What is the most intuitive trading reaction? For many trend traders, the answer is simple: go long in a bull market, and go short in a bear market.
But a paper published in March 2026, “Rethinking Trend Following: Optimal Regime-Dependent Allocation,” challenges this intuition. The core conclusion by author Valeriy Zakamulin is that while “reducing exposure” in bear markets is often reasonable, “going all-in short” may not be the best choice. From the perspective of maximizing the Sharpe ratio, bear-market positions should often be close to zero, and in some markets may still support a small long position.
Traditional trend following: go long in bull markets, go short in bear markets
Trend following is one of the most common tactical asset allocation methods in financial markets. Its basic logic is that recent performance of an asset contains information about expected short-term returns. Therefore, investors can increase exposure during uptrends, reduce exposure during downtrends, and even establish short positions.
The paper points out that time-series momentum strategies—Time-Series Momentum (TSM)—typically use an asset’s past 12-month excess returns as the basis for judgment. If the return over the past 12 months is positive, it is treated as a bull-market state, with an allocation of +1, meaning fully invested long. If the return over the past 12 months is negative, it is treated as a bear-market state, with an allocation of -1, meaning fully invested short.
The problem is that although this “+1 / -1” rule is simple, it carries a very strong assumption: being fully long in bull markets is rational, and bear markets should also be shorted at the same magnitude.
Zakamulin argues that this assumption has not been sufficiently proven. Traditional momentum literature mostly focuses on “how to identify trend signals,” such as using 12-month momentum, 1-month momentum, or combinations of signals at different speeds. But the truly important question is: once you already know the market is in a bull or bear regime, how much position should you actually take? In other words, the paper questions the practice of equating “being bearish” directly with “going all-in short.”
Market state is not the trading answer—it’s the allocation condition
The framework proposed by the paper is called Optimal Regime-Dependent Allocation, abbreviated as OPT. Its approach is to treat market state as conditional information, and then calculate the optimal exposure for each state based on the state’s conditional returns and risk.
In the traditional TSM strategy, once a bear signal appears, the position is fixed at -1. But in the OPT framework, a bear regime only tells investors that the market is in some state. Whether to short, go flat, reduce size, or hold a small long position must be determined by the state’s conditional average return and risk.
The paper’s mathematical conclusions can be simplified as follows: the optimal position depends not only on the average return in a given state, but also on the volatility and the second moment (second difference). If, in a bear state, the expected return is low but the risk is very high and the signal noise is large, then going fully short could actually reduce the overall Sharpe ratio.
This is the paper’s most important shift in perspective: a bear signal does not necessarily mean “you should bet against it.” It may only mean “the risk-return tradeoff of this market state is not worth taking a large position.”
Empirical evidence from US equities: the best bear positions are often close to zero
The paper first tests two strategies on the US overall stock market: traditional TSM and OPT. The training data starts in July 1926, and the optimal bull- and bear-regime positions are estimated using different training windows. Those estimated positions are then applied to subsequent out-of-sample periods, up to December 2025.
The results are clear: OPT’s annualized Sharpe ratio is higher than that of traditional TSM in all tested windows.
For example, with a training window from July 1926 to December 2003 and an out-of-sample evaluation from January 2004 to December 2025, traditional TSM’s annualized Sharpe ratio is 0.494, while OPT increases it to 0.727—a difference of 0.233. More importantly, the bear-market position estimated by OPT is not -1, but only 0.01, which is almost equivalent to being flat.
This implies that in the US stock market sample, the optimal strategy in bear markets is not aggressive shorting, but holding almost no risky assets. The authors also note that the estimated Bear-regime weights are usually close to zero, and may even be slightly positive. This suggests that going all-in short is neither necessary nor the optimal choice for maximizing risk-adjusted performance.
Not just the overall market: the same holds for large caps, small caps, value, and growth
The paper then applies the same approach to four US stock portfolios: large-cap, small-cap, value, and growth. The results again support OPT. Based on out-of-sample evaluations from 2004 to 2025:
For large caps, the TSM annualized Sharpe ratio is 0.553, OPT rises to 0.790, and the bear-market position is 0.13.
For small caps, TSM is only 0.074, OPT increases to 0.366, and the bear-market position is 0.31.
For value stocks, TSM is 0.689, OPT is 0.856, and the bear-market position is 0.03.
For growth stocks, TSM is 0.092, OPT rises to 0.521, and the bear-market position is 0.27.
What’s most interesting about these data is not only that OPT beats TSM across the board, but also that none of the bear-market positions are negative. That means the data does not support “bear markets should be shorted all-in” in these four US stock portfolios. Instead, it indicates that exposure should be sharply reduced in bear markets, and even some long exposure may still be retained.
This is instructive for investors: many people conflate “avoiding bear markets” with “shorting bear markets,” but they are fundamentally different in portfolio management. Avoiding bear markets is risk control, while shorting bear markets is making a directional bet in the opposite direction.
International markets: going all-in short is also not a common optimal solution
The paper also tests 14 developed-country equity markets, including Australia, Belgium, France, Germany, Hong Kong, Italy, Japan, the Netherlands, Norway, Singapore, Spain, Sweden, Switzerland, and the United Kingdom.
The results still show OPT outperforming TSM across the board. Across the 14 countries’ average annualized Sharpe ratios, TSM is only 0.054, while OPT increases it to 0.295, with an average gap of 0.241. More importantly, the average bear-market position is 0.10, not -1.
Some countries particularly illustrate the point. For example, in the French market, traditional TSM’s Sharpe ratio is -0.239, but OPT improves it to 0.086, and the bear-market position is only 0.01. In Italy’s market, TSM is -0.226, and OPT remains negative at -0.084, but it is clearly improved; the bear-market position is -0.30. This shows that even if shorting has some value in certain markets, the optimal short position is typically far lower than the traditional TSM value of -1.
In other words, international market data supports a more conservative yet more effective conclusion: in bear markets, reduce risk exposure, and do not automatically equate bear signals with going all-in short. Shorting bear markets is not necessarily wrong, but going all-in short is likely too decisive. Especially because equity markets have risk premia over the long run, bear-market signals may reduce future return expectations, but they do not necessarily imply that future returns are negative enough to justify high-intensity short positions.
Finding the answer in the paper: Is shorting bear markets a good choice?
According to the paper’s answer: it is usually not the best default option.
More precisely, the most reasonable approach in bear markets is often to reduce exposure, stay close to flat, or—only in a small number of cases—establish a modest short position, rather than mechanically going all-in short. Empirical results from the US stock market, international markets, the four-state model, and multi-asset portfolios all point to the same direction: optimizing positions matters more than simply improving signals.
This also shifts the focus of trend-following strategies from “Did I get the bull/bear call right?” to “How much risk did I take in different states?” The true value of bear signals may not be to encourage investors to short bravely, but to remind investors that the risk-return tradeoff is different now, and positions should be handed back from trading intuition to risk allocation.
Is this article titled “Is Shorting Bear Markets a Good Choice? The paper’s counterintuitive conclusion: Going bearish without shorting—reducing exposure is the way to go” first appeared on Lian News ABMedia.
