Stock markets, trading and investing truly intrigue me. What originally piqued my interest in the stock market was the perceived ability to make endless profit. However as I got more and more into the markets I began to realise that the holy grail of endless profit, as it’s name suggests, is an elusive object or goal of great significance that only a handful of people have discover, or even perhaps noone has discovered.
Despite the illusion of the holy grail being shattered for me, I still believe that it is possible to extract above average returns from the market. Rather than looking for the holy grail, I now look to find small edges with the belief that all it takes to extract these above aveage returns is a small edge, or combinations of small edges that will aggregate and compliment each other.
One of the podcasts I listen to, Chat With Traders, recently had an interview with Nick Radge, the author of Unholy Grails a new road to wealth. In the podcast Nick outlined a strategy trend trading strategy that he also describes in his book that uses bollinger bands as tool to identify new trends that are forming.
In this blog post, I will outline the bollinger bands trading strategy that Nick Radge describes with some tweaks that I have back tested, discuss the results of the back test and conclude with my opinions on the strategy.
The strategy is largely based off a 100 day simple moving average (SMA) asymetrical bollinger band. The key points:
- Entry: Price closes above +3 s.d. (upper bollinger band) from the 100 day SMA of the stock
- Exit: Price closes below the highest point of the -1 s.d. (lower bollinger band) from the 100 day SMA of the stock. Or the stock is removed from the trading universe
- Trading universe is the ASX300. The entry trigger is only valid when the ASX300 is above its 75 day SMA
- The portfolio will hold a maximum of 20 stocks at any point.
- Start portfolio value is $50k
The exit criteria has been tweaked from the original strategy described. The exit is now functioning as a stop loss that can only move upwards. Meaning that if the stock’s -1 s.d. from the 100 day SMA starts to move lower, the exit position will maintain the highest position it has been at during the life of the trade. This tweak was made as I found that the stock could slowly move down over longer periods and never satisfy the exit criteria despite showing clear consistent downward trends.
The Test Results
The test will be broken into two sections. The first is based on actual share market data from August 2001 through till the beginning of September 2019, the second section will then be many iterations of slightly randomised data over the the same period, the data will be adjusted plus or minus 3 percent. The second section is designed to be many stress tests of the strategy.
The test data chosen has some very large events in the stock market history, which includes:
- The dot come crash, which resulted in a major crash from 11th of March 2000 through till 9th of October 2004
- The global financal crisis of 2007-2008
- August 2011 stock market fall
- 2015-16 stock market sell off
- 2018 US stock market downturn
For simiplisity, dividends will not be considered. Over the period considered the ASX300 experienced an apprciation of 99.6%. Which will be our target with the strategy.
Actual historical back test
The actual historical back test, shows signicant gains, but also some significant dradowns.
The following graphs show the following interesting features
- The total gain on the strategy is 2037.6%. Outperforming the ASX 300 over the period by 1937.4%
- The strategy closed 463 positions during the period, with a win percentage of 49% and a win loss ratio of 3.3:1, meaning that a winning trade was 3.3 times more profitable than what was lost in a losing trade.
- The average hold period of a winning trade was 310 days, approaching the 1 year capital gains threshold here in Australia, whereas a losing trade was held for around 113 days.
- Strong run up to the GFC in 2007/2008, with some small periods of drawdown. Drawdowns up to this period experience 3 sharp drops between 25 to 30%, but recovered quickly.
- From the GFC until the beginning of 2016 was very turbulent for the strategy, It looks like the strategy still beats the market over this period, and does experience some periods of strong upward movement, the movement just did not carry on. The strategy even spent 3 years between 2010 and 2013 all in drawdown.
- The strategy performs really strongly from 2018 through till the 2018 US stock market downtown.
- During the 2018 US stock market downturn, the strategy experiences a spike drawdown of up 40% for a very short period of time before recovering close to all time highs.
Randomised historical back testing
After the historical back test provided positive results, I would like to inverstigate further with some stress testing of the strategy. By running 1000 iterations of randomised back testing we can see guage how well the strategy could perform with additional volatility. To add randomisation to the back test I multiplied the end of day closes by a random number between 97.00 and 103.00%, giving an additional variablity of 6%.
When taking all of iterations and their results we can see a similar average capital flows. However the biggest difference visible is from from the GFC through till 2016 shows a much stronger more clear upward trend, without the same long period of drawdown the actual test experienced
Interestingly the randomised results do not typically experience the same overall profit, as can be shown in the below scatter plot, the randomised results also seem to experience lower maximum drawdowns as well. This indicates that one should expect lower returns, and also lower drawdowns.
Additionally the average number of trades, and average hold period of winning and losing trades are very similar, however the average win percentage falls to 45% and the win/loss ratio also rises to 3.5:1.
Overall, the bollinger band breakout strategy does seem to create an edge within the market, that could lead to extracting above average returns when compared to the market.
The relatively simple entry and exit criteria should be easy to implement and stick too, however the relatively often and large drawdowns may really play on your trading psycology, especially during a period of prolonged drawdown.
I think with further tweaks and corrections, especially to the exit criteria, the drawdowns could be reduced in terms of amount of drawdown experienced, but it would be unlikely if you could reduce the number of drawdown events.