Category: DIY Investors

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. ↩︎

an anchor on the shore

News: No change to Bank of Canada Rates

RobHLeave a comment

Stop me if you’ve heard this before, but yesterday the Bank of Canada left interest rates unchanged — again. This is of interest to DIY investors only because it drives the kinds of rates you can expect from high-interest savings accounts and the like. I’ve updated the page that tracks some HISAs and cash-like ETFs over here.

The overnight rate remains at 2.25%. That makes 6 straight rate announcements with no change. The next one for the Bank of Canada happens on September 2nd. Next up is the US Federal Reserve on July 28,

a person is holding a package with their hands

What companies are in my retirement portfolio?

RobH1 Comment

I cover what I hold in my retirement portfolio every month (latest update is here), but as you can see, I only hold ETFs. What’s underneath the ETFs? Let’s take a look!

Calculating this is straightforward:

  1. Use the percentages I have of each ETF from the June update
  2. Look at the “Holdings” section of ETF or in the case of AOA/XGRO/XEQT, “Aggregate Underlying Holdings”1 and write down the percentages of each stock. I stopped when I accounted for 20% of each ETF.
  3. Multiply (1) by (2) and add them all up.

By doing this exercise, I can now account for the 20 companies that make up 20% of my retirement portfolio:

These 20 companies are really the tip of my personal investment iceberg — my aggregate retirement portfolio has at least 32,000 holdings, since that’s what’s actually inside AOA, the one ETF I own with the most holdings2. But the fate of these 20 companies will have an outsized influence on the overall performance of the portfolio. The only surprise for me is that there is only one company outside North America in the list. I would have expected one or two more.

It will be interesting to see what kind of differences I see once I complete kicking USD out of my retirement portfolio and go back to a list of only seven magnificent ETFs. I’m not sure there will be big changes, given that I’m keeping my asset allocations constant throughout, but I’m expecting some differences3. We’ll see come January 2027, I guess.

Just for fun, and given my recent article on the SpaceX IPO, I wanted to know how much I held of that particular company. AOA hasn’t included it yet4, but XGRO and XEQT have already. As of June 30, 0.02% of my portfolio is invested in SpaceX.

  1. In most cases, as of June 30. VFV only shows to May 30. ↩︎
  2. XGRO has 22000, for reference. I think the extras in AOA come mostly from fixed income (bond) products since they have broader geographic coverage. ↩︎
  3. The biggest changes won’t be visible since they occur in the bond side of my holdings. AOA and XGRO have very different approaches to this part of the market, but as it’s only 15% of my retirement holdings, they don’t show up as a “top” holding in my retirement portfolio . ↩︎
  4. And won’t until SpaceX is profitable. AOA includes IVV, which is an S&P 500 index fund, and the S&P 500 requires companies to be profitable before inclusion in that index. ↩︎