Discussion in 'Data Sets and Feeds' started by harrytrader, Dec 17, 2002.
I would like to see how my model behaved during these old days
I ran some tech analysis on daily Dow Jones Industrials closes of 1932, got some interesting results, looked like it would have been a tradeable market.
I have daily datas but only closing datas not high, low if you have them could you send them to me I will publish the results and we could compare your analysis with mine ?
I can do with closing datas but it's much better to have high and low since my model is used to predict tops and bottoms.
I like designing systems that would have worked in the past, since I don't care about the present anyhoo.
Haha I don't design models for the past. My model is designed once 4 years ago and HAS NO PARAMETER NO OPTIMISATION FOR ANY MARKET FOR ANY TIME PAST OR FUTURE can you imagine that this only could be possible ?
For example you can ask me ANY INDICE OR STOCK and I can produce the prediction model straigth away without even optimising anything. This is a zero tolerance degree of liberty ie the model and even the values of the model is the SAME for ALL INDICES STOCKS PAST AND FUTURE.
You can see an example for today @
there's nothing as old as yesterday's newspaper...
what good would those prices/data be for present day's trading?
this data using your model?
Harry, I only had closes, daily. Charted, it looked like a thinly traded issue of the current era.
The usefulness of it was I could see that my personal favorite tech analysis that seems to work in the last few years would have worked even in 1932. That is a good thing to know, it surprised me. I don't have to be so concerned about "curve fitting" and all those headaches too much, maybe an occasional optimization, or optimization on a set of growth issues versus a set of bottom fishery issues for example, otherwise it's a stable system.
I visited a trader's office a few weeks ago and saw some Gann or Fibo stuff that filled in with signals where my own favorite stuff would not tell me anything, will be interesting to test that new stuff on the 1932 data some day. Then there is HarryTrader's stuff which I don't comprehend yet but it looks very interesting.
Harry, if you find that data and get some results it would be interesting to see how you do.
You made a joke but in fact I have already thought about creating artificial datas and to compare with stochastic process (for example using Poisson's law to generate crash) creation to see the difference.
This is my last intervention in this forum. I won't come back any more since this is a commercial forum and I can't post what I want to freely post.
So good continuation to all.
Many People say that stock market has changed that it works diffrently than yesterday. I want to see if is true because as far as I can say it works the same for my model so for me it is not true that the market has changed its behavior: it's just the proportion but from the structure point of view it has not changed. Since there is rarely a structure model of the market (in fact the scientist who will discover it will probably receive the nobel price) one can't perceive the persistence.
On practical point of view this will prove that my model is robust horizotally (applying to any market) and vertically (appying to any time).
Separate names with a comma.