Although P123 is not aimed at daytraders, it seems the technical factors could be used for “swing trading”. It also seems like you can use the backtesting here to test a simple pullback model, by defining an uptrend and pullback, and applying to an entire universe.
It also seems like all other backtesing sites only allow you to test a strategy on one or a few instruments at a time. Is this what others have found too? In other words I can test a moving average strategy on stocks ABC, XYZ, etc. but I can’t backtest this strategy on all stocks at once over a period of time.
What I am getting at, is trying to understand if certain technical set ups or strategies have any validity by testing over a vast universe as opposed to cherry picking a set of stocks or ETFs. It seems like all the technical GURUs out there say such and such a set up have a high probability of working, and other than anecdotal evidence I have never seen a clear test.
I made similar experiences. When backtesting technical systems from books or articles with P123, with rules as near as possible to the original, in the original time frame or in the last periode, my results were always much worse.
I think, it is much more difficult to find / develop a working technical system than a fundamental one. Reasonably safe technical systems generally have a low profit / trade, which makes them difficult to trade.
I am not a technical trader but I know a good one and I have read statistics of one of the best European individual traders of the last 10 years with audited performance. These guys don’t have extraordinary set ups, but they have extraordinary skills about knowing when to cut their losses and money management in general. In short, they have a very bad percentage of winning trades (far below 50%), but a stratospheric average gain/average loss ratio (above 4 or even 5).
That almost sounds like a trend following strategy, where you get chopped a lot in sideways markets, hence the low win %, but make large gains when a trend resumes.
It is interesting that P123 allows you to screen and backtest some fairly common technical strategies, and most of them fall pretty short of what the GURUs claim.
I remember picking up any first book on technical analysis at the CBOE library. I was legitimately curious about whether this pattern stuff would work, but the authors or salespeople only provided anecdotal or limited empirical data regarding their efficacies.
One of the reasons is that both the identification and interpretation of these patterns were subjective. It turns out that identifying patterns which are easy for the eye to spot is programmatically very difficult. And even then, they are up for interpretation. For example, a traditional chartist could defend a poor call by saying that there were patterns embedded embedded at different time scales which contradicted the main pattern or that the recent move was inscribed within a much larger and contradicting pattern.
For what pattern recognition I am capable of doing by long hand, P123 empowered me to evaluate (and debunk) many financial market myths.
While it did also allow me to validate persistent technical patterns, all of what I currently use are but refinements of Fama-French momentum and reversal.