Category: Tools

ai chat interface on laptop screen

Mini-Review: Questrade MCP support for AI agents

RobHLeave a comment

I’ve had AI on the brain lately; it’s partly because I’m always curious about new technology — but I also recognize that my own portfolio management is more labour-intensive than it needs to be. That’s partly because I enjoy the tinkering, partly because I like things done in a certain way. My spreadsheets help a lot with making things more mechanical, but I still have to manually update them to get useful results1.

Now, I’m far from an expert on this sort of thing, but I’ve been experimenting with Anthropic’s Claude2 for a few days, getting familiar with how it works, and downloading the desktop client to my laptop.

Coincidentally, this week, I noticed a new little popup in my Questrade portal this week which promoted a new capability, namely MCP support: How to Connect Questrade to AI Tools with MCP3.

“MCP” stands for “Model Context Protocol”. Gemini helpfully describes it as “… a universal USB-C port for artificial intelligence”. Clear enough. In other words, MCP allows Claude read-only access to my Questrade account, which sounded rather useful and instead of pages of documentation (typical for an API integration), I got a three step process to make the connection.

  • copy the MCP URL from Questrade
  • Plop it in Claude’s Custom Connectors setting
  • Login to Questrade and permit access from Claude

I successfully followed these three steps in roughly 60 seconds, and next thing I knew, Claude was able to give me a list of my Questrade accounts and balances4.

That’s easy stuff, admittedly, so I figured I would ask it to please calculate the number of XGRO shares I would have to sell in order to make my monthly RRIF minimum payment. (This is something I actually do every month; you can see the steps I have to take to get paid in retirement over at What do you have to actually DO to get paid?

Claude was happy to help me — and it knew enough to ask me how old I was and what the value of my RRIF was at the beginning of the year. (This is how RRIF minimum is calculated, if you’re new to how RRIFs work, you might want to give Demystifying RRIFs a read).

But since I already know what my monthly RRIF minimum payments are, I told Claude to please use those. And it correctly calculated how many shares to sell based on the live cash balance in the RRIF account and the live quote for XGRO.

Claude immediately offered to set up a monthly schedule to do this…but, unfortunately, Claude scheduling is one of those capabilities that requires a not-free Claude subscription. The entry level subscription would be $24/month, or around $300 a year. In the grand scheme of things, that’s not a ton of money, but as a rule I’m a bit reticent about subscription-based services since they have a way of slowly adding up and becoming a drain on your hard-earned savings5.

That was a silly, easy question to ask, and failing that test would have resulted in mockery.

So I asked Claude a more difficult question, namely to tell me how much Canadian Equity exposure I had across my entire portfolio.

Claude was quickly able to identify my 100% Canadian equity holdings (currently VCN, XIC and HXT), and correctly pointed out that my all-in-ones (XGRO, XEQT, ZEQT) were also sources of Canadian equity. With a confirmation to proceed, Claude was able to quickly create an aggregated table that showed all my Canadian equity holdings6.

But there remained a problem. Claude failed to uncover the Canadian equity portion of AOA, which is a USD-based 80/20 fund. My own calculation put the Canadian equity portion of AOA at 2.64% so it’s small, but not zero. But since I have a pretty large (if dwindling7) portion of my savings in AOA, this is a signifiant miss. I pointed this out to Claude who agreed (naturally) and fixed it8, along with a helpful and accurate observation9:

“Notes on AOA specifically: it’s a US-domiciled fund, so its 3.03% Canada weighting is small and comes through its international-developed-markets sleeve rather than a dedicated Canada allocation — nowhere near the home-country tilt you get from XEQT/XGRO/ZEQT. Its USD value was converted at the last USD/CAD rate we pulled (1.4087); today’s rate may differ slightly.”

Anyway, this experiment proved to be interesting, but without additional automation10, not particularly helpful to me. The key gaps right now are that it does not currently provide access to my spouse’s accounts (this I blame on Questrade’s design and decisions regarding how to treat TA accounts), and the lack of an MCP connector to Wealthsimple11 (where I have some retirement holdings). Have you connected your DIY accounts to an external tool? Tell me about it at comments@moneyengineer.ca!

  1. Questrade (my primary broker) has had API access for quite a long time, and I kept thinking I would figure out how to make it useful but I do find the security aspects of messing with this a little daunting. And there have been enough negative posts about its reliability and availability it makes me wonder whether it’s worth my time. ↩︎
  2. The free version, naturally. I am still a cheapskate. ↩︎
  3. It appears that the promotion of this capability is not showing up in everyone’s account. Whether it is being selectively enabled on a per account basis is unknown. ↩︎
  4. Well, not all of them. I’ve long complained about how Questrade handles accounts for which I have Trading Authorization (i.e. my spouse’s accounts). There’s no unified view from my login and this also shows up in my interactions with Questrade via Claude. Claude also doesn’t know about the accounts for which I have TA. ↩︎
  5. “Pay yourself first” (cfr The Wealthy Barber) cuts both ways. If you set aside money for savings before you ever see it land in your bank account, it is a bit like magic over time. If you are billed for a subscription service that automatically renews, it’s like magic for the provider of the subscription service. They’ve done the work to land a client, and now they can just watch the monthly revenues roll in, knowing that automatic renewals are unlikely to get noticed immediately, if ever… ↩︎
  6. It appears Claude took a look at the real time Canadian equity exposure of (for example) XEQT. I don’t myself bother doing this because XEQT (for example) has a stated objective of keeping the Canadian equity portion of its fund at 20% and it rebalances periodically to do this. So 20% is good enough for me. ↩︎
  7. I’m actively kicking USD assets out of my portfolio: Kicking USD out of my retirement portfolio ↩︎
  8. It set the percentage at 3%. ↩︎
  9. This wasn’t new or helpful to me since I own AOA, but I include it just to show the sorts of things Claude mentions. ↩︎
  10. Which may be possible; I’ll have to play with it a bit more ↩︎
  11. There are some 3rd party integrations I’ve seen, but I’d prefer it to come from the source. ↩︎

close up shot of rusty gears

AI and Financial Planning

RobH1 Comment

Although a fair bit of my retirement portfolio is wrapped up in AI companies, that’s not what I’m talking about this time. I’ve been exploring the use of AI in helping make investment decisions. And it’s a rapidly evolving and fascinating place, and ultimately, I think it’s a huge win for the DIY investor. There are many ways for you to use AI in your financial decisions today — you can use free and generic AI, a commercial financial product that uses AI under the hood, or a DIY financial advisor built from scratch using Claude. The choice is yours!

Free AI

Here I’m talking about using free tiers of AI to ask basic questions. Excellent for the lazy investor. Yesterday, for example, I asked Gemini1 “What’s the best foreign ETF for Canadian Investors”?

Gemini’s response when asked, “What’s the best foreign ETF for Canadian Investors”?

I rate this response a solid nine out of ten:

  • XAW and VXC are what I was actually after; they are both ETFs that allow you to invest in the world ex-Canada with one ticker. Like all such funds, they tilt pretty heavily towards the US (about 63%) because that’s the biggest market by far.
  • VIU is also a valid response, but excludes the US market. My question wasn’t overly specific.
  • VFV is on less solid ground as a response since it’s an S&P 500 ETF, meaning it only has US exposure. I suppose it’s technically a “foreign ETF”.
  • The answer preventing the perfect score here is XEQT, which, although featuring in the Magnificent Seven ETFs, is a wrong answer IMHO because it holds Canadian Equity.

I like Gemini because it always provides links indicating sources, and checking the source is always a good practice. This is a big time saver. I tried the same query on ChatGPT and Claude. ChatGPT’s answer was very close, but Claude figured that by “foreign” I meant, “Everything outside North America”, not “Everything outside Canada”. An interesting assumption…

Anyway, my experience using free AI for financial advice like this is generally pretty positive, but I’m using it like an assistant, not an advisor. I even see moneyengineer.ca show up as a source now and then, so what’s not to like 😉

Products that use commercial AI under the hood

Here you have products like Gilded2 or Truthifi3. Truthifi reminds me a bit of Passiv, whereas Gilded looks a bit like Optiml, products I’ve talked about in these pages. I’d expect to see more and more of these kinds of companies try to make a revenue stream by wrapping up commercial AI packages in a way that makes it easy for the AI neophyte to benefit. I very quickly took a look at Gilded, but the free version does not do very much at all, and targets people who haven’t retired yet. Truthifi looked more promising to me but only gets useful if you directly connect it to your brokerage accounts, something I’m not willing to do until I read a bit more about it.

DIY Financial Advisor AI

While researching this article, I was surprised to find an open source project that allows DIY financial advice based on Claude AI. The Canadian Finance Planner Skill on GitHub looks to be a comprehensive piece of work that targets Canadians, but requires a Claude subscription including Cowork to be useful. And a little bit of tech-savviness to get it running, but I presume anyone already paying for Claude Cowork qualifies.

I may have to invest some time and money into this one, it looks absolutely intriguing, covering the entire gamut of financial concerns from budgeting, to estate planning, insurance, taxation… In their words “Think of it as having your own AI financial coach on call who happens to know every corner of Canadian finance”. It’s a promise that seems a bargain for the cost of Claude Cowork at $24 monthly.

What about you? Are you using AI in your own financial decisions? Tell me about it at comments@moneyengineer.ca!

  1. Gemini is Google’s own AI and, being lazy, the one I generally use because it doesn’t require anything other than typing the question directly into my browser bar. I don’t use ChatGPT or Claude as much for this reason. ↩︎
  2. I found out about Gilded from a recent Globe and Mail article by the perportedly retired Rob Carrick. ↩︎
  3. Discovered by Googling “personal finance canada AI” ↩︎

snowball toss in los penitentes mendoza

Review: Snowball Analytics

RobHLeave a comment

Snowball Analytics (link here) is a tool that markets itself as a “Simple and Powerful Portfolio Tracker” with “automated dividend tracking”.

I don’t remember when or where I first kicked the tires of this tool but I figured I’d give it another look. I’m only using the free version of Snowball, since my multi-asset tracker tracks my portfolio and also I don’t really care about dividends.

Startup

Snowball gives you full access to all its feature set for 2 weeks so you can really see if it’s worth paying for. That’s a plus in my view. But even the free version allows you to manually create a single portfolio with up to 10 holdings in it. All the paid-for versions include direct connection to your brokerage, and the ones I’m most familiar with (Questrade and Wealthsimple) are both supported.

For fun, I created a public portfolio that’s sort-of1 based on the percentages in my June 2026 portfolio update. You can see it here. What you see on these public portfolio pages is similar, but not identical, to the screens shown if you create your own login.

Dashboard

Gives a high level overview of what’s going on in the portfolio: what it’s worth, how much you made today/all time, and, atypically, Passive Income aka yield.

One weird thing Snowball does is insist on putting your assets into “Categories” which you can define yourself or let it assign them automatically. It doesn’t really know what to do with ETFs and so it classifies them all as “Funds”, which is about as helpful a category as “Unclassified”. I flagged this with support, but I don’t really expect much. If you click through the pie charts, it does show what’s underneath, so this is mostly a cosmetic issue.

Pie Chart, like the one I show every month

Also shown on the dashboard are the timing and predicted value of future dividend payments, which is nice.

Future passive income from dividends shown on the Dashboard

Dividend tracking

Here we see what Snowball’s focus is. The dividend screen is quite comprehensive, and will warm the heart of the dividend-focused investor. I can’t really comment if the numbers presented on this screen are sufficient for that audience. My portfolio focus is on absolute growth, and I’m completely indifferent as to whether that growth comes from dividends or unit price appreciation. 

Anyway, I’m happy to hear whether others out there find this sort of dividend analysis useful, let me know at comments@moneyengineer.ca!

Analytics

Here you have a bunch of views of the makeup and performance of your loaded portfolio. For the holder of all-in-one ETFs, these pages are a mixed bag of helpful and not-helpful. On some screens, Snowball isn’t able to look deeply enough into what’s actually inside an all-in-one and thus produces unhelpful results. For example, I know my portfolio is 80% equity. Snowball has other ideas, especially when it comes to Canadian ETFs:

Asset class distribution has obvious problems

There’s problems here:

  • XGRO/XEQT/HXT are missing asset classes. Their percentages don’t add up to 100%
  • A lot of Canadian all-in-one ETF asset classes are classified as “Other”, which is wrong. XGRO (for example) should line up exactly as AOA does, with 80% stocks and 20% bonds, more or less. For XGRO, the bond percentage is right, but the other 80% is suspect.
  • XIC/VFV/HXT are 100% Canadian equity, but Snowball seems lost here.

The performance widget seems to be a lot better; here you can backtest your portfolio against a bunch of benchmarks. Unfortunately, none of the included benchmarks includes bonds, but you can see how my portfolio stacks up against a global equity pool, namely the MSCI World index:

Growth charts are helpful to see how your portfolio compares to a benchmark

I did see a few problems here, too:

  • Not all of my portfolio holdings have been around for (for example) 5 years. So how is the tool generating 5 year return charts?
  • When you select “all” as a timeframe, my portfolio shows a one year performance. Puzzling.

The fact that my portfolio lags the MSCI World is unsurprising since my portfolio has 20% bonds, and the MSCI World has none. Bonds add stability to a portfolio at the expense of growth. Sometimes I wonder if that stability is really worth it.

Conclusion

Snowball, like many tools, struggles with understanding what’s actually inside an all-in-one ETF like XEQT or XGRO. For me, that sort of shortcoming is a deal-breaker. If you were a holder of individual stocks that cared about passive (dividend) income, it might be worth your time.

What do you think? Let me know at comments@moneyengineer.ca!

  1. “sort of” because I have more than 10 holdings in my portfolio. I got most of them, and scaled it to start as a 100k CAD portfolio. ↩︎

crop sportswoman checking information on tracker

Mini Review: Portfolio Tracker

RobHLeave a comment

Over the years I built my own portfolio tracker (the multi-asset tracker) and I’ve shared it on this website1. I’ll be the first to admit it’s not terribly user-friendly, which is somewhat understandable since I built it for myself.

If you want to take a look at another Google Sheet tracker that is fully documented and more user friendly, then you might want to take a look at Portfolio Tracker.

I’m not sure where I first encountered this tool; possibly on Reddit or perhaps the Financial Wisdom Forum.

Anyway, the brains behind this tool are substantial, and I bow to the organization and wizardry of the author. For people who adhere to the idea of asset allocation as a way to make investment decisions, you won’t find a better fit.

So, in a nutshell, what does it do?

  • It allows you to track the value of your portfolio across multiple brokerage accounts using multiple data sources for near-real time quotes2
  • It allows you to set up your own asset classes to track and to set individual targets for each, both at the portfolio level and at the account level3.
  • It will show you how far off you are from your targets and make high-level recommendations for where (what account) to buy/sell to get back on track

One very nice feature of Portfolio Tracker is that a given asset can be divided up into multiple asset classes. For fans of asset allocation funds (like me) this is a critical feature. With this feature you can accurately depict that (say) XEQT is 25% Canadian equity and 45% US Equity.

Once I figured out the terms used in Portfolio Tracker, it was pretty straightforward to enter my own portfolio across the 5 RRIF, 2 TFSAs and 3 non-registered accounts.

What confused me at the beginning about Portfolio Tracker is that it has more layers than I’m used to:

  • It starts with asset. Like XGRO, AOA or ZMMK. So far so good.
  • Assets belong to one (or more) asset classes. If more than one, the percentage has to add up to 100%. Asset classes are where I focus my attention: Portfolio Tracker has more of them than I need4 by default but you can define as many or as few as you like.
  • Asset classes in turn belong to a unique Asset Category)5.; a given Asset Category can be the parent of multiple asset classes (e.g. US Small Cap and US Total Market asset classes are both included in the US Equity Asset Category)
  • And asset categories roll up into Parents (stocks, bonds, short-term)

One minor point of confusion to the Canadian user is the inclusion of TIPS which is a uniquely US investment vehicle. In Canada you can buy real return bonds or buy ETFs that hold TIPS if you wish. I don’t bother with either myself.

The only limitation I could find in this tool was that it didn’t support multiple currencies. If you hold USD assets (as I do), that is a very serious limitation, but one that I could (and did) correct myself pretty easily with a few changes. When I sent a note to the provided support email on that limitation, the author promptly replied and admitted it was not the first time someone had asked about it.

I recommend this tool as a user-friendly introduction to tracking your own portfolio.

  1. I’ve built a new version based on pivot tables; on my to-do list is to make it generic enough to share. The new design lifts some ideas from Portfolio Tracker, in fact. ↩︎
  2. At one time my own tracker did this too but as it requires web scraping code it breaks pretty frequently, and in the mean time googlefinance() has become much more reliable ↩︎
  3. I’ve only really cared about portfolio level, but I have some broad rules about what goes where at the account level. TFSA: Equity only. RRIF: only place outside of the cash cushion where cash can be held. And the only place I hold bond funds. Non registered: equity only. ↩︎
  4. It divides US Equity into small cap (“US Small Cap”) and total market (“US Total Market”). This particular example I found a bit weird since logically “US Small Cap” is normally considered part (albeit a very small part) of the “US Total Market”. ↩︎
  5. Asset Categories for me are a level of detail I don’t need. If you set them to be the same as your Asset Classes then they effectively aren’t used. Although probably best to given them “AC”names so you don’t get mixed up, e.g. US Equity asset class belongs to US Equity AC asset category. ↩︎

abandoned plane wreck on icelandic landscape

PortfolioPilot Review: Any better?

RobHLeave a comment

I last took a look at PortfolioPilot a year ago, and since their email frequency seems to have ticked up, I figured I’d give it another look.

If you don’t feel like looking at the old review, the TL/DR is “good promise, errors in the data make me hesitant to recommend it”.

And guess what? Nothing has changed in that regard. All the lovely formatting in the world, all the tailored recommendations, all the graphs and charts are pretty much useless if PortfolioPilot can’t accurately reflect what’s underneath the ETFs in my portfolio. 

And, as I’ll show with a few examples, both of which I reported to support, the errors are not minor.

Now, bear in mind that I’m a big fan of all-in-ones, and these are essentially “funds of funds”, so if you DON’T hold these kinds of assets, then your perception of PortfolioPilot’s usefulness might be quite different. For me, if PortfolioPilot doesn’t have an accurate handle on what I actually own, it can’t have an accurate handle on anything else: the risk calculations, the forecasts, the recommendations — all are suspect.

Problem number 1: PortfolioPilot doesn’t know what’s in XGRO

So here’s the breakdown PortfolioPilot shows when you give it a portfolio with just XGRO in it:

PortfolioPilot’s Assessment of the “By Holdings” look-through of XGRO

The first few entries are accurate, per the XGRO product page. As of April 17, 2026, it reports:

  • 36.31% in ITOT, the iShares Core S&P Total US Stock Market1
  • 20.31% in XIC, the iShares Core S&P/TSX Capped Composite2
  • 20.21% in XEF, the iShares Core MSCI EAFE ETF3

So it’s got about 3/4 of the holdings right so far. PortfolioPilot now reports that XGRO holds 12.3% in a BondBloxx ETF. This is dead wrong. I’ve never heard of it, and “BB rated USD High Yield” sounds rather speculative, not something I’d want to invest 10% of my hard-earned money in. How can this kind of error creep in? My friend google gives a hint for those in the know:

Google Gemini’s take on BondBloxx BB Rated HY Corporate Bond ETF

The clue? The symbol of this ETF per Google Gemini is “XBB”. XGRO does not hold XBB on any US market4. XGRO does, however, hold XBB.TO, which, admittedly, is also a bond fund, but its description is a lot more boring:

Why XBB? Low cost, broad exposure to the Canadian investment grade bond market

XBB by Bondbloxx is clearly a much different animal than XBB by Blackrock, and that’s a pretty big miss.

There’s more to shake your head at, though. PortfolioPilot has “other equities” sitting at 7% of the portfolio. This is also wrong. XGRO is an 80/20 fund, which means it’s 80% equity. 75% of it we’ve already talked about (ITOT, XIC, XEF), and the other 5% is the next line in the PortfolioPilot report, namely XEC, the iShares MSCI Emerging Markets fund. So by PortfolioPilot’s estimation, XGRO is about 88% equity, which is off by 8 percentage points.

Anyway, two pretty serious errors for a fund that makes up 15% of my retirement portfolio.

But perhaps it’ll do better with a fund based in the USA?

Problem #2: PortfolioPilot doesn’t know what’s in AOA either

AOA is even a more important fund for me at the moment: it’s 50% of my portfolio, give or take. So what does PortfolioPilot have to say about what’s underneath?

PortfolioPilot’s Assessment of the “By Holdings” look-through of AOA

I don’t really know where to begin with this breakdown. Perhaps it’s faster to point out what it has right:

  • iShares US Aggregate Bond ETF (IUSB) percentage is correct

The rest is pretty much random:

  • PortfolioPilot claims the top holding of AOA is a Vanguard fund. Given that AOA is a product of iShares (a major competitor of Vanguard) this seems rather unlikely. And it is, I assure you, completely wrong.
  • PortfolioPilot correctly says that AOA holds the S&P 500 ETF (IVV) but the percentage is totally wrong. Per the AOA product page, it sits at about 45%
  • The other three major holdings (namely “other”, iShares Real Estate and SPDR Gold) are all wrong. AOA holds none of these.

Of course, my test is a very small sample, but important to me. If you do use PortfolioPilot, I’d make very sure that it accurately reflects what you actually own; otherwise the rest of the service cannot possibly work correctly. I’ll let you know if/when the situation at PortfolioPilot changes, but until it does, I’m not trusting it even at its free tier.

  1. AKA “US Equity” to my way of thinking of asset allocation ↩︎
  2. AKA “Canadian Equity” ↩︎
  3. AKA “International Equity” ↩︎
  4. Recall that PortfolioPilot is a US based tool that happens to support Canadian-listed ETFs, but to find them you have to add “.TO” to the end of the Canadian symbol ↩︎