Kumar,
At the start of this thread, you say you welcome feedback. That is a proper request.
It would also be proper if you are posting in order to generate interest in your model – to attract subscribers.
Both goals are fine. But it’s hard, or maybe impossible, to do both at the same time.
If you want feedback, then stop giving clues as to what you are doing. You need to open your entire trading system by setting the model to public visibility so we can see the details and give you good-quality feedback.
If you want to generate interest in your model, which is what I assume you really want since you have not completely opened it up, then it is my opinion that you have not succeeded.
A model that describes itself as being Value and Quality should rank higher than 79 and 72 under the Value and Quality style scores. That leads me to suspect that there may be things in your model that you may think of as Value and Quality but which in fact, are not. I know, for example, that many on p123, particularly those who take purely quantitative approaches, think they are doing value when they use ratios like PS, EV/EBITDA etc. based on quarterly sales or EBITDA numbers. They do it because it has “worked” in simulations designed to maximize backward returns. But neither is an acceptable example of Value or Quality because the latter needs to be based on figures that have high probabilities of being persistent – it’s hard enough top get that when using 12-month figures – using figures from just one quarter, which don’t even encompass a full annual cycle different quarters have different characteristics) is low-persistence in the extreme and is the extreme opposite of value or quality. That sort of thing is momentum. Yet your models’ momentum score is very low, the lowest of the three. Now, I have no idea what you’re doing in terms of genuine fundamental modeling.
Yet your performance is pretty good – too good. You said at one point in this thread that:
And, you checked “NO” for use of hedging.
My reaction: No way! There is no way a long model (which is the only kind permitted in Smart Alpha) could gain 38.39% in 2008 nor could such a model have gained 58.38% in 2001 and 48.87% in 2002. That’s why I asked the question I did about your use of hedge-like factors in the context of Buy rules. And now, you say you did use some sort of IBD type system. While IBD does have fundamental aspects of its work, it is also very well known for technical, and use of backward-looking technicals (since even technical folks got killed in 2008) is the only way you could have achieved the simulated 2008 performance. In other words, I even wonder if what you did was really all that consistent with what IBD does. It looks to me like plain old data mining including use of SPEPS series factors.
So putting all this together, I do not see a Value Quality model enhanced by use of IBD technical principles. All the clues I see point to heavy data mining.
I may be wrong. But it is important that you and other Smart Alpha strategists be aware of what the numbers say – which, in many cases, does not match verbal descriptions. If you want to persuade me that I am wrong, you would need to completely open up your trading system.
