Premium Market Analysis, Trader Education, Software, and Trading Strategies. Thirty Years Of Skin In The Game

Trading Strategies

My Favorite Trading Strategy

Photo by Burak The Weekender.

My favorite trading strategy must fulfill eight criteria. In this article, I list the criteria and also briefly discuss my favorite trading strategy that meets them all.

All traders do not have the same preferences for trading strategies, and these may change over time. Depending on their style, traders favor different strategies. In this article, I list what I believe are good properties of a trading strategy and briefly discuss my favorite one.

Properties of a good trading strategy

  1. The foundation is sound.
  2. The turnover is low.
  3. There is diversification.
  4. The execution is easy.
  5. The return is reasonable.
  6. Sharpe is close to 1.
  7. One parameter maximum.
  8. Performance is robust.

1. When developing trading strategies, the majority of traders usually follow a reverse path: they first develop a strategy and then try to figure out whether it has a sound foundation. The reverse path can lead to losses and frustration. A correct path is a top-down approach that involves first considering a sound foundation for a strategy and then developing one based on it.

For example, machine learning and genetic programming models can generate millions of trading strategies from historical data, but most, if not all, do not have a sound foundation, or what I usually refer to in podcasts and social media posts as “economic value”.

In blogs and on social media, there are often examples of strategies with extremely high returns in backtests, and there are a host of reasons for the lack of a foundation, but a frequent one is that these strategies reflect the potential of a market maker when taking the opposite side. For example, higher-frequency backtests using limit orders for entry signals ignore the “impact” and reactions of other market participants. In actual trading, high hypothetical returns can easily transform into high realized losses when the foundation of a trading strategy is not sound.

2. A low turnover offers multiple benefits, like low transaction costs, the minimization of the impact of slippage, and what many traders often underestimate: fewer execution errors. Professionals and institutions have a low-cost structure and the quant power to minimize the impact of high turnover, but in the case of retail, frequent trading acts as wealth redistribution that benefits the sell-side.

One way of lowering turnover is by looking for specific rare anomalies in price action. This is far from trivial and requires advanced quantitative skills that most traders do not possess or are unwilling to invest in acquiring because they have the wrong conception of what constitutes an anomaly in price action. Indicator and chart pattern signals are not anomalies but more often random formations. These patterns may look good in backtests due to over-fitting, but in reality, the expectation is zero before trading costs and slippage. These are discussed in Chapter 5 of my book Fooled by Technical Analysis: The perils of charting, backtesting and data-mining. 

Another way of limiting turnover is by increasing the timeframe, for example, from daily to weekly. I favor this approach but there is an impact on trade sample size and the potential loss of statistical significance. The increase in timeframe must be combined with (1) above, i.e., there must be a compelling and sound foundation for operating with a slower trading frequency. Trading strategies tend to have lower robustness with a slower trade frequency. Therefore, careful analysis must be made to analyze the impact of the lower frequency. If the strategy has a sound foundation, there are higher chances for success.

3. A good strategy must provide sufficient diversification, preferably in the ensemble and time domains.

Ensemble-domain diversification is realized when the strategy is applied separately to different markets and the results are positive in stocks, bonds, and commodities. A trivial example is a dynamic allocation to stocks, bonds, and gold.

Time-domain diversification is present when the strategy trades different markets over time. An example is a trend-following strategy based on momentum or breakouts.

Time-domain diversification could be inferior when compared to ensemble-domain diversification, but both can provide a hedge, or what is usually referred to as “convexity”, during particularly bad times in one of the markets.

4. Ease of execution is important but underestimated due to the (false) belief that nowadays programming can solve all the issues and everything can be automated. Some professional traders prefer strategies that they can continue to execute even if they have no access to their computers. I prefer strategies with simple rules I can remember and execute from any computer with access to market data and basic charts. This is also a criterion for a more robust edge: If something is too complicated and requires a background algorithm to generate trades, then unless it is making markets, it has a low probability of competing in real life. Complexity and randomness are inextricably related, and although market makers and high-frequency traders make money from it, this is only because the structure of the game is designed to favor them. The simplest source of profitability for market makers is the bid-ask spread, and even if prices are random, they profit.

5 and 6. The annualized return of the strategy must be reasonable and offer some alpha above the risk-free return. In a higher-rate environment, as is currently the case, most traders should stay on the sidelines because there is no alpha. Higher interest rates make it more difficult to profit without leverage. In the 1980s, when rates were too high, some traders used to buy zero-coupon bonds and then used the discount with high leverage to generate alpha. Some firms also offered this scheme in the form of “guaranteed capital”. For most, the result was a loss of “trading capital” in a few months and then waiting for several years to realize the risk-free return after the bond matured.

What constitutes a reasonable return depends on prevailing market conditions. At present, in my opinion, very few traders have an edge over the T-Bills and should not be trading.

The Sharpe ratio must be as close to 1 as possible. There is a tendency to discount the value of the Sharpe ratio, especially by funds that cannot realize a high enough value. If the strategy does not provide “convexity” during bad times, then a low Sharpe ratio is an indication of high risk.

Duration and drawdown depth are related to the Sharpe ratio:

  • Drawdown duration is proportional to 1/(Sharpe)^2
  • Drawdown depth is proportional to 1/Sharpe

See this article for more details.

If the denominator (volatility) is high, a low Sharpe ratio means there is a high chance of ruin.

7 and 8. I cannot overemphasize the importance of limiting the number of parameters used in a strategy. I prefer strategies with no parameters other than the number of markets to trade, but this is not always possible. One parameter is the maximum, but that is already too high and can lead to over-fitting. The impact of a parameter on robustness can be analyzed with stochastic modeling. Often, that involves changing the parameter values and looking at the distribution of key performance metrics.

My favorite trading strategy

My favorite trading strategy has the following properties:

  • The strategy is based on cross-sectional momentum.
  • The strategy trades ETFs in several different markets.
  • The timeframe is weekly.
  • The strategy has only one parameter.
  • The annualized return since 2004 is 9.4%, and Sharpe is 0.9 (backtest).
  • The strategy has made 272 trades, or about 14 trades per year on average.
  • The maximum drawdown is 11.5% (backtest).
  • In 2022, the strategy ended the year with a small loss of less than 1%.

The equity curve, annual returns, drawdown profile, and weekly return histogram are shown in the backtest below.


In the same backtest period (12/31/2003-06/04/2023), the SPY ETF has an annualized return of 9.1%, a maximum drawdown of 55.2%, and a Sharpe ratio of 0.51.

This strategy has had nearly zero correlation with the SPY ETF in the backtest period. Because the foundation is cross-sectional momentum and it is sound, this strategy (ETFNRW) is included in the ensemble of six strategies we use for our weekly Market Signals reports. 


Trading strategies must have a sound foundation that has economic value. Usually, traders first look for strategies and then attempt to rationalize their foundation, but the correct process, in my opinion, is the reverse. Increasing the timeframe has benefits, but there can be an impact on robustness, so minimizing the number of parameters of a strategy is of paramount importance. A low turnover offers solid advantages in the case of retail traders. Finally, strategies must be simple and easy to remember. Complex strategies may have advantages for certain market participants, but in the case of retail, there are more disadvantages than benefits. Complexity usually favors the selling side.

Premium Content

Online Books
Premium Articles
Systematic Market Signals

By subscribing you have immediate access to hundreds of articles. Premium Articles subscribers have immediate access to more than two hundred articles and All in One subscribers have access to all premium articles, books, premium insights, and market signals content.



Free Book

Subscribe for free notifications of new posts and updates from the Price Action Lab Blog and receive a PDF of the book “Profitability and Systematic Trading” (Wiley, 2008) free of charge.



Disclaimer:  No part of the analysis in this blog constitutes a trade recommendation. The past performance of any trading system or methodology is not necessarily indicative of future results. Read the full disclaimer here.

Charting and backtesting program: Amibroker. Data provider: Norgate Data