Overview of Quantitative Trading Strategies › Categories of Quantitative Trading Strategies

Quantitative strategies use many approaches across market segments and holding periods. Understanding these categories provides a framework for thinking about how different strategies work and what skills they require.

Statistical arbitrage strategies identify assets whose prices have diverged from their historical or predicted relationships, betting that prices will revert to normal levels. The classic example is pairs trading: if two historically correlated stocks diverge significantly, you short the outperformer and buy the underperformer, profiting when the relationship normalizes.

The core assumption underlying mean reversion strategies is that price deviations are temporary and driven by noise rather than fundamental changes. This works best in markets where:

• Assets have strong economic linkages (two oil companies, a stock and its ADR)
• Arbitrage mechanisms exist to enforce pricing relationships
• Deviations are caused by temporary factors like order flow imbalances

Mean reversion strategies typically have holding periods ranging from days to weeks and require sophisticated risk management because the assumption of reversion can fail catastrophically when regime changes occur.

Trend following strategies take the opposite view: rather than betting on reversion, they bet that existing price trends will continue. As we observed in Part III, Chapter 1 on the stylized facts of financial returns, asset returns exhibit positive autocorrelation at medium-term horizons, an anomaly that contradicts weak-form market efficiency.

Momentum strategies identify assets that have performed well (or poorly) over recent periods and bet that the trend continues. The economic explanations include:

• Underreaction: Investors initially underreact to new information, causing prices to adjust gradually
• Behavioral feedback: Rising prices attract more buyers, creating self-reinforcing cycles
• Risk-based explanations: Momentum returns may compensate for crash risk during trend reversals

Trend following strategies tend to have longer holding periods (weeks to months) and can be applied across asset classes. They also tend to perform well during market crises when trends become extreme, and gives useful diversification.

Factor investing systematically captures returns associated with characteristics like value, quality, momentum, and low volatility. As we discussed in Part IV, Chapter 3, these factors have historically delivered positive risk-adjusted returns.

Long/short equity strategies extend this concept by constructing portfolios that are long stocks with favorable factor exposures and short stocks with unfavorable exposures. The resulting portfolio has limited net market exposure but concentrated factor exposure, aiming to generate returns from stock selection rather than market direction.

Building on our coverage of factor models, a typical long/short equity portfolio might target a return structure that separates stock selection skill from market direction. The return of such a portfolio can be expressed as:

$$r_{\text{portfolio}} = \alpha + \beta_{\text{market}} \cdot r_{\text{market}} + \beta_{\text{value}} \cdot f_{\text{value}} + \beta_{\text{momentum}} \cdot f_{\text{momentum}} + \epsilon$$

This equation reveals the multiple dimensions of return generation in a long/short strategy:

• $r_{\text{portfolio}}$: return of the strategy, the total performance we observe
• $\alpha$: excess return generated by stock selection. This is the value added by choosing specific stocks within each factor group.
• $\beta_{\text{market}}$: exposure to the broad market (targeted $\approx 0$). By maintaining near-zero market beta, the strategy aims to be immune to overall market direction.
• $r_{\text{market}}$: market return
• $\beta_{\text{value}}, \beta_{\text{momentum}}$: exposures to factor risk premia (targeted $> 0$). These positive exposures mean the strategy is intentionally harvesting the value and momentum premiums.
• $f_{\text{value}}, f_{\text{momentum}}$: returns of the value and momentum factors
• $\epsilon$: residual noise

Volatility strategies exploit discrepancies in how volatility is priced across instruments or time periods. As we explored extensively in Part III on options pricing, implied volatility often differs from realized volatility, and the volatility surface exhibits persistent patterns.

Common volatility strategies include:

• Variance risk premium harvesting: Selling options to capture the tendency for implied volatility to exceed realized volatility
• Volatility surface arbitrage: Exploiting inconsistencies across strikes and expirations
• Dispersion trading: Trading the relationship between index volatility and single-stock volatility

These strategies require deep understanding of options mechanics and the Greeks we covered in Part III, Chapter 7.

Market makers provide liquidity by continuously offering to buy and sell assets, earning the bid-ask spread as compensation. This is fundamentally different from directional trading; the goal is not to predict price movements but to facilitate trading and manage inventory risk.

Market making strategies require:

• Fast execution capabilities
• Sophisticated inventory management
• Real-time risk monitoring
• Understanding of order flow dynamics

The returns from market making come not from alpha in the traditional sense but from providing a valuable service to other market participants.

High-frequency trading (HFT) operates on timeframes measured in microseconds to seconds, requiring specialized technology infrastructure. HFT strategies include:

• Latency arbitrage: Exploiting speed advantages to capture price discrepancies across venues
• Electronic market making: Providing liquidity at high speed
• Statistical patterns: Detecting very short-term predictable patterns in order flow

HFT is technologically intensive and has become increasingly competitive, with returns driven more by infrastructure investment than model sophistication.