The second implies data has been used selectively to arrive at the desired solution, which would obviously render the outcome invalid. I was extremely careful to avoid the latter - I developed the strategy mainly for my own investment purposes.
My comment related to this was the methology you used for coming up with your strategy. Test a basic premise, add a little bit of this and a little bit of that. Find some timing indicator that would have kept you out the worst of times. With enough this and thats you can always find a strategy that looks great, even if the result may be somewhat random. Your process isn't very different from what lots of other people are doing with their strategies, including strategies people use here. There's also lots of additional statistics you could add to your research, like varying parameters somewhat and see how that affects results, hold more/less stock (top 5, top 15, 5-15), analyze actual trades and figure out how much of your returns are due to very few trades.
Also, it is built on sound underlying principes. High RS coupled with a value factor is well known as a powerful combination, and the Timing Indicator enhanced performance still further by keeping you out of the market when it gets into one of its periodic blue funks (like right now).
Momentum effects have been documented, yes. But using this as evidence for strategies indicating a +30% yearly returns going forward is stretching it.
The market for timing indicators is very fickle; vendors appear with great backtested strategies and disappear within a year or two later since their timing usually falls down in forward testing, only to reappear with a new great timing system based on a revised backtested strategy.
You are also correct in questioning whether it works in real time as well as in back testing.
The algorithm was frozen in May '06, since when it has shown a gain of 38%, versus a fall of 8% in the All Share Index. Only time will tell, of course, but I am personally pretty confident
I hope that will work out for you and your customers.
I may have misread your material, but disclosing the facts about when the system went live and not certainly would make it easier to evaluate the research versus marketing you do (which was what I was asked to do). The front page certainly does not disclose any facts about what is backtested returns and what is not. Also, you might want to consider sharing a little bit more information from your live system without requiring logging in, possibly lagging timing indicators and similar to avoid interfering with you fee-based business.
The second implies data has
The second implies data has been used selectively to arrive at the desired solution, which would obviously render the outcome invalid. I was extremely careful to avoid the latter - I developed the strategy mainly for my own investment purposes.
My comment related to this was the methology you used for coming up with your strategy. Test a basic premise, add a little bit of this and a little bit of that. Find some timing indicator that would have kept you out the worst of times. With enough this and thats you can always find a strategy that looks great, even if the result may be somewhat random. Your process isn't very different from what lots of other people are doing with their strategies, including strategies people use here. There's also lots of additional statistics you could add to your research, like varying parameters somewhat and see how that affects results, hold more/less stock (top 5, top 15, 5-15), analyze actual trades and figure out how much of your returns are due to very few trades.
Also, it is built on sound underlying principes. High RS coupled with a value factor is well known as a powerful combination, and the Timing Indicator enhanced performance still further by keeping you out of the market when it gets into one of its periodic blue funks (like right now).
Momentum effects have been documented, yes. But using this as evidence for strategies indicating a +30% yearly returns going forward is stretching it.
The market for timing indicators is very fickle; vendors appear with great backtested strategies and disappear within a year or two later since their timing usually falls down in forward testing, only to reappear with a new great timing system based on a revised backtested strategy.
You are also correct in questioning whether it works in real time as well as in back testing.
The algorithm was frozen in May '06, since when it has shown a gain of 38%, versus a fall of 8% in the All Share Index. Only time will tell, of course, but I am personally pretty confident
I hope that will work out for you and your customers.
I may have misread your material, but disclosing the facts about when the system went live and not certainly would make it easier to evaluate the research versus marketing you do (which was what I was asked to do). The front page certainly does not disclose any facts about what is backtested returns and what is not. Also, you might want to consider sharing a little bit more information from your live system without requiring logging in, possibly lagging timing indicators and similar to avoid interfering with you fee-based business.
Keelix