A.I.M. Users Bulletin Board

[SFSecurity]

Hi Clive, Have a sunny weekend yourself. Rain is predicted here in San Francisco, but who knows. Weather forecasts are a lot like prognostications by stock analysts - take them with a very large grain of salt as they tend to be more inaccurate than not.

Thank you for the excellent discussion of the issues of accounting for actual gains involved with backtesting. I intend to read and reread it a bunch of times to extract the most out of your contribution to this forum.

BTW, I should have posted the actual link I use to get prices from Yahoo. https://finance.yahoo.com/q/hp?s=VFISX&a=09&b=28&c=1991&d=11&e=13&f=2014&g=d Change the symbol (VFISX here) to what you wish to see. Then change the last character to what you want:

 d = daily 
 w = weekly 
 m = monthly 
 v = dividends only

Now to adjusted versus daily close. I've looked at a number of different stocks and noticed an interesting aspect to adjusted prices. For the most part the further into the past you go the wider the divergence between the adjusted and the closing price. Yes, it seems to change based on dividends by lowering the adjusted price prior to the date of the dividend so that makes the historical price lower, and increasing the AIM trading signals. This seems to make the gains higher in backtesting a stock. Is this realistic? For VFISX, I'm not sure but it might be, for others I think it actually seems to ignore inflation's impact as we will see when we look at ADS.

First of all inflation between 7/14/2000 and now is approximately 38.5% but the change in adjusted price for VFISX is about 65.6%, or about 19.6% more than inflation. The dividends for that period total $4.693/share with the bulk of it in early years as the current dividend is $0.005 and the early dividend was $0.051. I'm not sure the proper way to account for inflation considering the ten to one drop but let's assume that the actually payout was equal for every period, this would mean that the real income would be somewhere around $3.50, I would guess. Possibly a bit less given the early years weighted dividends. Mashing it all together it would seem that the adjusted closing price for VFISX is very close to correct adjusted for inflation and dividends received.

Now take ADS, (it only goes back to June of 2001) as far as I can tell it has never paid a dividend and never had a stock split. The adjusted price is exactly the same for every monthly period. So adjusted price would not account for inflation so this would not reflect at least inflation. This would mean that the current price of $274.32, 18.82 times the initial price of $15.00, is not realistic. Not sure of the best way to state the current actual value of an investment made back then in current purchasing power, but I think that it would be about $197.87, based on an inverse calculation at http://www.in2013dollars.com/.

So I ran ADS data through your spreadsheet and B/H does better than AIM, and this is with a beta of 1.62 according to Yahoo.

  4.69 Gain Factor	        18.29 Gain Factor 
 13.67	years			13.67	years 
 11.98	%CAGR			23.70	%CAGR 
 AIM				B/H	 
 					 
 		Avg Cash			 
 		66.98%			 
 

So, I'm not sure what is the best way to backtest given this. Yet another puzzle to solve to develop a good methodology that is closer to reality. What I have done is limit my backtesting to about 2 years to reduce the effect of inflation and dividends.

On another note, is there an easy way to upload a spreadsheet here or do I need to use something like dropbox and a link?

Best,

Allen