I was going to title this post something like ‘bond quant performance during the bond meltdown’ or something a bit more hyperbolic than the current title. But there’s enough hyperbole in the financial news without me adding to the mix. Just google ‘bond meltdown’ and you’ll see what I mean. Anyway, in this post I want to take a quick look at how the Bond Quant model has performed during the recent bond selloff. I’ll look at performance from the end of 2015 through yesterday.
First, let’s take a look at some benchmarks for broad bond portfolios. Here’s the YTD performance for BND, AGG, BNDX, and IEF (pretty much the benchmark).
Bonds were having quite the year until summer. The bond market peaked in summer, July 8th to be precise, not before the election. Now, IEF is negative for the year, BND or AGG are at just under 1.5% for the year, and international bonds are about 3.5% YTD. Now, lets take a look at the basic Bond Quant which holds the top 3 bond ETFs, ranked by 6 month momentum.
Note: ignore the benchmark in the screenshots below. P123 forces you to choose a benchmark and there are no bond benchmarks in the tool.
The Bond 3 portfolio us up 6.05% YTD with a drawdown of just over 6% from its high in early July. Not too bad. Basically up to July, performance was mainly driven by long treasuries which then transitioned into junk and emerging bonds. The more aggressive version of the bond quant model only holds the top holding, Bond 1. Here is its performance throughout 2016.
Even better with a YTD return of 9.74% and a max drawdown of around 5%. Here you can more clearly see the switch in performance drivers since the portfolio has the opportunity to be 100% long the best performing ETF. There have been only two trades since the beginning of the year. The portfolio was long high yield munis at the beginning, switched to long treasuries in late Feb, and went to junk bonds in August. Pretty simple, yet effective, without all the hype.
In summary, as we should expect from a quant momentum model, the bond quant model has generated good performance relative to its benchmarks while avoiding the worst of the selloff in the bond market.
Note: all the quant models and performance charts are done with Portfolio123.com .
23 thoughts on “ Bond quant performance during the bond selloff ”
Paul, I’ve been trying to work through implementation of these strategies. The problem I have is that my retirement accounts don’t allow margin, and don’t allow buying/selling on the same day.
So, anytime I want to rebalance or reallocate in response to a signal, I can’t do it other than by selling and then waiting three trading days to take the new positions.
Anyone else have to deal with this? I think it definitely “matters” to not have your money invested 15% of the time, and to have to make your buys at least three trading days in. It’s frustrating.
The (other) frustrating aspect is that high-turnover strategies seem like good candidates for deferred-tax accounts since you are going to have a bunch of short-term capital gains.
Does anyone account for this by using low-turnover strategies in these 401(k) and IRA accounts that you can’t rebalance in quickly (and if so, which ones)?
That wold be very irritating indeed. I don’t have that issue in my IRAs but they’re all rollover IRAs. Can you switch brokerages?
Thanks, Paul. I use Schwab, who is the custodian for my 401(k). I can’t change that, but I can change the broker for my and my wife’s other IRAs. Which broker do you use for IRAs?
TD and Wells Fargo. Would recommend TD, not Wells.
I was running one very similar in my 401k with a 8-week rebalance because of Vanguards buy/sell rules (which could be violated if the portfolio sold something and tried to rebuy it 4 weeks later). Unfortunately it looks like bonds start dropping, having a shorter rebalance period works a lot better! My tiny backtest showed no difference, but we’ve been in a solid bond bull market…
I think your system is more robust anyway.
Sorry, I meant that I am running one (still). It showed 4.4% total for the year, but is flat since when I started in May.
Shorter rebalancing is probably better though. Have you noticed 3-month (even shorter) momentum signals work better with bonds? Maybe there’s not enough data to be sure though?
4 week re-balance works the best. weekly re-balance is a bit worse but may be emotionally satisfying.
6 month momentum works best over the long run. of course, there are period where shorter momentum signals do better but usually the increase in trading is not worth it.
Hi Paul, Congrats for your blog, it´s impressive.
One question, What software is more friendly for developing fundamental systems?
Thanks Charly. I use portfolio123.com I find it to be a good mix of a powerful tool, yet not that complicated. It does require some basic programming skills but its not too bad.
Thank you for your answer Paul,
Is there any tutorial or similar to develop the quant fundamental systems in portfolio123?
There are some how-tos and references at Portfolio123.
Good work Paul. And yes, it’s probably better you didn’t use “Meltdown” in the title 😉
Are you using a timing mechanism (e.g. – 200 day SMA, 12 mo abs momentum, or 6 mo abs momentum) on this bond quant portfolio. Or, are you just using the 6 month momentum of the bonds without any timing? Thanks.
Hey B, the system is described in the original post. It uses the 6 month absolute performance as a crash filter. 6 month total return has to be greater than zero. That has helped the top 3 system this year. It had only 1 ETF as a holding for a bit. Now its up to two since EMB crept back up.
I have a question about your Quant portfolios in general(VC2, TMC, Utilities, Consumer Staples, etc).
If you were going to pick one time a year to make and remake the portfolios each year, do you think there is one best time? For example, every January 2 of each year to reconstitute the portfolio. Thank you.
Historically, from What Works on Wall Street, January has been the best month. In recent years, last 20 or so, Sept has been the best month. With Jan and Dec being the runners up. In general, I would say anytime in Q4, plus January, are good times. I spread mine out so I don’t have too much to do in one day, or week, or even month. I even had a couple of quant ports I started in June, historically not good, which did just fine, which I have since re-aligned to the Dec-Jan period.
Thanks for your update on this excellent, novel strategy.
I had a few questions I’d appreciate your help with. And I’m new to your blog, so forgive me if you’ve covered these before.
How did you select the number and various ETF categories in this strategy? Did you consider correlations? In a rising interest rate environment like the 70’s, will performance be marginal? What role might this TAA bond strategy play in one’s overall portfolio?
Thanks so much!
I basically chose a broad swath of the bond market categories. Yes, but in general all bonds are correlated very highly. The exception would be the stock like high yield bonds and maybe emerging market bonds.
In a rising rate environment, like from 1940 to 1981, intermediate bonds still did OK, and more importantly provided an effective diversifier to stocks. Intermediate gov’t bonds do this the best.
I use this strategy as my bond allocation in my portfolio. No buy and hold bonds for me.
One more–how about adding a convertible bond ETF, like CWB?
Would you list the bond ETFs you are using in the Bond Quant ETF strategy. From what I read looks like it would be TLT, JNK, IGOV, SHY, IEF, TIP, LQD, MUB. Is this same group of symbols you are using to select the top 3 funds or top 1 fund every month with a 6 month look back period?
Also have you checked the results for 2015?
How do they compare with 2016?
Yes, Yes (search for quant performance 2015 and you’ll see the results). See my latest post for 2016 results.
Any plans to keep up the Bond Google sheet since this is not available on Allocate Smartly?
Thanks for your excellent, ongoing work.
I like the idea of using a quant bond strategy as an element of my investing portfolio and have looked at your posts on this before. But when I try to get down to the details and reproduce the results you show I have not succeeded with either 1 ETF or 3 ETFs traded each month. To keep things simple, below are the results I come up with for the last two years trading just 1 ETF each month.
Looks like you are confirming the list of ETFs you use for this strategy is TLT, JNK, IGOV, SHY, IEF, TIP, LQD, MUB. Trading 1 ETF based on past 6 months relative momentum at the close on the last day of each month I get the following results.
For 2016 I get 15.17% CAGR with max draw down of -3.11% (better than what you show) with the following trades: Jan – Apr TLT, May – Jun IGOV, Jul TLT, Aug – Dec JNK (also different from what you indicate).
For 2015 I get -5.31% CAGR with max draw down of -15.07 (quite a bit different from your results) with the following trades: Jan – May TLT, Jun – Aug JNK, Sep SHY, Oct IEF, Nov – Dec MUB.
I would really like to get the consistently positive returns you show for this strategy. Please help me identify where I’m going wrong trying to reproduce your results.
I have been a bit all over the place with this strategy especially with the ETFs I chose to include. And it has changed since the original post. So here is the original strategy and the modifications I’ve made since then.
The ETFs I use in the original published bond strategy are IEF IGOV JNK MUB SHY TIP TLT VCIT . Ranking is 6 month total return with an absolute return filter of 6 month total return > 0. Max position size for top 1 is 100%, and 33.3% for top 3. Re-balance every 4 weeks.
This original strategy for top 1 returned -5.5% for 2015 and 20.5% for 2016. There is an error in my 2015 quant performance report. The list of bonds ETFs for that report was different. I’ll fix the old post. The top 3 version returned -2.3 for 2015 and 7.6% for 2016.
I have since added international (BNDX) and emerging market bonds (EMB) to the original ETF list and removed municipal bonds (MUB). All future bond quant results will include these two ETFs. The top 3 version with these ETFs returned -3.3% for 2015 and 9.5% for 2016. The top 1 version returned -5.9% for 2015 and 20.5% for 2016.
Hope that clears things up.
Comments are closed.