How Asset Classes Behave
Systematic trading starts from an observation that different asset classes have structurally different drivers. A strategy that works for equities fails for commodities not because markets are different in principle, but because the underlying fundamentals and participant behaviors differ. Understanding those differences is prerequisite to designing strategies that actually hold up.
Cash Equities
The dominant strategy considerations are sector and cap-size rotation, stop-loss margin sizing, and volatility sentiment analysis. Corporate events (earnings, buybacks, dividend changes) create momentum and mean-reversion signals that are distinct from the baseline.
Several behavioral biases are structurally exploitable:
- Lottery effect: investors systematically overpay for high-volatility stocks hoping for outsized returns. This creates a return premium for low-volatility stocks over long holding periods.
- Representativeness bias: investors extrapolate from a few visible winners (the AMDs, the Nvidias) to conclude all high-volatility stocks are likely to perform similarly. They aren’t.
- Asymmetric crisis behavior: low-volatility stocks underperform during the late phase of a bull market (when risk appetite is highest) but dramatically outperform during crises. The strategy is positive-carry most of the time with occasional sharp drawdowns.
Fixed Income
Bond-specific factors: yield-to-maturity calculation, credit rating dynamics, interest rate duration risk, and the structural differences between government bonds, investment-grade corporate, and high-yield bonds. The primary trading signal is yield spread movement against macro rate expectations.
Duration risk is the central variable. A portfolio of long-maturity bonds with high duration has high interest rate sensitivity — a 1% rate rise translates into significant mark-to-market losses even if the bonds never default. Managing duration exposure is as important as managing credit exposure.
Derivatives
Key considerations: time value decay as the primary headwind for options buyers, gamma/delta relationships that change behavior near expiry, open interest analysis as a positioning signal (where is the market crowded?), and expiration-week volatility patterns.
Options markets are dominated by two opposing forces: buyers paying for insurance and sellers collecting premium. The sustainable edge for systematic traders is in the pricing of volatility — either finding situations where implied volatility is structurally mispriced relative to realized volatility, or exploiting the term structure and skew patterns that arise from institutional demand for protection.
Commodities
Demand-supply is the primary driver, but the signal is dominated by noise at short horizons. What adds edge: transportation and storage cost seasonality (contango/backwardation dynamics), geopolitical supply risk (which has fat tails — low probability, high impact), and the link between commodity cycles and macroeconomic phases.
The commodity-macro relationship runs bidirectionally: commodity prices affect inflation (and thus central bank policy), while monetary policy cycles affect commodity demand through industrial activity.
Forex
Macroeconomic policy is the dominant driver at medium-to-long horizons. Political tension and economic calendar releases (CPI, unemployment, PMI) create short-term volatility. The established correlations between oil, gold, and USD create a natural factor structure: oil USD → petrodollar flows; gold USD → risk-off demand.
Carry trades (borrowing in low-rate currencies, investing in high-rate currencies) are the most well-documented systematic strategy in forex — and also the most subject to crash risk, as carry unwinds tend to be sudden and severe.
Algorithmic Trading Paradigms
Momentum and Trend Following
Price momentum persists across asset classes over medium horizons (1–12 months). The mechanism is disputed — behavioral (anchoring, herding) vs. risk-based (compensation for bearing crash risk) — but the empirical pattern is robust. Trend following has historically provided positive returns with low correlation to equities, making it valuable as a portfolio diversifier.
Implementation: identify trend direction (moving average crossovers, breakout systems), size positions based on volatility targeting, and apply stop-losses to limit drawdown.
Value and Mean Reversion
Assets that have become expensive relative to fundamentals tend to revert. This works at both the individual asset level (pairs trading — two correlated assets that have diverged) and at the portfolio level (value factor in equities).
The challenge: the signal-to-noise ratio is low, reversion timing is uncertain, and value trades can lose money for years before working (the “value is dead” problem of 2017–2019).
Statistical Arbitrage
Pairs trading and cointegration: if two assets share a common stochastic trend, their spread is stationary and can be traded. The classic example is two companies in the same sector — their prices move together in the long run but diverge short-term.
Implementation requires: identifying cointegrated pairs (Engle-Granger or Johansen test), estimating the hedge ratio, and setting entry/exit thresholds based on the spread’s historical distribution.
Market Making
Market makers provide liquidity by quoting both bid and ask, capturing the spread as compensation. The inventory risk comes from accumulating unwanted directional exposure as the market moves against the quotes.
Inventory risk model: the market maker adjusts quotes based on current inventory and risk appetite. With long inventory, shift quotes down (bid lower, ask lower) to attract selling. The optimal quote skew is proportional to inventory size, spread parameter, and remaining trading horizon.
Adverse selection model: adjust quotes based on the probability that a counterparty is informed (i.e., knows something you don’t). A large order against your quote in a stock with high short interest is more likely to be informed than a random small retail order. The market maker widens the spread when adverse selection risk is elevated.
Both models arrive at the same practical intuition: wider spreads when risk is high (whether inventory risk or information asymmetry), tighter spreads when conditions are favorable.
High-Frequency Microstructure
At the fastest time scales, the dominant dynamics are order book microstructure: queue priority, latency, and the order flow imbalance signal.
Order flow imbalance (OFI): the imbalance between buyer-initiated and seller-initiated trades is predictive of short-term price direction. A market experiencing sustained buying pressure (OFI persistently positive) tends to move up in the next few milliseconds to seconds.
Queue position: in a first-come-first-served exchange, queue position at the best bid/ask determines whether a limit order gets filled when the market touches that price. HFT strategies for improving queue position include passive order placement at the optimal level based on predicted queue dynamics.
Latency: at the HFT level, latency is the primary competitive advantage. Co-location (placing servers physically adjacent to the exchange matching engine), specialized network hardware, and kernel bypass networking are standard requirements. The latency arms race has largely run its course in mature markets — the marginal benefit of further latency reduction is small.
See Quantitative Trading Playbook and The Trader Mindset for the practitioner-facing synthesis of these ideas.