February 28, 2017
You often hear academics and traders say: "all trading strategies fail over time". They don't provide proof but will provide examples to make their point. And usually, for the examples they present, I agree, those strategies should fail. It is as if their selected trading strategies were designed to fail in the first place, and therefore, no one should be surprised if eventually, they do fail.
There are exceptions, but I do not see them as such.
I view them as better strategy designs. They don't need to be elaborate or complicated. But, they don't need to be simple, either. However, they will have to adhere to common sense and the math of the game.
For instance, I have highlighted my preferred trading strategy (DEVX8) a few times on my website. Each time doing more and more elaborate tests to see how it would have behaved under different trading conditions. And this strategy did not fail, even over a 20-year test. There are reasons for this.
You do a simulation to find out if the trading procedures you want to set up would have worked in the past. You certainly can not test the next 20 years without living them first. If you wanted to test the next 20 years, you would reach the end of test with nothing to show for it but a test result, good or bad, and would have wasted 20 years of your time.
A Closer Look at DEVX8
DEVX8 is a strange kind of stock trading strategy. It trades following random-like functions and was designed to obey the following general principles. It takes a trade only if the price is above its initial price, and disregard stocks operating below it. Thereby minimizing the impact of any weak stock in its portfolio. When a profit is available, it might be taken but is not forced to do so, making the bet that there might be more profits just by waiting some more. Proceeds from sales are reinvested.
DEVX8 is designed biased to the upside. Its primary mission is to build up a portfolio inventory, gradually accumulating shares for the long term. To this end, it simply buys more shares than it sells.
On average, over the past 8 years, the US market has been mainly going up. Over this time, a trading strategy needed only one decision which was to buy shares. But it would also have been the case for the past 50 or 100 years. In an up market, you buy shares until they stop rising. You want to outperform averages then you reinvest ongoing profits.
Say you invest your money at a 10% CAGR, but each year you take out the 10% return. You automatically forego compounding for your 10% withdrawal. And even after 20 years, you will end up with the same amount you started with. At least, you would have had the use of the 10% annual return as income. But, this would not have built you a portfolio. It would have maintained it with no appreciation. A(t) = A(0)∙(1 + r - r)^t.
To get compounding, you have to let your portfolio grow with time. In the end, there will be three important numbers: your initial capital, your growth rate (CAGR), and how much time this compounding was allowed to operate. The formula: A(t) = A(0)∙(1 + r + α)^t. Where α represents the premium you bring to the game (your skills, your trading methodology).
In DEVX8, a stock price series is divided into 3 trading zones: buy, sell, and hold as is. Maybe the best way to describe this is with an idealized price movement chart.
A Simplified Model
(click to enlarge)
In the above chart, buying (yellow dots) occurs within a buying zone (green line segments) during the upswing. While selling (white dots) is done on the rise after having crossed a no-trade zone (yellow line segments). This is done for every swing of significance. Based on such a design, all trades can be closed profitably when prices go up.
It is the reason for the high hit rate seen in the DEVX8 trading strategy. It is not that you are predicting that the price will rise. It is simply that you are selling when you do have a profit in hand. The objective being to do this cycle after cycle, and use the generated profits to buy even more shares.
That is the main purpose of the trading strategy: to accumulate shares for the long term. So your interest, evidently, is in the best-performing stocks out there. They are easy to detect. Their stock prices are going up, on average, and perform above market averages. It is how you outperform averages. It is by buying stocks that are already performing above market averages.
DEVX8 was shown to be controllable. You could set the degree of trade aggressiveness (see related articles). In this test, I did not want to push it. I only wanted to show that the strategy did not break down and continued its mission of accumulating more shares.
A 3-month walk forward test could be considered sufficient to show if the principles on which a strategy is based has merit. The law of large numbers will help substantiate claims for averages. Even more so if the test duration is already for 20+ years and does thousands and thousands of trades.
This test is done on top of the November 2016 simulation which already was a one-year walk forward test. Therefore, this test is to show that it kept on ticking. The strategy did not fail.
The Test
I opted to use the moderate settings: 40, 95, 80, 30, 30, 30. The same setting as was used in the Controlling a Stock Trading Strategy article (see figure #4).
The current simulation was run once and had for results:
DEVX8 - Feb. 17 2017 with controls: 40, 95, 80, 30, 30, 30
(click to enlarge)
The strategy ended with a net liquidating value of $ 848,404,062 from its $ 5 million start, some 20.96 years ago (5,449 trading days). It maintained an average CAGR of 27.76% over the period.
During this time, the strategy made a total of 58,605 trades of which 34,565 were closed generating $ 672,966,400 in profits. Most of it still in cash since the ending cash reserves totaled $ 436,049,709. The program could have been put on a more aggressive stance as was demonstrated in the Buy & Weak Hold article. The available cash reserves were more than sufficient to increase the program's trading activity.
The average hit rate was 92.95%. This was not a matter of luck. There is a simple reason for this. Trades were only closed if at a profit. To get a rate lower than 100%, you have to have some still opened positions in the red. While a 100% hit rate only stated that you were at a high for the price series which will happen every time new highs are hit, or the current price is above the highest purchase on record.
Since I have used ABT as an example in the other tests, here is its position profit distribution:
ABT - Feb. 17, 2017 – The Percent Profit per Trade Distribution
(click to enlarge)
Each of the 6,060 dots represents a trade. Blue dots are profitable positions while red dots are for positions still in the red. As can be observed, there are not that many red dots, and they are not losing much when compared to the blue dots. In fact, only 111 out of the 6,060 trades taken are seeing some red. Given time, they too might go positive and join the triangular area of blue dots on the left. Out of 6,060 trades, 5,949 are showing a profit, of which 3,161 have been closed with an average profit of 92.38%.
For comparison purposes, here is the November 14th, 2016 test which was performed using the same program and the same settings:
DEVX8 - Nov. 14 2016 with controls: 40, 95, 80, 30, 30, 30
(click to enlarge)
See Controlling a Stock Trading Strategy for a detailed explanation of the above chart.
I added a difference table to compare the two tests. All changes are due to the added 64 trading days.
DEVX8 - Differences: Feb. 17, 2017 – Nov. 14, 2016
(click to enlarge)
Over the three added months, the strategy's account increased by $ 40.9 million. Most of it coming from still opened positions. The strategy added 2,563 trades. Profits in still opened positions increased by $ 40.58 million.
The primary mission was to acquire more shares. Over the 3 months period, the strategy added 505,905 shares to its ongoing inventory, raising the number of shares to 3,993,806.
A short-term trader could not reach these levels. This kind of trades are not available to him. On average, 90% profit trades are not available on a daily basis. Neither does a 92% hit rate.
To get these, you have to wait for them. Play your long-term strategy which incidentally consists mostly of waiting.
But still, the strategy did 58,605 trades over the 20.96 years. That is an average of 10.75 trades a day. All done at market at the opening bell the next trading day. Something that could almost be done by hand.
Or, given to a machine that would take a few minutes a day to execute its trading orders. One would be left with monitoring the portfolio to see it stays on course.
Since the entries and exits were randomly set, it would imply that precise entries and exits were not that critical to the profit-making process. One could have randomized the whole thing; entries and exits which is what the program does.
Conclusion
If you accumulate shares in the DEVX8 manner over the next 20 years, you will win. It is as simple as that. DEVX8 is like a generic trading strategy. It was designed to be scalable up or down, and be controllable.
It behaves the same as if using other trading methods in the sense that entries and exits often appear as if randomly taken which in this case they are.
One could convert his/her trading strategy to do somewhat the same things as presented here and obtain similar or even better results.
It can be a more than an effective way of building a long-term retirement account with a minimum of effort: a few minutes of a machine every day.
Test Results
Just as in previous DEVX8 tests, this test was also run once. Here are the charts for each of the stocks in the portfolio:
ABT - Feb. 17, 2017
(click to enlarge)
ALL - Feb. 17, 2017
(click to enlarge)
BIIB - Feb. 17, 2017
(click to enlarge)
CVS - Feb. 17, 2017
(click to enlarge)
FDX - Feb. 17, 2017
(click to enlarge)
GD - Feb. 17, 2017
(click to enlarge)
GILD - Feb. 17, 2017
(click to enlarge)
HD - Feb. 17, 2017
(click to enlarge)
LMT - Feb. 17, 2017
(click to enlarge)
LOW - Feb. 17, 2017
(click to enlarge)
See related articles:
Controlling a Stock Trading Strategy
Created... February 28, 2017, © Guy R. Fleury. All rights reserved.