Statistical Analysis of Intraday Data - Any worthwhile resources?

So, the consistency between win% proves (or is the best indication) that the out-of-sample is still benefiting from the same pattern as the in-sample. And, that you didn't produce a positive or negative out of sample result by sheer luck. Correct?
 
xandman said:
So, the consistency between win% proves (or is the best indication) that the out-of-sample is still benefiting from the same pattern as the in-sample. And, that you didn't produce a positive or negative out of sample result by sheer luck. Correct?
More...

Yes, but only when random entries also show consistency between in-sample & out-of-sample.
 
I had to take a little break in my work, but I'm now back on track and interested in completing this in not too long.

Basically, at this point, I'm "done". I'm just spending time thinking about stuff I may not have thought of before I move on to testing and programming. I have one question:

It would be interesting to objectively define each day in my data set as being either in an up trend or down trend.

This could easily be determined by using for instance a moving average. I'm sure I must have heard somewhere that above the 200-day MA is an up trend and below is a down trend.

Any better? It doesn't have to be super precise. The idea is that days in a down trend environment would behave differently than in an uptrend. The 200 day MA would be a simple way to determine this. Thoughts?

Of course, a market could also be in a range bound environment, but I'm not sure how I could quantify that.

Possibly with the market being contained by two averages or bands. Hmm...

Suggestions are welcome. :)
 
I've been tinkering with it a little and I'm thinking perhaps the 200-day average is too slow and that perhaps I'll have to use the 50-day average or similar.

Then: If day close above 50-day average --> Up trend. And vice versa.
 
Howard said:
I've been tinkering with it a little and I'm thinking perhaps the 200-day average is too slow and that perhaps I'll have to use the 50-day average or similar.

Then: If day close above 50-day average --> Up trend. And vice versa.
More...

You may want to watch the first part of the video of Simons' interview and read this article.
 
Sergio77 said:
You may want to watch the first part of the video of Simons' interview and read this article.
More...

Thank you for that article. I read it, but I'm not sure if it made me any wiser. : )

To clarify; I'm not interested in creating a trend following system. I'm merely interested in an objective way of determining if we're currently in an up trend or down trend environment. It can be crude, just so that I'm able to differentiate in my data set. I may NOT need this though. It was just an idea I had.

Upon writing this, I'm thinking these two, respectively, could possibly serve as a metric to quantify trend:

1. If >60% of the last N days is positive --> Up trend.

2. If today's close > 2X N day close --> Up trend.

And vice versa.

Thoughts?
 
I've been looking at several moving averages, but I'm thinking I'll just go with the 30-day Moving Average using a Close below as a down trend and vice versa.

Not perfect, but it should be a way to differentiate between an up trend and down trend environment. The idea being that market patterns are different in the two conditions.

Thoughts?
 
Howard said:
I've been looking at several moving averages, but I'm thinking I'll just go with the 30-day Moving Average using a Close below as a down trend and vice versa.

Not perfect, but it should be a way to differentiate between an up trend and down trend environment. The idea being that market patterns are different in the two conditions.

Thoughts?
More...

TA predicts price constraints which enable anticipating price reactions until new constraints are defined. The biggest and best investment funds have traders predicting those constraints ....and predicting when price has indicated it's respecting the current constraints or establishing new constraints. This is an unquestionable fact.

Howard, schizo, baro-san, i am nobody, surf and everyone else is free to use this info in any manner u like. However I want to caution everyone and surf in particular, if u want to present examples of descriptive (non-predictive) TA then keep your examples of the worst of the worst TA off this thread. Everyone knows not all TA is predictive and Howard likely isn't interested in seeing examples of non-predictive TA.

Howard, it's unlikely anyone will provide u tangible models of predictive TA but now u know the preferred measure that determines the relative validity of the components you're using to quantify if today is part of an uptrend or downtrend. I encourage u to flag and report disruptive posts if u feel they dilute the content of your thread or make it more difficult for you to decipher truth.
 
You must log in or register to reply here.
Back
Top