometimes flood exchanges with enormous numbers of orders that are immediately canceled — a practice called "quote stuffing." This achieves multiple purposes:
HFT firms place large orders they intend to cancel before execution, creating false impressions of supply or demand. A firm might place a large sell order to push the price down, buy at the lower price, then cancel the sell order — all within milliseconds.
THE DAILY HFT TAX ON INVESTORS:
Conservative estimate of latency arbitrage costs:
Average trade: $0.01-0.02 per share extracted by HFT
Average daily US equity volume: ~7 billion shares
Daily HFT extraction: $70M-$140M
Annual HFT extraction: $17B-$35B
This is a hidden tax that appears nowhere on any trade confirmation.
The investor sees only their executed price, not the price they
SHOULD have gotten without HFT interference.
Who pays this tax?
- Mutual fund investors (through worse execution in the fund)
- Pension fund beneficiaries
- 401(k) holders
- Individual investors placing market orders
- Anyone who trades stocks
The regulatory backdrop for the entire HFT problem was Regulation NMS (National Market System), implemented by the SEC in 2007. Reg NMS was intended to ensure that investors received the best available price across all exchanges. In practice, it fragmented the market across 13 exchanges and dozens of dark pools, creating the exact complexity that HFT firms exploit.
Before Reg NMS, most trading occurred on the NYSE or NASDAQ. After Reg NMS, an order to buy 10,000 shares might need to be routed to 10 different venues simultaneously. The time it takes for orders to reach different venues creates the latency gaps that HFT firms exploit.
MARKET FRAGMENTATION:
PUBLIC EXCHANGES (13):
NYSE, NASDAQ, BATS (now Cboe BZX), BATS Y, Direct Edge A,
Direct Edge X, NYSE Arca, NYSE MKT, NASDAQ BX, NASDAQ PSX,
IEX, CHX, and others
DARK POOLS (40+):
Credit Suisse CrossFinder, Goldman Sigma X, Morgan Stanley MS Pool,
Barclays LX, UBS ATS, and dozens more
INTERNALIZERS:
Broker-dealers who execute orders internally before routing
to exchanges (e.g., Citadel Securities, Virtu Financial)
An investor's order must navigate this maze to find the best price.
At every step, HFT firms are positioned to intercept and profit.
The complexity of the modern market structure is itself a tax on investors. The more complex the system, the more opportunities for sophisticated players to exploit less sophisticated ones. Simplicity would benefit investors but would reduce profits for exchanges, HFT firms, and broker-dealers.
Lewis traces the journey of a typical retail investor's order:
THE JOURNEY OF A RETAIL STOCK ORDER:
Step 1: You click "Buy 100 shares of AAPL at market" in your brokerage app
Step 2: Your broker does NOT send the order to an exchange.
Instead, it sells the order to an "internalizer" (a market maker
like Citadel Securities or Virtu Financial). The broker receives
payment — typically $0.001-0.004 per share — for this order flow.
Step 3: The internalizer executes the trade at a price that is
slightly better than the best exchange price (by a fraction of a cent).
This "price improvement" justifies the arrangement legally.
Step 4: The internalizer profits by:
a) Trading at a spread wider than they could achieve on exchanges
b) Using the information from your order to inform their own trading
c) Harvesting the "alpha" embedded in order flow patterns
RESULT:
You get a price that is technically "better" than the exchange price
by perhaps $0.001 per share. But the full picture is more complex:
the internalizer captures $0.01+ per share in economic value from
your order. You receive a tiny fraction of this as "price improvement."
THE JOURNEY OF AN INSTITUTIONAL STOCK ORDER:
Step 1: A mutual fund manager decides to buy 500,000 shares of XYZ
Step 2: The fund's trading desk breaks the order into smaller pieces
(algorithms chop it into 100-500 share slices) to minimize market impact
Step 3: Each slice is routed to various exchanges and dark pools
based on the algorithm's logic
Step 4: At the first exchange, HFT firms detect the slice:
- They recognize the algorithmic pattern
- They predict that more buy orders are coming
- They race to buy shares at other venues before the algo's orders arrive
Step 5: By the time the fund's algorithm has executed 500,000 shares,
the price has moved against it by $0.02-0.10 per share
Step 6: Total cost of HFT interference:
500,000 shares × $0.05 average slippage = $25,000 on a single trade
A large fund executing thousands of such trades per year:
$25,000 × 1,000 trades = $25 million per year in hidden costs
These costs are ultimately borne by the fund's investors.
Lewis exposes the payment for order flow (PFOF) model as a fundamental conflict of interest. Brokers like Robinhood, TD Ameritrade, and others receive billions of dollars annually from market makers in exchange for routing customer orders to them. The broker's incentive is to route orders where it receives the highest payment, not where the customer gets the best execution.
VISIBLE TRADING COSTS:
Commission: $0-$10 per trade (many brokers now $0)
Regulatory fees: Fractions of a penny
Total visible: ~$0-$10 per trade
INVISIBLE TRADING COSTS:
Bid-ask spread: $0.01-$0.05 per share
Market impact: $0.01-$0.10 per share (for larger orders)
HFT extraction: $0.01-$0.02 per share
Adverse selection: Variable
Opportunity cost of failed executions: Variable
Total invisible on a 1,000-share order: $30-$170
The invisible costs are 10-100x the visible costs.
"Commission-free" trading is not free — the costs are hidden.
Every order you place reveals information to the market. A market order reveals that you want to trade immediately and are willing to accept the current price. A limit order reveals your price expectations. A large order reveals directional conviction. HFT firms aggregate this information in real time and trade on it.
Much of the liquidity displayed on exchange order books is not real — it is posted by HFT firms that will cancel the order before it can be executed if market conditions change. Lewis describes situations where displayed liquidity of 50,000 shares evaporated to 2,000 shares the instant someone tried to trade against it.
After discovering how HFT firms exploited speed advantages, Katsuyama first built a tool called "Thor" at RBC that sent orders to all exchanges simultaneously, timed to arrive at the same moment. This eliminated the latency advantage that HFT firms exploited. Thor was effective, but it was only available to RBC clients.
Katsuyama left RBC to found the Investors Exchange (IEX) — a stock exchange designed to be fair for all participants by neutralizing speed advantages.
IEX's key innovation was a 350-microsecond delay (the "speed bump") imposed on all incoming orders. This delay was long enough to prevent HFT firms from exploiting latency differences between exchanges but short enough to be imperceptible to human traders.
HOW THE IEX SPEED BUMP WORKS:
Traditional exchange:
HFT order reaches exchange in 50 microseconds
Investor order reaches exchange in 200 microseconds
HFT has 150 microsecond advantage → can front-run
IEX exchange:
HFT order reaches exchange in 50 microseconds
+ 350 microsecond speed bump = effective arrival at 400 microseconds
Investor order reaches exchange in 200 microseconds
+ 350 microsecond speed bump = effective arrival at 550 microseconds
HFT still arrives first, but the speed bump neutralizes the
latency advantage they use for predatory strategies because
IEX's own pricing is determined by a system that also has the
350 microsecond delay — so the HFT firm cannot act on stale
quotes from other exchanges.
PHYSICAL IMPLEMENTATION:
38 miles of fiber optic cable coiled in a box
Light travels at ~124 miles per millisecond through fiber
38 miles / 124 miles per ms = ~0.35 ms = 350 microseconds
| Feature | Traditional Exchange | IEX |
|---|---|---|
| Speed advantage | Rewarded through co-location | Neutralized by speed bump |
| Order types | 100+ complex order types that benefit HFT | Simple, transparent order types |
| Rebates | Maker-taker model creating perverse incentives | No rebates (flat fee) |
| Dark pool access | HFT firms included | No HFT predatory activity |
| Revenue model | Sell speed, data, and complexity | Simple transaction fees |
IEX faced enormous opposition from HFT firms, established exchanges, and broker-dealers that profited from the existing system. Despite this, it attracted significant volume from institutional investors and eventually won approval as a registered national securities exchange in 2016.
For most individual investors making long-term investments, HFT costs are relatively small in the context of long-term returns. If you are buying a stock you plan to hold for 5 years and your total return is 50%, the $0.02 per share lost to HFT on entry and exit is negligible.
For investors who trade frequently, use market orders on volatile stocks, or invest through actively managed funds with high turnover, HFT costs accumulate significantly. A fund that turns over its portfolio once per year incurs HFT costs on every trade, and those costs are deducted from your returns.
MARKET MICROSTRUCTURE LESSONS FOR INDIVIDUAL INVESTORS:
1. "Commission-free" trading is not free
Your broker sells your order flow to market makers
The market maker profits from your trade
You bear hidden costs through worse execution
2. Market orders are the most expensive order type
They guarantee execution but not price
They reveal your urgency, which is exploited
Use limit orders whenever possible
3. The price you see is not always the price you get
Displayed quotes can evaporate before your order arrives
The more volatile the stock, the greater the slippage risk
4. Dark pools may not be working in your interest
Your broker may route your order to their own dark pool
The dark pool may be filled with predatory HFT firms
5. Larger orders are more expensive (per share) than smaller ones
Market impact increases with order size
Breaking orders into pieces helps but is not foolproof
6. Trading at market open and close is more expensive
Volatility and HFT activity peak at these times
Mid-day trading typically has lower execution costs
STRATEGY 1: USE LIMIT ORDERS
Instead of: "Buy 500 shares of XYZ at market"
Use: "Buy 500 shares of XYZ at limit $50.10"
This sets a maximum price you will pay.
It prevents being filled at an artificially high price.
The risk: your order may not execute if the price moves away.
STRATEGY 2: TRADE LESS FREQUENTLY
Every trade incurs HFT costs.
Long-term investors who trade once per year pay HFT costs once.
Active traders who trade daily pay HFT costs 250 times per year.
The best defense against HFT is infrequent trading.
STRATEGY 3: AVOID VOLATILE MOMENTS
Market open (9:30-10:00 AM): Wide spreads, high volatility
Market close (3:30-4:00 PM): Heavy activity, complex dynamics
News events: Spreads widen, HFT activity spikes
Best time to trade: Mid-day, when spreads are tightest
STRATEGY 4: USE BROKERS WITH GOOD EXECUTION QUALITY
Compare execution quality reports (Rule 606 reports)
Look for: Price improvement per share, fill rates
Consider brokers that route to IEX or other fair venues
STRATEGY 5: INVEST IN INDEX FUNDS
Index funds trade massive blocks with institutional tools
They negotiate better execution than individuals
Their low turnover minimizes total HFT costs
Vanguard, in particular, is known for execution quality
STRATEGY 6: USE ICEBERG/RESERVE ORDERS (for larger orders)
Display only a portion of your total order
As the displayed portion fills, more is automatically shown
This reduces information leakage to HFT algorithms
For fund managers and institutional traders:
The fundamental lesson from Flash Boys is that market structure matters. The rules of the game — how exchanges operate, how orders are routed, who has speed advantages — determine outcomes. Individual investors cannot change market structure, but they can adapt their behavior to minimize the costs it imposes.
"The United States stock market, the most iconic market in global capitalism, is rigged."
"The average investor had no idea that when he or she placed an order, it was being front-run by high-frequency traders who had paid for the right to see it first."
"The value of speed had been created, out of thin air, by the exchanges — which, having been converted from not-for-profit utilities to for-profit enterprises, now had an incentive to sell it."
"The stock market is no longer a place where investors are matched directly with companies seeking capital. It is a place where speculators are matched with other speculators, and the interests of the original investors are secondary."
"Dark pools were created to give institutional investors a safe place to trade without being front-run. They became the opposite — a place where institutional investors were routinely front-run."
"The best way to rob a bank is to own one. The best way to rob an investor is to own an exchange."
"The speed advantage wasn't a natural advantage like superior intelligence or better information. It was a bought advantage, pure and simple."
"Brad Katsuyama asked a question that should have occurred to a lot of people before him: If the market is supposed to be fair, why does it consistently move against you the instant you try to trade?"
"The problem wasn't that the market was complex. The problem was that complexity was profitable — for everyone except the investor."
"Fifty percent of all stock market trades were now being made by high-frequency trading firms that held positions for an average of not years, not months, not days, not hours, not minutes — but for fractions of a second."
"The IEX exchange was built on a simple idea: the stock market should work for investors, not against them."
"The single most important thing an individual investor can do to reduce the impact of HFT on their portfolio is to trade less."
Flash Boys exposed the hidden infrastructure of predatory high-frequency trading and forced a public reckoning about market fairness. For individual investors, the practical lessons are clear: use limit orders, trade infrequently, invest in low-turnover index funds or concentrated value portfolios, and understand that "commission-free" trading comes with hidden costs. The best defense against a rigged market structure is to interact with it as little as possible — which, conveniently, is also the best strategy for long-term investment returns.