Regular portfolio reviews are something that poor performing investors tend to miss out on. If you pick stocks and own a portfolio, you’ll need to review your performance at least annually. How else would you know if your money is actually working for you?
As we head into July 2021, it is a good time for me and my Early Retirement Masterclass (ERM) community to review our performance thus far. I’ll be sharing how our portfolios have performed below.
But first:
What is the ERM Portfolio?
I retired at 39, and dividend payouts have been my main source of income. As part of my early retirement masterclass, students build a dividend-generating portfolio every run. I’ll then use my trainer fees to invest in the portfolio that was built in that particular class. This gives me skin in the game, forces me to analyse market changes and company announcements seriously, and hopefully assures my graduates that I’ve got their backs.
I have since ran 21 batches of the ERM which has resulted in a combined portfolio of stocks, REITs and Business Trusts of 15-20 counters, per batch.
Initial stock picks employ a mixture of fundamental and technical factor models, and students are then made to research each counter in groups to decide whether to keep or discard these stock picks. The program thus employs both qualitative and quantitative criteria in stock picking, and through research, students can have a greater sense of ownership of their stock picks.
For folk who prefer learning lecture style, I walk through the analysis in greater detail here:
How did the ERM Portfolio Performed in 1H2021
Overall performance (25 June 2021)
The overall unleveraged performance of the portfolio is about 6.97%.
The portfolio picks by students tend to be conservative, with a beta of about 0.8. Overall, the student portfolios are 20% less risky than the overall Singapore Stock market.
The dividend yield on cost is about 5.24%, indicating a slight bias for higher payouts.
Against the STI
The portfolio has outperformed the rest of the STI overall, producing time-weighted returns three times that of an equivalent purchase of the STI ETF.
There is slight underperformance in 2021 as ERM has avoided the restructuring of Temasek-linked shares, but we are cautiously optimistic that we can catch up before the year-end.
Learning continues after the course
The portfolio is vast, with 69 stock picks made by 540 students. Unfortunately, with such a vast number of decisions made, the program has made some lousy stock picks, including the notorious Eagle Hospitality Trust, which claimed to have double-digit yields but had never actually paid a dividend in practice.
One of the convictions of the program is that we love to study our investment mistakes and spend much more time looking at mistakes instead of examining our best stock picks.
As a result, we can derive superior performance from avoiding bad choices more than actively the best investments in the markets.
Top 5 and Worst 5 stock picks
The following table shows the five worst picks in the history of the program :
Still, ERM has made excellent moves.
With counters like Propnex earning 200%, the program tracks growth investing trends by building a dedicated Tech portfolio for each batch moving forward.
However, a single stock don’t make up a portfolio. And market conditions play a role in the portfolio performances. As part of my promise to invest my trainer fees into each portfolio, I don’t get to time the markets, which gives us some good data points.
Let’s drill deeper into the best and worst portfolios out of the 21 batches made by the ERM program.
Note: The criteria for selection is to choose from the portfolios that have been built for over a year, as current portfolios may have an overly skewed XIRR on a small gain or loss that may not reflect its effectiveness as time moves forward.
For this article, we will use the timestamp of 26 June 2021 to compare the portfolios against each other.
Worst ERM Portfolio of 1H2021 – Batch 7
The worst performing portfolio is Batch 7 that was conducted in September 2019. However, it still had a positive internal rate of return of 0.36%, with losses primarily offset by dividends collected. The portfolio seemed to be created at a time when dividend yields were low, being below 5% when it was made.
Batch 7 was conducted using an older style of strategy selection. Students were told to select stocks with a low PE, but at that time, it was a tossup between high dividend stocks and large market capitalisation counters.
The performance was unfortunate because the portfolio was built just before the dividend factor started to underperform for REITs. Historically, high dividend REITs were significant investments, but as more investors piled into high dividend counters, they began to lose their ability to outperform markets and growth REITs like Keppel DC REIT started to dominate after that.
The other characteristics were similar to all the ERM portfolios, characterised by a low beta.
When we examine the individual positions of the portfolio, we see the following :
The worst performing stock was First REIT. Going back to the historical record on how the portfolio was built, students were offered an 8% dividend yield on First REIT at that time and were still unaware of the issues from the sponsor. Had the high market capitalisation factor were used at that time, First REIT would not have been chosen for the portfolio.
The second worst-performing stock was Comfort Delgro, but that is a more forgivable decision as we’re not out of the pandemic yet, and transport counters will need more time to recover from their losses.
Overall, Batch 7 was not unlucky in terms of market timing. However, Batch 11, the batch that was unfortunate enough to be conducted just before the pandemic crash, did better than Batch 7. It was unlucky in that we were fixated on high-dividends as an investment strategy that would begin to lose favour after that. Another issue was that Batch 7 was a permissive class that rejected only 25% of the quantitative models’ stocks.
The course has become a lot more robust.
For example, our factor models now encompass 4-5 factors, so the performance would be impacted less if one factor becomes obsolete in the markets. We’ve also shifted our emphasis to more qualitative bias as students now employ the democratic process to reject 50% of the stocks chosen.
Best ERM Portfolio of 1H2021 – Batch 12
This batch was conducted in the middle of the pandemic crash in March 2020, and the portfolio generated the best returns of any batch achieved by the ERM program. It was also the first variant to be conducted online as the lockdown no longer allowed us to run face-to-face classes.
If you had invested about $17,000 into the portfolio, the portfolio would have earned $6,000, which would more than offset double the course fees. The initial investment could be as low as $10,000 if you opened a margin account and invested on leverage.
The hallmarks of a winning portfolio are all present if you look at the raw data. The beta was low at 0.72 even by ERM standards. When we constructed the portfolio, yields were so high. It could be built at close to 7% dividends.
The final XIRR of 30.77% speaks for itself.
The portfolio contains the following stocks:
Batch 12 was fortunate in that none of their investments had any losses, showing how advantageous it is to build a portfolio when there was blood on the streets. The quantitative model was also able to flag counters like YZJ for a 75% gain.
What is singularly unique about the portfolio is that the pandemic crash was so bad it was feasible to include some retail bonds in the mix. This was a one-off incident in the history of the program, and it required the yield to maturity of a retail bond to exceed margin financial costs for a leveraged account.
Obviously, if you have 20-20 hindsight, excluding the bonds would result in an even better performance.
How can we improve our performance in the future?
In summary, we learnt from our worst-performing batch that portfolios could perform quite badly if there an overemphasis on a particular factor that will soon go out of fashion.
The solution would be to add more factors and build models more robust against changing investing trends.
Our own program has evolved to account for this. Conversely, from our best performing portfolio, it shows the power of rushing headlong into markets in the middle of a crash. Market crashes are indeed an investor’s best friend.
