I'm bit stumped on this one. I setup 7 asset classes:
US Large Cap
US Small Cap
Ex-US Developed
Ex-US Emerging
US Bonds
Ex-US Bonds
TIPS
Any suggestions as where to find historical data for historic classes that would approximate the above?
@humsby One place to start might be some of the sources on the Portfolio Charts site: https://portfoliocharts.com/user-guide/data-sources. There's a ton of entry points that might lead you in the desired direction.
@chrisb I will check it out. Thanks.
I think one of my issues is that some of those assets of mine haven't been identifiable since 1928 so the historical simulation won't work. I will just have to use the S&P and treasuries as a rough approximation.
Yes, we have some tilts to small and small value and the tracking error vs the S&P has been quite large and painful as we watch the S&P rocket to the moon while we have worse volatility in down periods and less upside in strong market conditions. But we've stuck it out long enough that giving up and switching would cost unpleasant amounts of capital gains.
Our experience is not unique. I understand that several of the Fama-French factors lost their predictive power after they were published. Hard to say whether that was because they were coincidental results, so were never real, or maybe the world changed and the old rules don't apply or maybe things will change again tomorrow and they'll come back in style.
Predicting anything about the stock market is impossible, slicing and dicing a portfolio and hoping for more accuracy/better results seems like a double impossibility. To me, the main goal of historical results is to illustrate the huge variability that is possible, so you can plan for what you would do if long term poor market conditions return.
@ricke Yeah, at one point I had a small cap tilt with two Russell 2000 index funds. When I wished to simplify my portfolio, I had a lot more gains in those funds than I wanted to pay tax on. So those will get whittled down a bit between now and when I start SS and RMDs. However, those are easily just tossed into the S&P performance since there isn't much of a tilt there now and it will decrease over time.
The asset classes that I really would like to be able to find a good historical database for would be the international sectors. Currently, about 30% of my equity positions are in Ex-US funds. 20% of those equity positions are in emerging markets. So over all, 17% of my portfolio is international and it would be nice to model that with historical data.
Then fixed income has me confused. I have about 5% of my portfolio in TIPS index funds, 3% in non-US bond funds, and 10% in an intermediate Bond fund. In Pralana I just used the 10 year Treasury to model it.
@ricke Yeah, at one point I had a small cap tilt with two Russell 2000 index funds. When I wished to simplify my portfolio, I had a lot more gains in those funds than I wanted to pay tax on. So those will get whittled down a bit between now and when I start SS and RMDs. However, those are easily just tossed into the S&P performance since there isn't much of a tilt there now and it will decrease over time.
The asset classes that I really would like to be able to find a good historical database for would be the international sectors. Currently, about 30% of my equity positions are in Ex-US funds. 20% of those equity positions are in emerging markets. So over all, 17% of my portfolio is international and it would be nice to model that with historical data.
Then fixed income has me confused. I have about 5% of my portfolio in TIPS index funds, 3% in non-US bond funds, and 10% in an intermediate Bond fund. In Pralana I just used the 10 year Treasury to model it.
I was trying to figure out what data source Pralana was using for stocks, bonds and cash. Bonds appear to be Damodoran's data (NYU) for 10 year Treasuries, cash matches Damodoran's 3 month T-bill, but stocks don't quite match Damodoran's data.
You should realize when you start looking for datasets that even the 10 year and 30 year Treasuries are somewhat synthetic data as you go back in time as the Treasury didn't regularly sell them. Back in the day, the government used to run surpluses at times, so there wasn't always a need and until fairly recently (say the 1950's), they didn't settle on 10 and 30 year as sort of standard instruments.
International not only suffers from data problems, it suffers from WWII problems. Europe had about the only reasonably well-developed market economies outside the US prior to WWII and most of the industrial base in Europe was bombed to rubble, so I would guess stocks behaved accordingly.
TIPS were introduced in 1997, if you run across data from before then, it's some kind of synthetic construction.
Professor Edward McQuarrie wrote a paper on long data sets with references to various other books and papers:
https://www.tandfonline.com/doi/full/10.1080/0015198X.2023.2268556?scroll=top&needAccess=true
