In simple terms, momentum investing puts money in winning stocks and avoids, or shorts, losing stocks.

The momentum effect has been found by numerous studies to generate long-term excess returns to investors. However, negative returns from momentum strategies over the first decade of this century have cast doubts on the future of momentum investing.

The main concern is that the momentum effect may be due to an unidentified anomaly, so future returns may defy rational predictions. In addition, it is possible that overinvestment in momentum strategies has arbitraged away all available returns. If either of these concerns were to be substantiated, momentum investing would become considerably less attractive to investors.

Given that studies of stock market returns have limited themselves to data since 1927, the authors decided to backtest a further 125 years of returns. Constructing a dataset going back that far was quite a feat because the authors had to combine three known 19th century and early 20th century datasets into one. They then analyzed the newly created 1801–1926 dataset to test for the price momentum premium.

The authors believed that looking at returns since 1801 would provide more insight into factors behind stock market returns and enable them to judge whether the momentum effect is durable. In particular, they looked for momentum drawdowns in the pre-1927 period to see how often long periods of negative returns occurred. They also explored whether hedging could avoid drawdowns and improve returns of momentum strategies.

The authors identified three main findings, each with promising implications for investors.

The first is that the momentum effect is robust. The previously untested pre-1927 data confirm the existence of a price momentum premium. The authors found that over the 212 years from 1801 to 2012, excess returns from their momentum strategy averaged about 0.4% a month.

The second finding is that, historically, momentum investing has not worked for a number of long periods, so the negative returns during the decade up to 2010 are not an unusual phenomenon. Since 1801, there have been seven 10-year periods during which the momentum strategy generated negative returns.

The third finding is that the market state (rising or falling) has the largest impact on momentum returns: The longer the market state lasts, the more market exposure contributes to momentum returns. Therefore, momentum strategies become riskier in long market states. Why? Consider a long bull market. As it develops, momentum returns become more and more correlated with market returns. So, if the market falls fast and far, momentum strategies do too.

The authors found that dynamically hedging out beta—essentially, removing the effects of changes in market state as they occur—increased returns from the momentum strategy from 0.4% a month to 0.7% a month.

Investors may be reassured to know that momentum investing has produced excess returns for a lot longer than previously thought—200 years at least. The finding that the main contributor to the momentum effect is the market state provides further evidence that excess returns are not due to an anomaly and can be replicated.

The concerns over the underperformance of momentum strategies over the first decade of this century will also be soothed by the finding that such strategies have generated negative returns in seven separate 10-year periods since 1801. Why is this positive news for investors? Because despite numerous periods of underperformance, momentum strategies still overperform over the (very) long term. So, today’s investors may well be rewarded by allocating to momentum strategies—if they are prepared to be patient.

Perhaps the key takeaway for portfolio managers is that dynamic hedging adds significantly to the performance of momentum strategies. Hedging the portfolio at market turning points can enable momentum strategies to hugely outperform an unhedged portfolio during periods marked by large market reversals.