Before I start a new chapter about how to develop and test your strategy before backtesting, I want to add closure for my last article(link last article).
As I mentioned before, the test result of Moving average crossover momentum I posted yesterday(link last article) shows that the beta of the strategy is high(means the strategy is high market risk exposure) and I did not take commission cost into consideration. Therefore, I decided to fix these problems for the new adjusted strategy. You can check the changes I made from the screenshots that I upload here.
In short, I set up the daily universe based on stocks Dollar Volume of them and only choose those whose dollar volume are at the top 1% among all stocks we can get access to. There are still 5 stocks in both long and short baskets for this model and the rebalancing frequency is still the same every 15 days. But I change the look-back period into 60 days because here I use a new moving averages strategy. The mechanism behind this one is as the picture below:
I use the Moving Average Convergence Divergence as a measurement to select the stocks and allocate them into my long or short basket. Basically, MACD uses Exponential Moving Average(EMA), which is similar to simple moving average except that more weight is given to the latest data.
If MACD > 0, the short-term average is above the long-term average, which signals upward momentum. I will add qualified stocks into my long portfolio. The opposite is true when the MACD < 0 which implies downward momentum and I will add those stocks into my short portfolio.
Besides, A signal line, also known as the trigger line, is created by taking a nine-period EMA of the MACD(functioning as a trigger for buy and sell signals). Thus, you can see three components in MACD method, fastperiod(12), slowperiod(26) and signalperiod(9).
Moreover, I include the commission cost in my strategy this time and the cost per share is $0.01. Here is the backtesting outcome of the new Time-series momentum strategy I developed after these adjustments. You can see from the metrics that the beta becomes much smaller here which is only -0.06(relative market neutral strategy). The Sharpe ratio is 1.52 which is pretty much the same as the one before.
This graph is the result from the same strategy but without any commission fees(easier for you to compare with the one I showed last time). It is not surprising that the total returns, Sharpe Ratio, and Alpha are all higher and Max Drawdown is smaller when we get rid of commission cost.
But still, this new strategy has its own issues such as it is sensitive to the rebalancing frequency and look-back period which I will pay extra attention to and try to find a solution with regard to this problem for my future strategies.
Hopefully, this article can provide a new perspective for you about how to measure and adjust the strategy. Please feel free to leave and comments or ideas here. I would love to hear from you.