I set myself last year a goal of doing one blog post a month. I have an idea for an interesting series of posts related to forecast strength, which I'd hoped to find time to research and post about. However, I've been quite busy marking exams and pushing through the production trading code for pysystemtrade (it's now at the point where I can trade manually; auto trading is next. Expect a lot of posts once it's finished explaining how to use it for live trading).
So instead here's a post about systematic investing rather than trading. Although this blog is ostensibly for both traders and investors, I do tend to focus more on the former. To be fair this is because I try to only spend a few days a year rebalancing my long only investment portfolio, whereas I trade every day (or rather my computer does).
However the recent activity in the markets, and resulting churning of my portfolio, made me realise that I needed to seriously update my long only investment toolkit. At some points during March I was trading my investing portfolio every day. Even now, I expect to be rebalancing monthly as I gradually get back to my strategic weighting.
I though I might as well share with you the results of that updating.
So, in this post I:
- Come up with a list of curated ETF's to match the major asset classes, regions and categories.
- Use the principles in my second book, Smart Portfolios, to create a model portfolio of these ETFs.
- The model portfolio can be found in this google docs spreadsheet, and I hope to update it every month or so.
(I have to say this every time: This link is for a read only version of the spreadsheet. If you want to modify it then do not ask for edit permissions. Think about it - why would I allow anyone to do that? Instead download it or copy it inside google docs).
A curated list of ETFs
First a caveat: this is a list of UK listed ETFs. I am doing this for my own purposes, so it doesn't make sense for me to do a load more work for US or European readers. However hopefully you can use the principles here (and covered in more detail in Smart Portfolios) to choose equivalents. I used the excellent service, justetf.com, to research this list. Most of these ETFs are listed on European exchanges as well, so if you search the ticker you can find an alternative easily. In the US I would use etfdb.com.
Using my top down methodology I decided to use the following categorisation:
- US equities
- Beta (the whole market), High yield, Value
- European equities
- Beta ....
- UK equities
- Beta ....
- Asian equities
- Beta ....
- Emerging market equities
- Beta ....
- US bonds
- Government, Corporate IG, High yield
- European bonds
- UK bonds
- Asian bonds
- Emerging market bonds
There are no alternative assets here. In practice my alternative asset is effectively my trading account, which gives me exposure to a large variety of risk factors. This has a target allocation of 25% of my risk, but that isn't included here. I also have a ragtag of a few property and gold ETFs that I am excluding for simplicity.
I went for the Beta, High Yield, Value split because this is a particular bias I wanted in my equity portfolio. I did not want to go any deeper into the structure (eg down to equity sectors), as I didn't want this portfolio to end up requiring too much work on a regular basis. In practice I invest in individual UK equities and there may be other places in my real portfolio where I go into a more granular portfolio than implied here.
I also needed a global equity and global bond ETF to do my tactical asset allocation (of which more later).
Here is how I selected the ETFs:
- Select the appropriate category
- Where possible, exclude accumulating funds (since I wanted to use dividend yield as a valuation metric. In practice I might choose to invest in the accumulated fund depending on whether the ETF was in a taxable or non taxable account)
- Exclude leveraged funds
- Where possible, exclude currency hedged funds (unless a hedged fund was much much cheaper). I explain the problem with these in the book
- Choose the lowest TER (as readers of my book will know there is more to costs than just this single figure; nevertheless I didn't want to spend too much time on this exercise)
There are other selection criteria discussed in the book, but again in the interests of time I just stuck to the points above.
I was also interested in introducing a degree of diversification across providers, since historically my portfolio was very heavy in iShares and latterly Vanguard (thus creating some potential counterparty concentration risk). However I was pleasantly surprised by how much more competition there is in the market now. The 28 ETFs ended coming from 10 providers, and although iShares (7 funds) and Vanguard (5 funds) are still the most popular there is plenty of diversification here.
I was also pleasantly surprised by how cheap the market has become since I last did this for Smart Portfolios about 3 years ago. The average TER was just 0.22%, with 17 funds coming in at 0.2% or less.
Here is the list of funds. Note that there may be data errors here which I am not responsible for.
Unavailable indicates there was no suitable fund available with the given criteria. In some cases I've stretched the term 'Value' to include things like Quality if no pure value fund was available.
Note: IEFV is an accumulating fund; the yield I use will be taken from the distributing fund (IEDL) but do not buy this fund as it only has £2m AUM.
The model portfolio
Broadly speaking the portfolio will be done in my usual top down way, with the tactical weighting done using my favourite methods:
- Equity / Bond with a strategic 90/10 allocation (which equates to about 80/20 in cash terms). Tactical weighting, relative momentum (discussed in this post as well as the book)
- Regional allocation with strategic weights (equal split for bonds, less equal for equities).
- Intra-regional allocation with strategic weights (discussed below)
- 'Intra-asset allocation done as a single process using relative dividend yields'
'Intra-asset allocation done as a single process' might need some explaining. Don't worry, I will explain everything.
Let's dive in to the google docs spreadsheet
In sheet 'ETF list' I include the information from the curated list above, but also add columns for the price and 1 year dividend history. I then calculate the yield. The reason I do this is that the just ETF site only includes dividends as a premium product. So every time I update the sheet, I just need to change the price and the yield will update automatically. Every now and then I will need to change the 1 year dividend history (these normally pay quarterly, so 4 times a year should do).
The 'calculations' sheets deducts the TER from the dividend yield to get a net dividend (a crude way of handling costs), and then converts that to a Sharpe Ratio (SR).
In the sheet 'returns' I put the one year total return for my global asset class ETFs (I don't actually invest in any global ETFs). These are used in the momentum model on the next sheet 'asset' to calculate the adjusted asset class risk weights (cash weights are also shown for information, but not used).
The 'bonds' and 'equity' sheets do the allocation within each asset class as described above. I calculate the strategic risk weights inside each asset class across and then within regions. At this point I have a risk weight for every ETF inside it's asset class. I then use the relative dividend yield SR within the asset class to adjust those risk weights.
Finally in the sheet 'Total' I use the asset class risk weights (adjusted for momentum) to calculate my final risk weights for each ETF. These are then converted to cash weights.
Note: The SR scaling adjustments are different in each sheet. That's because I've recalibrated them so that they are more like my trading rules. Broadly speaking the SR differences have different meaning across different places.
In the asset allocation section, it's not unusual for equities to go up by 30% a year (about 2 SR units) with bonds flat; a SR difference of about 2 units. But within bonds you would be surprised to see one corporate bond regional ETF with a yield of 10% (about 2 SR units) and another with a yield of 0%. Similarly, you'd rarely see a developed regional equity ETF with a yield of 44% (about 2 SR units) whilst another had a zero yield. So I've changed the SR adjustments to reflect this.
Using the model portfolio
My own monthly routine will now look something like this:
- Update the ETF prices, and possibly the dividends
- Compare my own portfolio weights
- Consider selling and buying to match those weights if they are too far out of line with my current portfolio weights
The question of whether to buy or sell is complex and is covered in 80 or so pages of part four of Smart Portfolios, and I certainly won't be repeating it here!
At some point I hope to go back to annual rebalancing, but if the market moves a lot I can quickly fire up this spreadsheet and do an ad-hoc rebalance if required.
|I realise I wrote this whole post without including a picture, so here's a picture of my second book.|
I have all of your books which I have found very interesting reading, and have used the thinking of Smart Portfolios to 'make sense of' my random collection of investment and unit trusts, which is gradually morphing into a collection of ETFs with underlying rationale. Thanks! I have one question - in this piece you use some standard figures for volatility which I recognise from the book. Why don't you update with actual volatilities? Is it because it doesn't make much difference? On the other hand, it wouldn't take much effort for an annual rebalance either.
Too much effort. I haven't found a cheap or free data source yet that makes it easy to get these numbers.Delete
That makes sense. (Although I've found volatility figures on justetf.com, which you mentioned.) Thanks again for excellent books and blog.Delete
nice post and thank you for sharing the spreadsheet. I also have all your books and owe to them most of my financial education, I'm really thankful for that.
I noticed that there are some differences between the approach presented here and the one in Smart Portfolios (namely, the fact that the yield forecast is computed as a difference from the average instead of the median yield, and also that no momentum forecast is used). Is there a specific reason for these updates, or was it just to keep this post simpler?
Mean vs median wont' make much difference, but you're right I usually prefer the median. I'll edit the spreadsheet accordingly.Delete
In the book I actually say that I prefer to use yields alone for setting weights inside asset classes, and momentum for setting the weights of asset classes. So this approach is consistent with that opinion.
I follow the comment of Lorenzo. From the book I understand that the asset allocation is done with the momentum model only, and within the asset classes the combination of momentum and yield model is preferered. Also the example of Patricia (book p 351 and further) uses both models inside the asset classes. Could you point me where in the book you actually say to use only the yield model inside asset classes?Delete
You're right, I actually just checked back with the book and it says to use both models. Well it has been a few years since I wrote it. Perhaps a more accurate statement is that I decided only to use yields in the spreadsheet when allocating below the asset class level as it makes things much simpler.Delete
thanks for all the great content on your blog and books! Really appreciate it.
I have a question with regards to investing in a savings plan setting - as opposed to investing an initial amount. (Something like an ETF savings plan). I find it tough to reconcile some of the core concepts when managing a lump sum (e.g. broad diversification, risk control), with managing an investment annuity of sorts. Particularly if monthly inflow is below some threshold X, which would be needed to adequately relfect the desired asset allocation with each increment.
Hence, do you happen to know any good sources (books, blogs, podcasts, etc.) that cover this topic on a similarly advanced level as you usually do? Also, did you ever consider blogging about it? (or maybe you have and I just missed it, shame on me then...)
I don't know of anything, will add it to my list of things to think about.Delete
Thank you for your books and all the other great content.
I can see you've decided to add value factor ETFs into the portfolio while in the book you've concluded it's not worth to pay extra fee relative to market cap weighted alternatives. Did you find out that extra geometric return from these value factors are worth more expensive ETFs?
You're right that this is inconsistent with Smart Portfolios, but I thought it worth including these just for completeness and if people prefer (for example) to get their exposure entirely in the form of Beta or entirely in the form of Value they can just twiddle the spreadsheet weights.Delete
Thank you for updating this portfolio. I am following it for my various analysis. I have a couple observations/questions.ReplyDelete
1) The equity trailing / adjustment factor table has values of -0.08, -0.06 etc.. I was using the -0.8, -0.6 etc. table, which is in your book Smart Portfolios and also used for the “Asset” spreadsheet calculations. Is it an updated one, or a typo?
2) Bonds. The US Gov cell “Sharp ratio net of dividend” takes the cell H12 of “Calculations”. That refers to UK, not US Gov. It should be H15 instead, unless I am mistaken.
Also, more generally, I would be interested to know whether the rather aggressive 90-10 split to equity is related to a higher propensity for risk in general, or to concerns for the Bond asset class due to possible rate hikes and the way Bond ETF are built.
A relatively high allocation to equity HY seems to indicate you expect to get income from this source rather than from Bonds?
1) I've adjusted these to reflect the typical range of Sharpe Ratios experienced for the bond/equity dividend yields. I guess this is in the spirit of a 'forecast scalar'.Delete
The 90:10 split is my personal preference, and reflects my high tolerance for risk (so it's optimised for geometric return not Sharpe Ratio). Feel free to modify it.
Thank you for the amazing book (I going to re-read it), and thank you for the portfolio spreadsheet.
Could you elaborate on two things regarding your portfolio:
1. Why did you select the "Beta, High Yield, Value" split? Is there anything you would recommend to read about this in more details? (I could not find anything about it online, and it is not covered in the book)
2. It seems to be that a significant chunk of your portfolio is in UK ETFs, which seems to be contrary to example portfolios in the book. Is this a "home bias" or do you have some other reasons for allocating a significant portion of your portfolio to UK equities?
1. No this is pretty arbitrary. The point of the exercise is not so much 'Do this' as to let you play around with the weights to suit your own needs.Delete
2. Honestly? Part of it's home bias. But there is another explanation. I don't actually hold any UK stock ETFs (except as placeholders), I own UK shares. I don't like buying foreign stocks as the tax gets complicated (and it's hard to do so in the tax sheltered vehicles I try and use for stocks). So theoretically my UK portfolio should have a slightly better risk adjusted return since it will be equal weighted not market cap weighted, unlike foreign ETFs. Any value premium will improve this further. Finally, if I look at my UK stock performance it has been extremely good so I guess that also justifies a higher weight.
But again, feel free to change the numbers and make them more like my book.
Firstly, another sincere thank you in the list of comments - for your help building my foundations as self thought and end up working in finance (as tech, not a finance guy).
My question: how to smartly allocate portfolios in case of earning, holding and spending in multiple ccys (EUR and GBP in my case, and probably I am not the only one out there)? I see your point at ignore ccy hedging (and probably here I have even more reasons to ignore it). Ultimately what matters is risk so converting allocations into one would probably do the job.
Also earning&spending may not be of same ratio across ccys. Would that justify playing with dist and acc funds (as also on tax, captial gain should be spread across years through dividends)?
PS: am I wrong or AAAG has done -8% 202104 -> 202204 ?
I'd say the theoretically correct answer is to hold your portfolio in different currencies in proportions roughly equal to the present value of your future spending. And in practice to get as close as possible to that.Delete
AAAG has indeed got hammered, there's been a tiny bit of a global bond sell off.