Several of us have frequently discussed what impact TLT might have when we look at various back-testing results. The “Rutherford 10” tends to be a difficult portfolio to outperform. Could this be due to the strong performance from TLT over the years tested? To see what impact TLT is having on performance, I asked Ernie Stokely to run tests on the “Rutherford 10” with and without TLT. Ernie has been most generous with his time in running many back-tests I’ve requested.
Below are the results. I hope we will have more to report later, but for now, readers can see that TLT is definitely an “enhancer.” One test is the standard “Rutherford 10” where TLT is part of the portfolio. The second run substitutes BIV for TLT.
That fourth column (first BIV column) should read BIV 2P instead of BIV 3P.
If you have questions, please post them in the Comments section.
While I understand the arguments and concerns regarding the inclusion of TLT in our portfolio lists (and I too have warned about the dangers of overemphasizing it) – e.g. using it as a proxy for cash) I’m not sure that I could argue a good case for excluding it – it is simply a (hopefully) low correlated asset to others in the asset lists. Could we use an alternative Bond fund? yes, of course we could, and I could make a case that BIV might be a better choice since the return/risk is better!! But then, do I have to question whether BIV should be excluded for bias? – or even an equity ETF such as VTI (that has performed exceptionally well over the test period).
In the end we want a list of assets with low correlations and representing different classes and we then let momentum sort them out – if TLT comes to the top and we are “lucky” – so be it and let’s go with it – we may not see it again in the next 10 years! Looking at it from a different angle I can’t remember seeing TIP in the top rankings since I’ve been tracking the portfolio – but this doesn’t mean it shouldn’t be there (and maybe replaced by a “high flyer”) but, if/when inflation comes back maybe we’ll be glad that it’s there.
David
David,
I totally agree with your above comment. By substituting BIV for TLT in no way should send a signal we eliminate TLT from the portfolio quiver. I asked Ernie to run the comparison as a way to see what impact TLT might be having on the performance of the Rutherford 10, as it continues to set a high performance standard.
One of the reasons this came up is that I culled through over 125 ETFs to find a group of 10 to 12 low correlated ETFs. When Ernie ran the tests, the low correlated group did not perform better than the Rutherford 10. The question arose once more – what is going on within the Rutherford 10 that makes it so formidable? If you will recall, the “Rutherford 10” is the outcome of merging the “Swensen Six” and the “Ivy 10” some years ago. This was even before I ran any correlation numbers. The “Rutherford 10” sort of came together by chance, and has worked quite well when compared to the VTTVX index fund. This sums up my reasoning for requesting another look as to what impact TLT has on performance.
What I find interesting is that adding more ETFs to the “Rutherford 10” does not enhance performance, but frequently diminishes the returns. Intuitively, I would not think this would be the case, but the interaction between the ETFs runs counter to my intuition.
Lowell
David, Lowell,
I’m in agreement that we shouldn’t exclude either BIV or TLT a priori. TLT is often a good candidate in a falling rate environment. That environment will likely end “any day now”, and at that point our selection algorithm should exclude TLT on its own. I’ve also seen that adding more funds to the universe for a selection algorithm can degrade performance. No selection algorithm can see the future, and will thus always make mistakes. The more potential candidates we present to the algorithm, the more likely the algorithm is to make a wrong decision.
Pete Arnold
Pete,
Your last sentence reminds me of an Einstein quote.
“Everything should be made as simple as possible, but not simpler.”
Lowell
Lowell,
Absolutely! No epicycles!
Pete