Dear socially responsible Fools,I wrote some software to do historical analysis of the Domini Four strategy (like the Foolish Four, but applied to the Domini 30.I wrote it up in a little essay, whose URL I include below. (This is my first post with a URL, si I'm curious to see if the URL is actually click-able.)The essay is not complete, but it has enough interesting information in it that I thought you all might enjoy it. If you see anything incorrect, please let me know! The software is also available, as mentioned in one of the appendices.http://www.cybermesa.com/~rosalia/sr/dfour-essay/dfour-essay.html
Very nice work!Hope you don't mind if I include a link to your message in the FAQ for the Dow Investing/Foolish Four board.Dave GoldmanPortland, OR
Thanks for your interest and effort in researching the Domini Four!A couple of things I noticed on first reading:1)Domini has recently revised how they divide their sectors. Past performance would have to be based on their old system, which would create a tremendous difference in each year's 30.2)The BSP strategy on which our Domini Four was based excludes non-dividend yielding stocks such as MSFT and CSCO.3)The DSI has been in place since '90; the DSEF since '91, so there is more room for backtesting.Where are you hoping to take this research, and why did you choose this particular strategy?Kelly
Very nice work!Thanks; it was fun!Hope you don't mind if I include a link to your message in the FAQ for the Dow Investing/Foolish Four board.No problem: I was going to post something to that board myself, but it's volume is higher, and I'm not sure if I have time to follow too much discussion :-)You're from Portland? I spent four happy years going to college there.
Thanks for your interest and effort in researching the Domini Four!You're welcome. I had a lot of fun, and I was really worried about owning Phillip Morris or Union Carbide.A couple of things I noticed on first reading:Only a couple? Phew.1)Domini has recently revised how they divide their sectors. Past performance would have to be based on their old system, which would create a tremendous difference in each year's 30.I partially disagree, mostly because I think it's not important. The "list of 30" is rather stable, and its component companies are big and have a lot of inertia, so I don't think it makes much difference to choose historically accurate "lists of 30". I used the 1999 30 companies as far back as June 1996.For example, I think that if the Dow index were frozen to how it was in 1998, little would change in all the Foolish Four simulations.Another thing to think about (didn't you do this one?) would be to get the list of 400 companies, sort them by descending capitalization, but select them on the basis of the Yahoo! categories instead of the Domini categories.2)The BSP strategy on which our Domini Four was based excludes non-dividend yielding stocks such as MSFT and CSCO.That's trivial: in the RP sorting they will be at the bottom of the list, so they will never come up. Notice that MSFT is in the Dow, so having it in the "list of 30" does not hurt too much. I guess the real effect is that you end up with a "list of 28".With the Domini list, since it is missing a lot of big companies, allowing a few more companies in by excluding the no-dividend ones (there are about 4 or 5 of them -- not just Cisco and Microsoft) would scare me a bit because we'd end up with several $5 billion companies which might not have the same resilience as the Dow Jones 30.Still, I want to generalize the software a bit so that people could put in a "strategy profile" for the various selections, and then we could do historical runs on that.3)The DSI has been in place since '90; the DSEF since '91, so there is more room for backtesting.I mention that it would be a simple change to go further, but my thinking is a bit different:(a) are you sure DESFX has been around that long? Yahoo only has it until about 1986. (b) How can I get historical DSI data? (c) As I mentioned above, I don't mind using the current list of 400 companies even for the past.Where are you hoping to take this research, and why did you choose this particular strategy?The "particular" strategy was chosen because I like the Foolish Four and wanted to avoid PM and UK. My interest is in seeing if I want to try investing in the Domini Four, and I did not want to do it blindly.Where do I want to take it? I'm almost done, but I'd like to:(a) Go further back in history (partly by tweaking my programs, but you can help me by helping me find DSI and DSEFX data going further into the past).(b) Compare to the Dow-based Foolish Four (should not be hard; it's just code, right? ha!)(c) Generalize and abstract these programs a bit so that they can be used as general evaluators for complex strategies. If there are any hackers here who want to work with me, let me know.
Vis-a-vis the analysis I wrote up, I should point out that:(a) currently the software does not do anything with the dividends. I should change it to take all the dividends and invest them with the next year's batch.(b) I might have mixed up the original Foolish Four with the recent RP variation. I used a 2-2-3-4-5 scheme, but I think that the RP variation is supposed to just 2-3-4-5. I'll look it up and fix things, or maybe do comparative runs.
Thanks so much for working on this! I'm just in the process of moving an IRAnnuity and previous employer 401(k) to an online brokerage account. I've decided to follow the Foolish Four and was glad to see PM was #1 - I was investing in them anyway. :) I know you still, as yet, have to add the Fool4 comparisons. I was wondering, though, the difference between picking the Domini 4 and choosing the 2nd 4 responsible companies from the Fool4? I was (so far) planning on investing in CAT, EK, SBC and GM. I haven't done any research yet to see if any of those 4 are currently socially irresponsible. I guess visiting http://www.domini.com/ will help me when I finish the IRA transfers.Thanks again,Pete
Reed, by any chance?
If you took your price and dividend data from Yahoo, it's possible that you've been led into a crystal-ball error. Search the Fool Four board for posts by TimberFool, MontanaFool or Dave Goldman (to name three who immediately come to mind) dealing with this issue. There may also be a mention of it (and references to explanations) in the FAQ for that board.Kind regards,JakeLife was formerly so simple
I see now that Dave has already posted to this thread, and not mentioned the issue, so I suspect I'm in error. I wrote after an extremely and unjustly quick scan of your article.(The issue, for what it's worth, and if I remember it properly, is that Yahoo adjusts its historical pricing for splits. Since in a split market capitalization stays constant while the number of shares changes, price changes to offset the change in number of shares. Since price is an input to the RP ratio, doing an RP sort using Yahoo's split-adjusted historical prices will select stocks which would not have been selected at the time, but which later split, bringing down their split-adjusted pricing retrospectively.)(Gummy, what a paragraph.)My apologies for mucking it up, if I did...Kind regards,Jakethere came down to the beach from the Altogether Uninhabited Interior one Rhinoceros with a horn on his nose, two piggy eyes, and few manners
MakeItJake wrote:I see now that Dave has already posted to this thread, and not mentioned the issue, so I suspect I'm in error.I haven't actually done any critical reading or thinking here, so my not saying something carries no meaning.Thanks, Jake, for pointing out this key issue!Dave GoldmanPortland, OR
my not saying something carries no meaningUmm ...Then I provisionally withdraw my earlier demurrer ... <g>Dave, what are you doing up at this hour?Kind regards,Jakeunder the blear eyed mooni am pelted with cast off shoonbut wotthehell wotthehell
rosalia3Thanks very very much for your excellent work!!I am just about to invest in the foolish four strategy (I know I'm a bit late). Now that I have read your article, I will definitely buy the responsible four.Thanks so much for this excellent work!I did a rough comparison to the historic RP4 results obtained from the book, "The Foolish Four" and it appears that the RF (Responsible Four) out performed the RP on average by about 10 percentage points for the years 1992 through 1996. These are the years the purchase was made. The book reports pruchases made 12/15/92 as results for 93. I used your Dec. 1 purchases and interpolated from the graph. I look forward to your comparison to the RP. Those results should definitly be posted on the Foolish Four Board!I have a couple of minor suggestions or questions.I would like to see a table providing the annual returns for each year. The graph is good, but I think it should be acompanied with the data so one doesn't have to guess if they want the actual numbers. You stated that starting in late December provided the best results but I didn't see the data to back that up. Looking at the graph and the data in the tables it looks like Sept. had the best return of 31.73%. I couldn't find a specific return for the late Dec start date. I presume it was higher because of your statement but the graph doesn't seem to bear that out.I found the graph comparing various start dates a bit busy. I couldn't see the blue line and I finally saw that the yellow line, which barely shows up, connected the dots at about the Dec. 20 date. Perhaps this one should be split up into two graphs.I didn't download the program and not being really technically capable (I used to be 5 yrs. ago) was a little afraid to do so. I'll wait for the refinements.You have indicated you will check this one out but I was suprised at the 2 2 3 4 5 stock selection. My understanding is that the RP4 no longer uses this pattern and just buys equal amounts of the stocks that rank 2 through 5. However, there is much work to be done and perhaps some of us could help. All the variations tried in the FF aproach ought to be tried here. Perhaps the 2 2 3 4 5 aproach is better for the responsible four. These would be refinements to the aproach you have taken here. Who knows, perhaps some creative person will come up with a new version.I like the name Responsible Four better than the Dominini Four. I know you are still in a draft mode but I suggest you eventually be consistent with the name you use. Like I said these are very minor suggestions and none of them really address the main substance of what you have done. You have done excellent work and I really apreciate it!Thanks, Chuck
Hmm. I just took another (quick) peek at your study, and I think that Jake is probably right -- if you've gathered your data from Yahoo! or some other readily-available source (I can't tell -- do you mention your source for prices and dividends somewhere?), then you are guilty of Backtesting Fundamental Error #1: using split-adjusted prices. Probably also Fundamental Error #2: trusting your sources.When Timberfool gathered his historical data for backtesting the Foolish Four, he had to spend a lot of time in the library (yes, the big, offline one) to determine all the actual prices and dividends. When you calculate, say, the FF picks for 1989, you need to know the actual 1989 prices, not the 1999 reported split-adjusted prices for 1989. And backtracking all those splits can apparently not be done accurately online.Not only splits are at issue, but also spin-offs, acquisitions and mergers. In particular, the break-up of the phone company apparently presents a significant challenge to the dutiful backtester.Finally, some reported data is simply wrong. I believe that Timberfool, in some cases anyhow, had to check two or more sources to determine accurate prices and dividends.In any case, if your starting data set consists of 1999 reports of historical split-adjusted prices, and also ignores spin-offs and acquisitions, then I'm afraid that all of your subsequent analyses partake of a major "crystal ball effect" -- your programs will routinely pick stocks based on their future performance rather than their actual historical price and dividend.Dave GoldmanPortland, OR
Yikes!!! Even worse! Tell me if I've misunderstood something here...It looks to me like your "Domini 30" companies were picked on 23 December 1999. And then you simply pick from these specific 30 companies for many dates in the past.As opposed to calculating the Domini 30 for each date in the past, and then picking 4 from those 30 for that start date.If the above understanding is correct, then you need to rethink your entire approach, and retract all of your current analyses and results. You are using future (Dec 99) data in picking your past stocks.Dave GoldmanPortland, OR
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |