Cross-sectional momentum (Jegadeesh and Titman, 1993) ranks stocks by their return over the past 12 months (excluding the most recent month, which exhibits a short-term reversal effect) and buys the top decile while shorting the bottom decile. The long/short portfolio is rebalanced monthly or quarterly. Historical annual alpha: approximately 5-6% in the US, consistent in international developed markets, weaker in emerging markets. The formation period that works best empirically is 12 months back, 1 month skip, 1-month forward performance.

Time-series momentum (Moskowitz, Ooi, and Pedersen, 2012) applies momentum to each asset independently against its own history: go long if the asset's return over the past 12 months is positive, short if negative. This captures momentum independently of cross-sectional rank — even an asset that is the worst performer in its universe might still have positive time-series momentum if its absolute return is positive. Time-series momentum is the basis for most trend-following CTA (commodity trading advisor) strategies across equity indices, bonds, currencies, and commodity futures.

The behavioral explanation for momentum: investors underreact to positive fundamental news initially (anchored to prior earnings expectations, slow to update), then overreact as the news becomes widely understood and analyst revisions arrive. Post-earnings announcement drift (PEAD) is the clearest empirical demonstration — prices continue moving in the earnings surprise direction for weeks after the announcement, showing that the initial reaction was insufficient. Confirmation bias and herding amplify these effects over longer horizons.

The risk-based explanation: momentum is compensation for bearing specific crash risk. Momentum portfolios experience sharp, sudden reversals during market turnarounds — when previously winning stocks reverse sharply and previously losing stocks bounce (short squeezes on the underperformers amplify losses on the short side). This momentum crash is so sharp that it dominates the otherwise strong long-run momentum premium. In April 2009, momentum lost approximately 50% in a single month as the market began recovering from the financial crisis — the worst month for any systematic equity factor in modern market history.

Transaction costs are momentum's primary drag at the implementation level. Monthly rebalancing of a momentum portfolio generates high turnover (~100-200% annually) — the strategy continuously sells stocks that have slowed and buys new leaders. Bid-ask spreads and market impact for less liquid names substantially erode gross momentum returns, particularly in small-cap universes. Reducing turnover (lower rebalancing frequency or higher signal thresholds before trading) reduces transaction costs but also reduces signal freshness.

Volatility-adjusted momentum (dividing momentum signal by the stock's recent volatility) improves the Sharpe Ratio by equalizing momentum bets across different volatility levels — preventing a few highly volatile stocks from dominating the portfolio. Industry-neutral momentum (ranking stocks within industries rather than across all industries) reduces the sector concentration that raw momentum creates, providing more diversified factor exposure. AIQ's momentum factor incorporates risk-adjusted momentum signals rather than raw price return to improve robustness across volatility regimes.