All-Time High Trend Strategies Are Not Without Costs and Risks

Barry J from TX posted over 1 year ago:

So, Charles, I only audited this course because I needed a PE credit and you told us there would be no math on the final exam. #1 How is this example of (1) a “just right” Mama Bear (2) CONTRIVED calculation of (3) an INVENTED statistic – “risk units” -- (a calculation so contorted that it reminds me of the difficulty of “sticking” an Olga Korbut dismount) NOT THE POSTER CHILD for violating the #1 FINANCIAL INDUSTRY ALL TIME TABU -- “TRYING TO TIME THE MARKET”? #2 I noticed you did not provide links so we can track down the original research of these brainiacs and ask them now much of tier personal CASH they earned from “capturing” all the “risk units” they earned over the 75 years they tortured the BACKTESTING dataset. #3 They found only 66,000 “eligible” trades over the 75 years. I estimate that investors make 66,000 trades every millisecond (or a total of 23,000,000 msec) EACH trading day of 250 days each year of all those 75 years the market was open to find 66,000 needles in their 75-year haystack. By way of comparison, markets would have been open 18,750 days when those 66,000 “gold ATR nuggets” were found … AFTER THE FACT. Remember, this strategy requires you track each ATR of EVERY stock day-over-day to identify each ATR. Then what? YOU yell “Stop the market!” and call a broker and have them put in a buy/sell order (that’s the way markets worked for at least 50 of those 75 years) and the “roundtrip cost of EACH buy-sell transaction cost an investor about $20. #4 What? This was only an experiment. I am shocked! Alert Captain Renault. Consider the similarities in this dialogue from the movie Casablanca with this strategy: Renault: What in heaven's name brought you to Casablanca? Rick: My health. I came to Casablanca for the waters [ATRs]. Renault: The waters? What waters? We're in the desert [of unautomated markets before the internet was commercialized in 1995 and zero trading costs were offered in 2019]. Rick: I was misinformed. #5 Charles, I’ll give you a pass on this one. I will pretend this article was 3-week late April Fool’s Day prank. #6 PS, Can I count this as my math credit?

Douglas R from Washington posted over 1 year ago:

The paper behind Mr. Rotblut’s article is “Does Trend Following Still Work on Stocks?” (2025) by Carlo Zarattini, Alberto Pagani, and Cole Wilcox, freely available at the link Mr. Rotblut kindly provides in his reviews. When reading finance research like this, I remind myself of the audience; most papers are written for other academic researchers or institutional investors, but not individual investors. Before accepting the findings, it is essential to me to evaluate whether the strategy is relevant and practical—or at least adaptable—for individuals. If not, the findings may be of little use to me. Having reviewed the paper through the lens of an individual investor, I have several comments on the article: Mr. Rotblut’s strongest critique is that “transaction costs are a significant consideration,” a point made in the title as well. I could agree, but this is unimportant because the strategy is not feasible for the vast majority of individual investors. The main obstacles are not the costs themselves, but the drivers of those costs. Implementing this strategy requires a fee-per-trade brokerage account with API access (such as IBRK), because automation is necessary. The strategy is beyond the capabilities of free Yahoo data and Excel; one would need a specialized proprietary platform or would have to write their own custom Python or R scripts. Furthermore, the high slippage stems from the extremely high trading frequency—about 880 trades per year over the backtest period. This level of activity makes automation a necessity, making it infeasible for the vast majority of individual investors. (The articles also fails to mention that the authors offer a “Turnover Control Mechanism” in the second half of their paper intended to reduce costs. I won’t describe it here either, since it is unimportant to individual investors when the strategy is not feasible.") Mr. Rotblut also highlights that the average holding period is 370 days, which would allow more trades to qualify for long-term capital gains tax treatment. However, this too is largely irrelevant for individual investors given the practical challenges of implementing the strategy. With an average of 880 trades per year, even if most qualified as long-term holdings, capital gains would be realized regularly, with almost none deferred. This high level of trading will result in frequent realization of gains, creating a tax drag that erodes the compounding benefits of pre-tax returns that might otherwise be possible with another selection strategy. Lastly, when he says, “While often grouped under the momentum umbrella, trend-following looks at a stock’s own price history,” it seems Mr. Rotblut insinuates that trend following is not properly categorized as a momentum strategy. Of course, the paper’s use of the breakout high method is not an example of cross-sectional momentum (relative momentum) that dominated the literature from the mid-1990s, but it is certainly an example of time series momentum (absolute momentum) that dominated the literature before this period and is mixed and mingled in the literature thereafter. I assert that momentum is where you find it, and this paper demonstrates that new price highs lead to even more highs from which one can profit. This point is missing from Rotblut’s article, and I think it is the single most valuable finding for individual investors. Momentum can be a powerful source of excess returns, and this paper illustrates an example—albeit one that is not feasible to implement—but nonetheless provides more evidence that momentum returns can be found in many ways. So, we individual investors should be encouraged to look for momentum strategies that are feasible to implement; there are many possibilities and we will find many that do not suffer from cost concerns or tax drags anymore so than other strategies.