Estimate how your portfolio would have done in past crises and in a shock you design yourself.
Write down every holding you own today, with its current value in Singapore dollars. The test that follows runs on what you hold now, so this list is the starting point for everything else.
In Lesson 2.6, Value at Risk, expected shortfall and fat tails, Marcus worked out a Value at Risk of about S$12,900, what a bad month might cost him, and an expected shortfall of about S$17,300, the cost of an average bad month. Both figures are invented examples, and both came from five years of his own history, which contained no real crisis. That lesson's warning applies here: historical figures like these are a floor, never a ceiling. A stress test estimates what your portfolio would lose in a specific severe scenario that you choose, whether or not it appears in your own records.
A historical stress test takes a past crisis and asks what today's holdings would have done in it. You probably didn't own all of your current holdings back then, so for each one you pick an index to stand in for it, called a proxy. You then multiply each holding's value by its proxy's loss over the period and add the results.
Measure the proxy losses in SGD rather than in the index's own currency. In 2008 the US dollar rose against most currencies, which cushioned the fall in US assets for a Singapore investor. A US-dollar index loss would overstate your loss on US assets in that kind of period. Lesson 3.5, The dollar cycle and what a strong dollar does to Asian assets, explains how this works.
Marcus holds about S$202,000, roughly 80% in shares and 20% in bonds and T-bills (all his figures here are invented). His holdings are a world equity ETF, STI exposure, a Singapore bank, an S-REIT, Larkspur (a small chip supplier), a US chip designer, an SGD bond fund with a duration of 6.5, and T-bills. He picks two periods. The 2008-style crisis is a long, deep fall in shares worldwide, with credit drying up. The 2020-style crash fell at a similar speed but was shorter, and even some bond funds dipped for a few weeks. He gives each holding a proxy and finds its peak-to-trough loss in SGD. The losses below are invented, chosen to be of a similar scale to the real episodes, with the 2008-style figure first and the 2020-style figure second.
World equity: minus 45%, minus 28% STI: minus 55%, minus 30% Singapore banks: minus 55%, minus 33% S-REITs: minus 60%, minus 35% Small chip suppliers: minus 65%, minus 40% US chip designers: minus 60%, minus 35% SGD bond fund: plus 3%, minus 2% T-bills: plus 1%, flat
Applied to his holdings, the 2008-style case costs about S$81,150, about 40.2% of the portfolio, with about S$41,000 of that from the world ETF. The 2020-style case costs about S$50,000, about 24.8%.
History only contains crises that actually happened. A designed, hypothetical shock lets you test what history has not shown. For a portfolio like Marcus's, the most worrying case is interest rates rising sharply while shares fall, so bonds and shares are hit together. 2022 showed this pattern on a smaller scale.
His shock has Singapore and US yields rising two percentage points while shares fall 20%. He reasons through each holding. The world ETF loses 20% and the STI 15%. The bank loses 10%, less than the market, because higher rates help a bank's margin. The S-REIT loses 25%, because rates hit property values. Larkspur loses 30% and the chip designer 35%, because growth-style companies are hurt more when distant profits are worth less at higher rates, as Lesson 3.3, Real rates and breakeven inflation from inflation-linked bonds, explains. The bond fund follows the duration arithmetic from Lesson 2.5, Duration is the risk number for bonds: duration 6.5 times a 2-point rise gives a loss of about 13%. T-bills lose half a percent.
The total is about S$37,400, about 18.5%. This is the smallest loss of the three scenarios, but it is the only one where bonds lose money at the same time as shares. His bonds lose about S$3,900 while his shares fall, so the 20% he holds for protection does not protect him in this case.
Percentages make losses feel abstract. Writing the loss in dollars shows whether you would really hold through it, and Lesson 2.2, Drawdown and time under water are the risks you feel, makes the case that this matters more than any ratio. Marcus sets S$81,150 beside things he knows: it is about ten years of his yearly deposits, and more than he has made from investing in five years.
He then checks his rulebook, which says no scenario may cost more than 40% of the portfolio. That is about S$80,800 today (40% of S$202,000). The 2008-style case breaks the limit narrowly, S$81,150 against S$80,800.
If a stress test breaks your rules, change the portfolio now, on a calm day. Waiting until the crisis means changing at the worst prices under the worst pressure. Marcus makes the trims his limits already required: he cuts the bank, the S-REIT and the chip designer back to target and adds the S$14,720 of proceeds to the world ETF. He reruns all three tests on the changed portfolio. The 2008-style loss is now about S$79,200, about 39.2% and inside the rule. The 2020-style loss is about 24.3% and the rates shock about 18.2%.
Marcus's changes are small. Reshuffling within the share portion barely moves a crisis loss, because the loss comes mainly from holding 80% in shares at all. To cut it substantially he would need to lower his overall asset allocation, and that decision belongs in his investment policy, covered in the course Build and run an ETF portfolio, module 8, Write your investment policy and review it every year. He also notes the rates shock as a question for his next yearly review: whether to move part of his bond fund to shorter duration.
Look up actual index data for your own test instead of using the figures above. For the activity, you'll need your holdings in SGD, a proxy index for each, and that proxy's loss in two historical periods, from the index provider's data, plus one shock of your own design. Multiply and add for each scenario, write each loss in dollars next to your deposits and gains, compare it to your rulebook limit, and if a rule breaks, make the change now and rerun the tests.
Use an index as a stand-in for each holding and estimate your portfolio's loss in SGD for two historical periods and one made-up shock.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).