Yes it looks like many of the changes occured due to TA changes. Although several new picks should be included due to new fundamental information being avaliable after the earnings reports.
Got ya. I wonder how the bottom(bad fundamentals and TA conditions=short signal) 10-40 performing , now if they are lagging {or down big} the top 30 , then you really got an edge.
The max picks generated is 35. However, historically the top picks have outperformed the bottom ones. So 1-5 outperformed 6-10 which outperformed 11-15 and so on. This can also be seen from the differences in the performances of Rank 1-10 and Rank 1-20 picks as Rank 1-10 is beating the Rank 1-20. I will try and cook up a short signal model and post the results of that whenever I get some time.
Yes such occurances do (and will) happen from time to time. But, based upon data, one is better off being on the correct side of momentum. A simple momentum test I ran, on the universe I use, in which the top 10 stocks based on the past 4 week (roughly a month) is picked with monthly rebalancing. It shows an annualized return of 30.4% since 1989 (naturally with a lot of volatility).
If I understood correctly you reset your portfolio once a month. This creates a discontinuity and introduces a delay between market changes and when your system adapts to the new reality. Have you studied the possibility and the effect of reducing the period between portfolio adjustments? I think that ideally these adjustments should happen as needed and not at fixed intervals, because any "granularity" will introduce an uncontrollable noise / distortion.
Yes I rebalance the portfolio once a month. I understand that distortion occuring but the key reason why I have not reduced holding periods is because: A) commissions/spreads/slippage begin to play an increasingly important role. B) decreasing holding periods will place greater emphasis on the "technical" part of the system. As the technical indicators I use is very simplistic I would guess that increased weighting will hurt results. I have not yet done any studies or tests dealing with smaller holding periods. I may pursue this as soon as time permits.
Going to close off all positions tomorrow at 3PM. As of now the model has gained (after realistic slippage/fills/commissions) +2.73% in February. Barring any disasters, it should post a market beating return for February.
Excel and Access. Database in access. A VB program is run which spits out results on excel. I got the data from a friend of mine. However, earlier in this discussion a firm called Baseline (now part of Thompson) provides fundamental data so you may want to check that out.