Estimated study time: 60 minutes
Content:
Algorithmic trading uses computer programs to execute securities transactions based on pre-defined rules, typically to achieve specific execution objectives with minimal market impact and transaction cost. At CFA Level 2, algorithmic trading is analyzed within the framework of trade cost analysis, market microstructure, and its implications for portfolio performance. Trade execution quality affects portfolio returns directly — the difference between the decision price and the actual execution price compounds over thousands of trades and can significantly erode the returns of active investment strategies. Understanding execution costs is essential for evaluating whether a strategy's gross alpha exceeds its implementation costs.
Implementation shortfall (also called the arrival price framework) is the primary metric for measuring total execution costs. Implementation shortfall = (Paper portfolio return - Actual portfolio return) = Opportunity cost + Explicit costs + Market impact costs. More formally: IS = [Actual Execution Price - Decision Price] / Decision Price * Portfolio Value, where the decision price is the price at the time the investment decision was made (often the prior day's close or the time of the order decision). IS captures: (1) explicit costs — commissions and fees; (2) market impact — the price moves against the trader as they trade (buying pressure pushes prices up); (3) delay costs — the price moves against the trader before execution begins; and (4) opportunity costs — the cost of orders that were never executed (the stock moved away before trading began).
Algorithmic trading strategies fall into several categories based on their execution objective. VWAP (Volume-Weighted Average Price) algorithms execute trades pro-rata with market volume over a specified period, targeting the VWAP benchmark. A VWAP algorithm minimizes market impact by spreading trades across the day in proportion to expected volume patterns (heavy volume at open and close). TWAP (Time-Weighted Average Price) algorithms execute trades evenly over a time period, independent of volume. TWAP is simpler but less adaptive to volume patterns. Arrival price (IS) algorithms aim to minimize implementation shortfall relative to the decision price — they trade more aggressively early when urgency is high and market conditions are favorable. Participation rate algorithms specify a maximum fraction of market volume (e.g., "participate at most 20% of market volume"), controlling market impact.
Market microstructure concepts underpin algorithmic trading design. The bid-ask spread is the primary source of explicit transaction cost for market orders. The effective spread = 2 * |Trade Price - Midpoint|, which is the realized cost of trading. Market impact is the adverse price movement caused by the trader's own orders — larger orders in less liquid securities experience greater impact. Adverse selection arises because informed traders (who know the true value) trade against market makers, causing systematic losses for passive (limit order) strategies. The Kyle lambda (price impact coefficient) quantifies price impact: delta_price = lambda * signed_order_flow. Dark pools and algorithmic matching networks allow large institutional orders to execute without revealing them to the public market, reducing market impact at the cost of potential adverse selection.
High-frequency trading (HFT) is a specialized form of algorithmic trading characterized by extremely short holding periods (microseconds to milliseconds), very high message rates, co-location at exchange data centers for speed advantages, and revenue from very small per-share margins on very large trading volumes. HFT strategies include market making (providing liquidity in exchange for the spread), statistical arbitrage (exploiting short-lived price discrepancies across related instruments), and latency arbitrage (exploiting stale quotes on slower venues). The CFA curriculum evaluates whether HFT improves or damages market quality. The evidence is mixed: HFT has narrowed bid-ask spreads (improved liquidity for small orders) but may have increased the risk of flash crashes and may extract value from slower institutional investors through latency arbitrage.
Key Terms:
Quiz Questions:
Q1. A portfolio manager decides to buy 100,000 shares of XYZ Corp at 10:00 AM when the market price is $50.00. The order takes 2 hours to execute. At 10:30 AM (before the broker begins executing), the price has moved to $50.50. The 100,000 shares are ultimately purchased at an average price of $51.00. Commissions are $0.05 per share. Calculate the implementation shortfall.
A) IS = (Execution Price - Decision Price + Commission) / Decision Price = ($51.00 - $50.00 + $0.05) / $50.00 = $1.05 / $50.00 = 2.10%. B) IS = (Execution Price - Decision Price) / Decision Price = $1.00 / $50.00 = 2.00% (excluding commissions). C) IS = Commission only = $0.05/$50.00 = 0.10%. D) IS = (Execution Price - Pre-trade Midpoint) = $51.00 - $50.25 = $0.75 per share.
Answer: A — Total implementation shortfall captures all costs versus the decision price: IS = (Avg. Execution Price - Decision Price + Commissions) / Decision Price = ($51.00 - $50.00 + $0.05) / $50.00 = $1.05 / $50.00 = 2.10%. Components: Delay cost = ($50.50 - $50.00)/$50.00 = 1.00% (price moved before execution began); Market impact = ($51.00 - $50.50)/$50.00 = 1.00% (price moved during execution); Commissions = $0.05/$50.00 = 0.10%. Total IS = 2.10%.
---
Q2. A portfolio manager needs to sell 500,000 shares of a stock that trades an average daily volume (ADV) of 2,000,000 shares. She has three algorithm choices: (A) VWAP over 1 day; (B) TWAP over 2 days; (C) Participation rate of 10% of volume over several days. Given the order represents 25% of ADV, which algorithm best balances urgency and market impact minimization?
A) VWAP over 1 day — executing 25% of ADV in one day is aggressive and will cause significant market impact due to the large daily participation rate. B) A participation rate algorithm at 10% of market volume over approximately 2.5 days of trading provides a reasonable balance: it limits the manager's participation to 10% of each period's volume (minimizing market impact) while completing the order within a manageable timeframe. C) TWAP over 2 days — fixed-time execution regardless of volume may create market impact when volume is thin. D) All algorithms are equally appropriate for this order size.
Answer: B — For a large order (25% of ADV), minimizing market impact by spreading the trade over multiple days at a controlled participation rate is typically optimal. A 10% participation rate algorithm: each day the algorithm trades at most 10% of market volume, taking approximately 2.5 days to complete the full order (500K / (10% * 2M ADV per day) = 2.5 days). This controls market impact significantly better than a single-day VWAP (which would represent 25% of daily volume — very aggressive). TWAP is suboptimal because it executes the same amount per time period regardless of volume, potentially creating high impact during low-volume periods.
---
Q3. An institutional trader observes that whenever his firm starts buying a thinly traded stock, the price rises 3-5% before the full order is complete. This phenomenon is best described as:
A) Alpha decay — the investment thesis becomes stale during execution. B) Market impact — the trader's own order flow pushes the price up against him, increasing the cost of completing the remaining position. C) Adverse selection — other traders know the stock's true value and are selling it at inflated prices. D) Bid-ask spread widening due to increased trading activity.
Answer: B — Market impact is the adverse price movement caused by the trader's own order flow. When a buyer is actively purchasing a large quantity, supply and demand dynamics push the price up — sellers require higher prices to part with shares as the buying pressure is sustained. This is the primary source of implicit transaction cost for large institutional orders. The trader's own activity reveals information about demand and causes prices to move against the direction of the trade. Mitigating market impact requires spreading orders over time and across venues.
---
Q4. A portfolio manager is evaluating execution quality for her equity trades over the past year. The VWAP benchmark performance shows that the desk has beaten VWAP by an average of 2 bps per trade. The implementation shortfall analysis shows an average IS of 45 bps per trade. Which metric provides a more complete measure of execution quality, and why?
A) VWAP performance (beating by 2 bps) is a better metric; it reflects real-time benchmark comparison. B) Implementation shortfall (45 bps) is a more complete measure because it captures the full cost relative to the decision price — including delays before trading begins, market impact during execution, and opportunity costs of unexecuted orders. VWAP performance measures execution quality only for the portion that was traded, ignoring delay costs and opportunity costs. C) Both metrics are equivalent and should be averaged for the true cost. D) Beating VWAP by 2 bps proves execution was excellent; IS of 45 bps is irrelevant.
Answer: B — VWAP is a common but limited execution benchmark because it only compares the execution price to the day's average price — it says nothing about the cost of delay before execution started or what happened to unfilled portions of the order. A desk can easily beat VWAP by deferring difficult trades until the end of the day or by prioritizing easy trades, which results in excellent VWAP performance but poor IS (because the initial delay and missed opportunities are costly). Implementation shortfall is the more economically complete measure because it captures all costs relative to the investment decision price.
---
Q5. A market maker in equity options quotes a bid of $2.50 and an offer of $2.60 on a call option. An investor purchases the call at $2.60. The midpoint is $2.55. What is the effective spread paid by the investor, and what does the bid-ask spread imply about the market maker's expected revenue per transaction?
A) Effective spread = 2 * ($2.60 - $2.55) = 2 * $0.05 = $0.10 per option. The market maker earns approximately half the quoted spread ($0.05) per completed transaction if the midpoint is the true fair value. B) Effective spread = $2.60 - $2.50 = $0.10 (quoted spread, not effective spread). C) Effective spread = $2.60 - $2.55 = $0.05 (one-sided spread, not doubled). D) The market maker earns the full $0.10 spread per transaction regardless of subsequent trades.
Answer: A — Effective spread = 2 * |Trade Price - Midpoint| = 2 * |$2.60 - $2.55| = 2 * $0.05 = $0.10. This equals the quoted spread in this example (because the trade occurred at the ask price exactly). The market maker's expected revenue per completed round-trip (buy and then sell, or vice versa) is approximately the full spread ($0.10 = $2.60 ask - $2.50 bid), or about half per transaction side. However, adverse selection risk reduces the market maker's effective realized spread — if the investor buying the call at $2.60 is informed and the option subsequently moves to $3.00, the market maker loses more than the spread revenue gained.
---