I find the whole problem kind of too complex. You have to somehow factor in forward rate tables and do a lot of complex math. You also have to specify a date we are projecting to or the problem is intractable. This is my rough ballpark guess, using a 2 year projection, from looking at forward rate tables 2 years from now. The return of all treasuries is reduced, especially LTT, because forward rates indicate the market expects all interest rates to rise. STT and ITT do not decrease very much compared to the original values because the original values did not include return from roll. So I adjusted up from 0.8% to 1.3% first, and then down to 0.5% based on market forward rates. LTT see the biggest adjustment down because they have the most duration exposure to the market expectation of rate increases, and the smallest benefit from roll. The original value was 1.87% which I adjust down to 0.7%. Also curious what EfficientInvestor thinksLTCM wrote: ↑Wed Sep 08, 2021 5:38 pmThat would be very useful!EfficientInvestor wrote: ↑Wed Sep 08, 2021 4:52 pm What I would really like to do at some point is to gather all historical yield data and develop an algorithm that determines the current optimal point on the curve. So this algorithm would be a function of roll and carry I suppose.
I liked your efficient portfolio approach listed above. What would you amend the rates to based on this discussion (if anything)?
We could specify a longer projection horizon than 2 years, like 30 years. The return of the 30 year would basically be the current YTM, plus roll since we would be rolling 30s, not holding to maturity. But the return of the ITT would go up dramatically based on market forward rates > 2%. Of course having a 30 year horizon kind of defeats the purpose of a dynamic strategy that changes its allocation based on current market rates. If we're deciding what to invest in for the next 3 months or a year, we should make a 1 year projection of returns, using forward rate tables for a year from now.
https://www.portfoliovisualizer.com/eff ... ints=false