When reading a post with a keyboard, you can type the keys , and . to move backwards and forward between posts! You can also press 'return' to read through posts one at a time. There's freedom in Shrewd'm!
- Manlobbi
Halls of Shrewd'm / US Policy❤
No. of Recommendations: 11
I thought that if this was a matter of interest it would be better to start a new thread.
After FlyingCircus mentioned the use of Dual Momentum for Fixed Income I spent some more time looking at it. My goal was decent, steady returns without too much trading.
I settled on just three asset classes (High Yield, Mortgage-backed, and Intermediate treasuries), with ultra-short bonds for the out-of-market asset. There is no absolute momentum asset. Look-back period is 4 months, traded monthly on the next close.
I tried this with ETFs (JNK,MBB,IEI,BIL), and with mutual funds in the same categories:
JNK = BUFHX
MBB = PTRIX
IEI = FSTGX
BIL = PTSHX
Using mutual funds allows the method to be tested back to January, 1998. From Jan 1998 until June 2023 it produced a CAGR of 6.87%, Sharpe 1.16 Sortino 2.48 MDD -5.12. 80 trades in 25 years. ETFs can be tested back to 2009, and produce similar result, as does weekly trading, which adds more churn for pretty much the same payoff, but offers evidence that there's nothing special about month-end trades.
Baltassar
No. of Recommendations: 2
I've glanced at DMFI a few times over the years. Never did it though.
PORTFOLIO VISUALIZER link (mutual funds):
https://www.portfoliovisualizer.com/test-market-ti...Dec'97-Jun'23 CAGR 6.90%
Took off from 2009 to now. Dec'08-Jun'23: CAGR 7.74%
But Dec'97 - Dec'08, these were essentially identical, DMFI 5.92%, Vanguard Total Bond 5.71%
Eleven (11) years of useless trading. Doesn't help, doesn't hurt.
PORTFOLIO VISUALIZER link (ETFs):
https://www.portfoliovisualizer.com/test-market-ti...Dec'08 - Jun'23 CAGR 7.36%
So, yeah, looks like the ETFs and the funds are interchangeable.
There is probably some over-tuning here. 4 month look back gives 7.2% whereas 3 month & 5 month give 5.6%. Still better than BND, though.
No. of Recommendations: 1
Yes, I noticed that 4 months was a sweet spot, but it is close to what I use for equities, so I figured I should just go with it. Almost any reasonable look-back seems to do OK (not carefully investigated).
Completely agree about useless trading for the first ten years. Which is a tell, really. The graphic representation that Portfolio Visualizer produces shows bog standard aggregate bond returns until 2008, and then the line really moves. If you look at the "timing periods" tab you see that in 2009 the thing spent 14 months in high yield bonds, which returned 41%. Something similar happens in 2012. I suspect that's the entire difference.
Personally, I've always thought of high yield bonds as an equity substitute, and that's certainly what they look like here. I think it would be hard to put together an out-performing DMFI model that doesn't include junk bonds for extra juice.
Baltassar
No. of Recommendations: 2
OK, I couldn't not look again.
Here's a stripped-down two-asset version that swaps between aggregate bonds and high-yield bonds, with cash as the out-of-market asset. The benchmark is high-yield bonds. Backtest goes to 1995. As before, 2008 is a visible break point, in terms of out-performance.
https://www.portfoliovisualizer.com/test-market-ti...If I already owned an aggregate bond fund like BND, I'd at least think about this.
Baltassar
No. of Recommendations: 0
I like this better, it's more in line with Global Equities Momentum (GEM). Just 2 assets.
The thing that keeps getting me about any bonds is Jim's comments about "return-free risk". And the fear that 4 month lookback is just a happy accident.
No. of Recommendations: 2
"Return-free risk" is the problem, for sure.
I don't think this kind of approach is a reason to hold bonds in lieu of equities. I use an ultra-short bond fund as a "cash" reserve. Strictly speaking it's not really an "investment" at all.
But if I held bonds as a means of moderating the volatility of a diversified portfolio, as is common, I'd think carefully about something like this. I agree that the granularity of the backtest could be better (really miss GTR1); and the trades will introduce some friction, and throw off taxable gains. But even so, if you can beat an aggregate bond fund by 100 basis points, with no increase in risk, and it's 30 or 40 percent of your portfolio, that's not nothing.
So it seems to me, anyway.
Baltassar
No. of Recommendations: 3
Just for shiggles I ran a portviz on what I had been doing the last few years - for the fixed income portion of the port, the best of JNK, preferreds (PFF, there are others), 7-10YRS (IEF) and BND. Same special 4 month lookback. Results
https://www.portfoliovisualizer.com/test-market-ti...Similar to the original DMFI constituents results, much better than straight BND - although helped immensely by being in Cash for most of last year.
No. of Recommendations: 0
Here's a stripped-down two-asset version that swaps between aggregate bonds and high-yield bonds, with cash as the out-of-market asset.
Baltassar, just to make sure I'm reading the absolute filter properly, is it correct that if neither of the 4-month bond fund returns beat the 4-month cash return, the next month would be held in cash?
Tails
No. of Recommendations: 1
Tails wrote:
is it correct that if neither of the 4-month bond fund returns beat the 4-month cash return, the next month would be held in cash?
Yes. That is my understanding, anyway.
I tried both short- and intermediate-term bonds as "absolute return" filters, but they did not improve either the results or the risk metrics. To my way of thinking the idea makes more sense for equities, which trend more strongly than bonds.
I also think an "absolute return" filter may limit the benefit of having High-Yield bonds as one of the picks. They can respond strongly and quickly to general market recoveries (e.g. 2009), when more vanilla bonds may take their time about it.
Baltassar
