Results are for informational purposes only and not financial advice; past performance does not guarantee future results.
Relevance and dislcaimer: this is not a groundbreaking tool. There are some other portfolio visualizers/analyzers out there, probably vastly more precise, scientific, and comprehensive than the one proposed here. There is some level of arbitrariness, and it is simplistic. However, it is based on solid data-backed facts, free, and takes just a few minutes to use - it's meant to be a quick sanity check, not the backbone of your strategy. Results are for informational purposes only and not financial advice; past performance does not guarantee future results.
Privacy and data:
- User data is not sold, traded, nor monetized under any circumstances.
- I regularly and systematically delete all stored data on google sites to ensure no long-term retention of user information.
- This program is a personal project run with a strong commitment to user privacy.
- While I strive to protect your information, I cannot guarantee entreprise-level security.
- Avoid entering personal information beyond what is strictly necessary (especially in the "Investible portfolio allocations (instrument name - %)" field).
- The app is provided "as is" without formal data protection guarantees.
Tool: a quantitative analysis program that classifies your portfolio as [good, medium, bad] for the dimensions [tax efficiency, complexity, diversification, inflation risk, recession risk] based on your particulars.
Builder's background: I currently work in quantitative trading. I have experience working as a quant analyst in wealth management funds in Switzerland and Hong Kong.
Motivation: I am very risk averse, and a few years ago I wanted to benchmark what I was doing against industry standards to see whether it matched my profile or not. The idea was to have a sanity check and a way to spot gross mis-allocations rather than a complete "perfect" robo-advisor/allocation algo. I am from a very small rural town and as I was advancing in my career I had a few relatives asking me to help with their finances - which is when I started generalizing the model to fit different life situations and scenarios. Recently my girlfriend was away meaning I had some time alone at home so I thought I'd try and generalize the model to make it more accessible - which is what I'm presenting here.
Methodology: I start from what an average "result" for this dimension is among US retail investors, I assume a distribution around this point, then look where your portfolio stands in this distribution. If you are in the top tercile you are "good", if you are in the middle tercile you are "medium", if you are in the bottom tercile you are "bad". The base allocation is always determined based on available literature but, more often than not, I have to make assumptions regarding the distribution. These assumptions are based on industry standards.
Example: a 2024 study by JP Morgan Chase (https://www.jpmorganchase.com/institute/all-topics/financial-health-wealth-creation/retail-risk-investors-portfolios-during-the-pandemic) of 500,000 US retail investors found the average number of positions in a portfolio to be 7. There are other articles pointing to this number, but I'm keeping things simple here. A distribution is however also defined by its skewness and standard deviation which are really hard to find in this case - so we have to make an assumption. For this case, a reasonable one would be SD=3.5 and skewness=1 (skipping explanations in this post). I can now tell you how your portfolio compares to others. The "simplicity" dimension is one of the most straightforward to calculate and isn't made dependent on personal factors. The other dimensions follow the same principle but have some level of adjustment for personal factors (once again, based on what the literature tells us) - which is done by shifting the average value based on the scenario in which you fall. These shifts are not arbitrary but, once again, based on what I can find in the litterature.
Robustness: I have not run any robust back-test nor have definite proof of anything; I have been using this approach (or, variations of it) for a few years and it has been pretty useful to me. I sometimes get weird results with edge-cases but, once again, the point of it is to call your attention on general factors you may have overlooked.