Momentum in stocks is an effect of rising stock prices entering a seemingly self-fulfilling virtuous cycle in which past returns are indicative of the future. Momentum is one of the oldest and most popular and studied trading strategies. There are two ways to compose momentum investment portfolio. One is time series momentum also called trend following, where particular trend may even concern the whole market during certain period of time such as US equity market during 2008-2018. The other is cross-sectional momentum that aims to target spread between winners and losers while keeping low overall market exposure. A classic equity momentum strategy is to construct a portfolio of best and worst performing stocks according to returns in the past month. We buy the top 10% of best stocks and short the 10% of worst performing stocks. Due to momentum effect, the past winners and losers have higher chance of continuing in their trend rather than changing course. However, in an arbitrage free market, a good past performance should not lead to a future price appreciation any more than a poor past performance. In other words, there should not be any positive (or negative) serial correlation in returns. Under the efficient market hypothesis, returns are said to be martingales in which future returns do not depend on the past realizations of prices. Momentum effect has been documented across different time periods and asset classes and it is now an established market anomaly.

Forming portfolios based on past three- to 12-month returns Jagadeesh and Titman (1993) find that momentum strategy does well for first three years after the portfolio formation, however tends to reverse significantly in the 3 to 5 year period. This was very good news for the momentum strategy. Conrad and Kaul (1998) challenged this result and claimed that momentum returns are brought about by the cross-sectional variation in portfolio stocks rather than a persistent trend. This result was later challenged and rebutted by Jagadeesh and Titman (2001). In addition, Fama and French (1996) show that a three-factor model of returns fails to explain intermediate-horizon price momentum. Furthermore, Chan, Jegadeesh, and Lakonishok (1996) show that intermediate-horizon return continuation can be partially explained by under-reaction to earnings news but that price momentum is not subsumed by earnings momentum. Rouwenhorst (1998) finds a similar pattern of intermediate-horizon price momentum in 12 other countries, suggesting that the effect is not likely due to a data snooping bias. Moreover, Moskowitz and Grinblatt (1999) find that momentum profits are mostly related to industry momentum.

Hong and Stein (1997) bring a key finding to the momentum strategy inquiry. They propose an information dissipation model which shows that analysts following only part of the market lead to a slow dissemination of good news that is then exploited by momentum traders. This initial under-reaction attracts arbitrage capital which eventually leads to mal-investment and overreaction at long horizon leading to reversal effect. Moreover, Hong, Terence and Stein (2000) find that momentum is most effective for less covered small-cap stocks and that analyst coverage reception is more pronounced for past losers than winners. Lee and Swaminathan (2000) affirm previous findings and add intermediate-horizon “underreaction” and long-horizon “overreaction” can be explained by trading volume and book to market value investing. As value stocks have lower trading volume and lower analyst coverage they tend to surprise markets more and become favoured. After some time, however, trading volume and analyst coverage increase, and as value stock P/E ratio increases, it falls out of favour and enters reversal.

Further literature finds momentum generally profitable even after considering trading costs, and more tax efficient than value, as it generates substantial short-term losses and lower dividend income. Also, 77% of mutual funds use momentum strategy (see Momentum in Mutual Fund Returns) with significant herding effect. However, herding does not demonstrate itself in buying and selling at the same time. Funds using momentum are found to be better off than those that do not. Empirical evidence also suggests that even simple strategies implemented by individual investors can be profitable (see Momentum Effect in Stocks in Small Portfolios). Curiously, momentum has been found to be pervasive phenomenon throughout different countries (see Momentum Factor Effect in Country Equity Indexes) and throughout institutional arrangements going back to Victorian era or Imperial Russia.

Overall, momentum effect continues to deliver excess returns and its existence is widely established. Main factors believed to generate momentum effects are under-reaction as well as over-reaction to information, trading volume, analyst behaviour and information dissemination. Interestingly, these driving forces have been found to be more stable and persistent than value and size effects. Later research finds that intermediate-run momentum and long run reversals (see for example Reversal Effect in International Equity ETFs) as found by DeBondt and Thaler (1985) are even part of the same cycle that can in turn be explained by value investing and trading volumes across time. Historically, momentum in stocks has been generally found to experience significant reversal after it runs out. This reversal can sometime become extreme as demonstrated by most financial crises that had strong prior run up period. This momentum crash can be mitigated by combining momentum strategy with other strategies, like Momentum Combined with Asset Growth Effect or Momentum and Reversal Combined with Volatility Effect in Stocks.