New Source For Older Stock Data

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
Tyler9000
Posts: 577
Joined: Fri Aug 21, 2015 11:57 am

New Source For Older Stock Data

Post by Tyler9000 »

As part of my ongoing personal research into historical data sources for backtesting purposes, I stumbled across a data source I've never seen before that I thought I'd share. It was compiled by Independence International Associates and includes monthly country and market segment indices from 1975 to mid-1997 for a huge number of things you don't normally find.

https://web.stanford.edu/~wfsharpe/rets/iia.htm

If you're concerned about credibility then there are a few nice things to note. First, the page is hosted on William Sharpe's Stanford website -- someone I respect. Second, look up that company name in the index of "Triumph of the Optimists", and you'll find DMS referenced the exact same data for international small caps. And finally, I was discussing this with Siamond and he found a paper (read the endnotes) that details how IIA actually had a hand in designing MSCI's value and growth indices. So clearly they're legit. In fact, after digging through the data I've personally concluded that you can pretty much think of it as similar to Fama French data but with far more realistic size and value filters.

The most important thing to note is that while their definition of "large cap" (top 70% of market capitalization) is consistent with most indices these days, they define the small cap breakpoints as 70%-100%. If you know anything about how indices work, that means that it's really more like Mid+Small by modern standards. And while there's not enough overlap with MSCI to draw direct conclusions, I have my own US stock data calibrated pretty well to MSCI that I can use to compare against the IIA methodology. As you can clearly see, IIA "small cap" is really much more similar to modern "mid caps". But at some point that's also being picky, as small and mid caps have reasonably low tracking error in normal portfolio percentages.

Image

As I mentioned earlier, the data hasn't been updated since June of 1997 and I'm not sure IIA even exists as an independent entity today. That may not make it particularly useful on its own, but once you realize that similar Fama French data picks up the slack since 1991 and other more recent data is available from index fund providers it opens up all kinds of opportunities for personal study. I'm currently playing with a few ideas, and I imagine the Simba crowd may also appreciate the potential for expanding and improving some of our older data series.

Enjoy.
:beer
User avatar
siamond
Posts: 5738
Joined: Mon May 28, 2012 5:50 am

Re: New Source For Older Stock Data

Post by siamond »

Thanks, Tyler. I'd like to emphasize that this is really a treasure trove of historical data, providing full total-return history for 22 countries, with breakdowns using the usual size/value factors, and groupings per geographical theater. Those series can be easily combined with either French-Fama or MSCI corresponding data series to provide a full history from 1975 until today. And there is no copyright of any sort on the data, from what we can tell.

For Simba, this should finally provide an answer to improve the International Small data series, potentially eliminating the unfortunate dependency on DFA (see detailed discussion in this thread), and going further back in time. We might also consider adding brand new data series, to be discussed. Very cool.
AlohaJoe
Posts: 5678
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: New Source For Older Stock Data

Post by AlohaJoe »

Nice find. I'm always a bit amazed that these treasure troves of historical data are just laying around and no one seems to know about them. It makes it less and less excusable for books, blogs, etc to only ever refer to "the S&P 500 since 1871" (or whatever) in their charts and backtests.
User avatar
siamond
Posts: 5738
Joined: Mon May 28, 2012 5:50 am

Re: New Source For Older Stock Data

Post by siamond »

I've been mulling over this IIA data set, pondering what would be desirable to extend Simba (besides the obvious move of extending the International Small data series). And... I am not sure this is desirable, actually.
1. International Small-Cap Value? Complementing MSCI World (or ACWI) Small Value? That one seems very tempting, but then there is no corresponding Vanguard nor iShares real-life fund to splice to...
2. adding regional (e.g. Europe, APAC) value and small data series? Or just small and small value? Maybe this is the right balance to give more of a flavor about factors' applicability besides the US? Same issue though, no Vanguard nor iShares funds to splice to.
3. adding a few per-country data series? And splice to corresponding iShares funds? I pondered about a sample like Canada, UK, Japan, but then the MSCI 1970+ data series are actually better, at least for total-market. And then it's a slippery slope, add one more country, one more country... Plus country-specific returns expressed in USD currency combined with US inflation is kind of weird.
4. adding a few per-country + factor (e.g. size, value) data series? Similar objections, plus it gets overly specialized.

About #2, another issue is that the Pacific data series are listed, but lead to broken links on the W. Sharpe Web page, aaarg. While Europe works well.

Overall, it seems to me that this IIA data set is well suited for various point studies, but its applicability for something more general-purpose and somewhat US-centric as Simba might be more limited. Thoughts?
User avatar
Topic Author
Tyler9000
Posts: 577
Joined: Fri Aug 21, 2015 11:57 am

Re: New Source For Older Stock Data

Post by Tyler9000 »

The part I personally find most interesting is the level of size and value factor data for World and Europe, but I admit that's more from a non-US perspective. European investors have different needs and index offerings from their US counterparts, and good historical data is even more difficult to find for those types of assets. I like being able to bring something new to the table and plan on using this info myself, but you're also absolutely right that unless you're also accounting for exchange rates and local inflation it is a lot less useful. At some point one could argue that it becomes a different type of tool than Simba is designed for, so I can understand the hesitancy to get too far away from the original charter.

That said, if I were to add one new asset to Simba from the IIA database it would be International SCV. While I agree that appropriate index fund availability is a problem and one would need to be smart about how to apply the information to their own portfolio expectations, I can also see how it would be useful for more accurately modeling the Swedroe MFT and Merriman Ultimate portfolios already in the spreadsheet. Both authors specifically call for International SCV, and Simba simply lumps that in with similar assets.
Last edited by Tyler9000 on Sun Apr 29, 2018 12:17 pm, edited 1 time in total.
User avatar
SimpleGift
Posts: 3869
Joined: Tue Feb 08, 2011 3:45 pm
Location: Central Oregon

Re: New Source For Older Stock Data

Post by SimpleGift »

siamond wrote: Sun Apr 29, 2018 9:29 amOverall, it seems to me that this IIA data set is well suited for various point studies, but its applicability for something more general-purpose and somewhat US-centric as Simba might be more limited. Thoughts?
The beauty of the Simba spreadsheet, in my view, is that it provides return series for asset classes that do-it-yourself investors can actually invest in and construct portfolios from today. Adding regional and individual country data series to the spreadsheet — though these may be of historical interest — likely won't add much practical value for today's U.S.-based investors, where broader market international index data is much more useful.

Thus the International Small-cap series is definitely of value to modern portfolio construction — and perhaps International Small-cap Value would be as well — but I doubt that adding Europe or Pacific regional small and value data series, much less individual country series, would offer much additional utility. Just my (U.S.-based!) perspective.
User avatar
siamond
Posts: 5738
Joined: Mon May 28, 2012 5:50 am

Re: New Source For Older Stock Data

Post by siamond »

Tyler9000 wrote: Sun Apr 29, 2018 11:28 amThat said, if I were to add one new asset to Simba from the IIA database it would be International SCV. While I agree that appropriate index fund availability is a problem and one would need to be smart about how to apply the information to their own portfolio expectations, I can also see how it would be useful for more accurately modeling the Swedroe MFT and Merriman Ultimate portfolios already in the spreadsheet. Both authors specifically call for International SCV, and Simba simply lumps that in with similar assets.
SimpleGift wrote: Sun Apr 29, 2018 12:13 pmThe beauty of the Simba spreadsheet, in my view, is that it provides return series for asset classes that do-it-yourself investors can actually invest in and construct portfolios from today. [...] Thus the International Small-cap series is definitely of value to modern portfolio construction — and perhaps International Small-cap Value would be as well
I agree that Int'l SCV is the most desirable option. At some point, Vanguard and/or iShares should fill this gap. In the meantime, checking threads like this one or that one, there is really no proper option where a real-life fund has a clear charter for Int'l SCV, besides DFA's DISVX. All the discussions are centered on Int'l Small-Cap funds which happen to have a bit of a value tilt, but not by their charter, nor by the index they follow. As to DISVX, it has a long track record (1995+), which would make for an easy splice, but then... it's DFA (and institutional)... Ok, let me mull this over a bit more, and maybe open a separate thread.
lazyday
Posts: 3798
Joined: Wed Mar 14, 2007 10:27 pm

Re: New Source For Older Stock Data

Post by lazyday »

Could individual country series better help backtesters to do out of sample tests?
User avatar
siamond
Posts: 5738
Joined: Mon May 28, 2012 5:50 am

Re: New Source For Older Stock Data

Post by siamond »

lazyday wrote: Sun Apr 29, 2018 1:46 pm Could individual country series better help backtesters to do out of sample tests?
This is exactly what I had in mind while listing option #3 above, but Simplegift is right that this would depart from Simba's general aim. Plus it would then make more sense to do it using local currency and local inflation, as I did in this point study (for which I used a heavily customized derivative version of Simba). Note that PortfolioCharts already allows to do something like that for a few countries, and I'm sure Tyler has plans for this IIA data...
User avatar
Topic Author
Tyler9000
Posts: 577
Joined: Fri Aug 21, 2015 11:57 am

Re: New Source For Older Stock Data

Post by Tyler9000 »

siamond wrote: Sun Apr 29, 2018 2:05 pm Note that PortfolioCharts already allows to do something like that for a few countries, and I'm sure Tyler has plans for this IIA data...
Yep! :D

viewtopic.php?f=10&t=171019&p=3921803#p3921803
Post Reply