A data-oriented approach to managing your portfolio

It's very common for retail investors to spend quite a considerable amount of time in conducting fundamental analysis in stocks and looking for undervalued stocks to invest in.

While that is certainly important, what is equally important is to understand how to optimise your asset allocation and balance the various components in your portfolio to achieve the best risk-adjusted returns possible.

And this is where I think we could definitely be a bit more data-oriented.

In general, there are two preliminary steps that one could adopt in ensuring an optimal balance or asseallocation in your portfolio.

First, conduct a correlation analysis on the mix of various assets/components in your portfolio to understand what are the correlation factors among the assets of your portfolio. It probably makes no sense if all the assets in your portfolio are strongly correlated to each other in performance. For instance, you may think that buying all the various different ARK ETFs is a good choice as each of the ETFs seems to be a representative of a megatrend (EV, Genomics, Fintech etc). However, if you were to run a correlation analysis on the various ARK ETFs, you would realise that most of the ARK ETFs have a strong correlation factor of ~0.8 and above. Hence, you are actually not as diversified as you like to think as the various ARK ETFs could have the same performance given from what we understand historically. In this case, there is then no reason to be owning all of them since they will be performing identically in all market conditions. Ideally, you would want to have a portfolio of assets which have slight correlation/negative correlation to each other so that your portfolio is well balanced against each other. I generally look for correlation factors of 0.5 or lesser between the bigger assets/components of my portfolio.

Second, run a backtest of your portfolio strategy. Backtesting is a technique used to understand how your portfolio strategy performs ex-post given historical data. It's a common method used by professional investors, but less so for the retail investors. While some might argue that past performances is no indication of future performances, I would argue that there are very few other data-backed techniques one can use to better understand/predict how their portfolio strategy will perform. By doing backtesting, you have an opportunity to understand how your portfolio strategy performs in different market conditions if your backtesting time horizon is long enough. This gives you some insights on how robust your portfolio strategy is. While the earlier correlation analysis helps you in creating a portfolio of uncorrelated stocks for better stability, it does not tell you how much of an allocation to each of these stocks is optimal. This is why doing a backtesting exercise will be important as it provides the data you would need to further balance/optimise the mix of assets/components of your portfolio to achieve your optimal results. Is there a certain time period which results in a huge drawdown of your portfolio through your backtesting exercise? If there is, you could identify the set of market conditions which happen in this certain time period and look out for general weakness in your portfolio allocation. Eg. If your portfolio had a huge drawdown during big drops of oil prices, your portfolio might be too heavy on oil stocks and you might want to trim your positions for such stocks.

How to do it?

Some of us might be thinking that doing the two exercises above requires significant coding skills and are worried that they do not have sufficient technical mastery to attempt them. Fortunately, there are quite a few online tools available for your use without the need for you to know how to even write a single line of code. For example, you could use Portfolio Visualizer to conduct both the correlation analysis and backtesting just by inputting your current asset allocations. It's extremely intuitive and you will be able to pick it up in no time. You are also able to obtain important metrics such as CAGR (Compound Annual Growth Rate), Sharpe Ratio and Maximum Drawdown for your portfolio from the tools available in the site to better assess the performance of your portfolio strategy. And the best part of it? All these tools are free :)

If you are looking at conducting backtesting for your SG stocks, you may consider PyInvesting as Portfolio Visualizer do not cover SG stocks.

Hope this kickstarts your data-oriented journey in managing your portfolio.

You can follow my blog here.