Review of Michael Covel's revised edition of Trend following
If you are a participant in the markets, you know that reliable knowledge and information is difficult to obtain. Markets shift, old techniques fade, and some inventive people figure out new ones. If you are new to the markets, you face a huge task of figuring out who to listen to, and who is a phony.
I am a voracious consumer of trading and finance books. A few years ago, I reviewed Michael Covel's first edition of Trendfollowing: How Great Traders Make Millions in Up or Down Markets.
Back then, I had some understanding of the issues that face trading system designers. I have much more now, but even then I could see the book was fluff.
The first edition was nothing more than a promotion piece for trend following fund managers, that you were paying for. The second edition is no different, and in regards to promotion it is probably worse.
In mathematics, there is philosophy called mathematical constructivism. In their school of thought, in order to prove something exists, you have to provide instruction or direction on how to calculate or produce some mathematical object. In short, telling me how to do something is a more valuable proof than merely telling me something is possible.
I now judge trading books by this standard. Tell me how to do something, and I will gladly pay for it. Tell me what others have done (and may now be doing just the opposite) while claiming you are telling me how to do it, is not only a waste of my time, but a wasteful slaughter of poor trees.
I have at least 5 complaints about the book:
1. Recommendations by fund managers are of questionable value. If you are trying to increase assets to manage (and they all are) will you trash a worthless book that paints you in a favorable light? I didn't think so.
2. Pointing to the most successful trend followers as proof is a fundamental flaw in his methodology. It suffers from survivorship bias, in that we don't know how many other funds who subscribed to a trend following method that had to close down. Without this, we have no idea how effective trend following is. (My experience is that trend following is effective, but one needs counter-trend systems for diversification purposes when trend following models are in drawdown. Welles Wilder described such a system in New Concepts in Technical Trading--a book I highly recommend).
3. I want to scream at the idiocy in claiming that trend followers "don't predict." This is word magic that con artists play on people to induce them to buy things--like worthless books.
The whole point of rigorous testing for trading rules is to get an estimate of MATHEMATICAL EXPECTATION--ie your "edge". If the model holds up--what can the trader EXPECT to see if he follows it? This is statistical prediction, as any actuary, or machine learning specialist, would tell you. But Covel shuns the word "predict", as I suspect it would depress his sales to uninformed, wanna-be traders--his target market.
4. Paul Rabar, on the side margin on p. 83, is quoted as saying "Whatever you use should be tested and applied in a rigorous fashion." With that, I totally agree. But if you look at the book, there is not a single equation, or computer program, that would help a trader implement a trend following methodology--a telling indictment that Covel is selling you sizzle, not steak.
5. Without knowing the context of his advice against profit targets, and how they affect the system, it is best described as unsubstantiated folklore. At worst, it could put you out of business. Traders, particularly in the options markets, continually adjust their positions to make sure they are Kelly optimal--ie. the percentage at risk will maximize the logarithm of their wealth--if their estimate of their expectation is accurate. Often, they will discount their estimate and trade at a fraction of what Kelly optimal implies.
(See this paper by Ed Thorp on the Kelly Criterion for financial markets.)
For example, if you have a trading method that is profitable, in that you make 0.50 for every 1.00 risked, about 25% of your account should be invested in the strategy, IF you are willing to suffer the drawdown it implies. (see Ralph Vince's Portfolio Management Formulas for details).
Now, suppose the model has reaped in lots of profits, and because you stuck with the model, the account equity invested in the strategy is 40%. Should you let it ride?
Well, it depends. Do you have good reason to believe that your edge has increased so that 40% is Kelly optimal? Can you live with the drawdown such a %f implies? If you have no reason to believe the strategy has changed much, then you should most definitely book those profits, bringing the amount invested down to 25% of the current account.
I had to learn this the hard way. It has been my biggest mistake in my trading--not taking profits when they are there.
6. He resorts to options pricing theory as a pseudo-justification that trend followers don't predict, and only benefit from infrequent, but large price moves that more than pay for the prior losses in trendless markets.
If that is the case, the whole point of backtesting trading rules is a waste of time. You can do that right now by simply buying out of the money puts and calls. That would be a legitimate claim at a trading style that is price neutral, and not out to "predict" the market. It is the strategy that Nassim Taleb implements in his fund.
But you still have to predict something--market volatility. There are no free lunches in the markets, regardless of what Covel implies.
In short, there is not much new in the revised edition. What is partially true in the book is not new to the trading literature, and what is new in the book is misleading at best, false at worst.
Labels: Book Reviews, Kelly betting, Michael Covel, money management, Trend following
posted by Rob Ryley @ 9:57 AM
0 Comments
