I will keep a log of a new model I have made to systematically invest in US equities. The model is a long-only investment method which uses fundamental as well as technical factor in order to generate a list of candidates. The model combines momentum investing (using simply technical factors), institutional holding, and fundamental factors (change in earnings, ROE, ROA). The model uses a ~2000 stock universe and picks, typically, are medium-large capitalization companies. The list will generate 35 candidates, ranked in terms of attractiveness, every month. Based upon testing the top 1-5 stocks will outperform the 30-35 ranked stocks. Thus, the picks can be used by low capitalization investors/traders as well as ones with a big bankroll. The bottomline goal is to beat the S&P500 while attaining an annualized sharpe ratio of 1.0. The model has been tested from 1989 to 2005. In that period the 1-10 ranked picks attained an annual return of +49% with a sharpe ratio of 1.56. It has been able to generate double digit profitable returns even during the bear market of 2000-2002 (although it did lose -3% in 2002 which is better than the -22% lost by the S&P500). It did, however, underperform the market in 1998 (+32% vs. +33% by the S&P). Here is a summary of performance based upon the ranks held: Rank 1-5: return +60% with sharpe of 1.57 Rank 1-10: return +49% with sharpe of 1.56 Rank 1-15: return +47% with sharpe of 1.66 Rank 1-20: return +41% with sharpe of 1.55 Rank 1-25: return +40% with sharpe of 1.56 Rank 1-30: return +40% with sharpe of 1.57 Rank 1-35: return +40% with sharpe of 1.57 Based upon these results the optimal number of positions to hold based upon risk adjusted returns is from the 1st through 15th. Keep in mind trading commissions have not been factored in. Of course past performance is no indication of future results. I am not investing real money based on this model just yet as I would like to "test drive" it a little first. Starting in February the first month's list shall be posted here. Happy investing/trading!
February trades: PBR: PETROLEO BRASILEIRO BER: BERKLEY W R CP CRDN: CERADYNE INC TKR: TIMKEN CO NEWP: NEWPORT CP VRTX: VERTEX PHARMACEUT IM: INGRAM MICRO A JNS: JANUS CAP GP CMN STK ILMN: ILLUMINA INC GT: GOODYEAR TIRE RUBBER A rather diverse bunch of picks. Industrial Groups: Oil Exploration/Drilling, Insurance, 3 Industrials, 2 Technology, 1 Pharma, 1 BioTech, 1 Finance. Average Mkt Cap: $12.61 Billion Average P/E: 21.37 (3 have none) Average P/B: 7.037
Do you have sufficient data to test your system against 1968-1982? That period might represent what we'll see during the next ten years.
The holding period is roughly a calender month. Sadly I dont have data from 1968-1982. It would be excellent to test out this model on an extended bear/ranging market. The model was built with in sample data from 1992-1996. The rest of the years is out of sample. So the only out of sample bear market conditions I could simulate was that during 2000-2002.
Alas, many wish to have such data, including myself. 2006 might reflect performance in 2004-2005. It's after 2006 that price action may get dicey again. FWIW.
Yes data like that would be valuable. I don't tend to make such macro forecasts on market actions. As I stated my goal was basically to beat the S&P500 by a sizable margin. If the ENTIRE market basically tanks then the model will simply not trade as the momentum based criteria will fail for the picks. One of the key criteria of the picks is to show an upward momentum. Way I see it even during extended bear markets, some sectors out there are moving and the goal is to identify those and capitalize on them.
mahras2: I have worked on systems in the last year combining some fundemental data and technical patterns. A question for you: I can obviously obtain technical data (closing price, highs, lows) for 20 years from a zillion sources but was wondering where you were able to get fundemental data going back more than 2 years? The best I have been able to do is get fundemental data (EPS, insider info, etc.) for the last 2 years. Thanks Mike
mahras2 , did you ever heard of Baseline co ? They have a feature called "different time periods" where one can roll back and see any data ( all the key stats that you mentioned ) as its was THEN. Perfect for your type of back testing . Thompson Financial bought them a couple of years ago .
I got the fundamental data from a friend of mine. He had the entire dataset since 1989 in a database and he sent it to me. I am not sure where you can get fundamental data from. Maybe someone else on these forums may know.