I have asked one of the programmers. But if I remember correctly on the asset classes that have long histories we use that entire history. Those with short histories (e.g., Pacific rim or perhaps REITS) there are shorter histories. I believe the minimum is ten years.

You can see this exactly on the Monte Carlo, Build Portfolios tab. Click on an asset, then use your right mouse button and click "view". That will show the series for each asset. For example, Large Cap Stocks goes back to 1926.

There may be a couple of DFA assets with only 9 years, but 10 years is the minimum for everything else.

Best,
Brian

Some of the DFA assets are very short. E.g., International Vector has only 6 years of data. Most are longer, though.

Many of the general-sounding funds like large stocks are many decades old and are based on reliable public sources, e.g., Ibbotson. The Large Stocks asset itself dates back to 1926.

I would have thought all data sets should be of the same length, the longer the better. I'm guessing ESPlanner does a regression of each asset onto available long term data sets and estimates or imputes missing data.

There is a nice tidy data set of annual returns for most asset classes-of-interest from 1972-2014 at https://www.portfoliovisualizer.com/historical-asset-class-returns. It looks like the data set that ought to be included with ESPlanner. If nothing else, you could run a side-by-side comparison between these data and ESPlanner's historical annual returns data. CPI-U is included so you could adjust nominal returns to real returns and visa versa.

Portfolio visualizer is a very cool tool.

I used Portfolio Visualizer data to produce some custom assets. With the new functionality in ESPlanner Plus, it's easy. I named them all with a prefix of "PV" to indicate the source. I also added the calendar year as a suffix as a reminder to update it when the year rolls around. I can edit the name accordingly, when I do.