I've always wondered what guideline to apply to determine the all-important "time-frame". Is ATIT (average time in trade) the metric to apply ? For instance, I trade and discover my ATIT is 60 minutes. What time frame is best ? 1 minute, 5 minute, or 10 minutes ? And within that time frame, what value should be used in the trend calculation : close, (H+L+C)/3 or VWAP or ???
quantitative trend determination
- Thread starter afx111
- Start date
Can't help you there I'm afraid. I pay almost no attention to the time dimension in my trading.
By reading headlines.
Neither one is perfectly sideways.
I assume you understood the gist of the post. If you add the percentages in a linear fashion, then yes, the signal is perfectly sideways.
Of course that wasn't the gist of it, was it?
Does this not look sideways to you?
You are able to disprove mathematical correlation on a sampled data sequence with your visual analysis looking out a million data points into the future, yet, you can't seem to grasp this sideways vs. trending concept using binary vs. continuous data with a simplified example?
Granted it's not compounded, although, I assume you are intelligent enough to imagine such a scenario can be made to corroborate the meat of my original point.
If you want to be an antagonist and chime in with your corrections, next time try to focus on the relevant point of the post; not minute tripe.
This is a perfect example of trend measurement and the importance of choosing a time frame: in a short time frame, this is perfectly trending...first up, then down....a scalpers delight. On a longer time frame, the linear regression line would be flat i.e. trendless....terrible for doing anything except selling option credit spreads.Quote from dtrader98:
[/B]
Bottomline: know your time frame, stick to your rules.
I think we can all agree that compounding is a pretty significant "granted".
So everyone is clear - the original example pattern of {+X% -X%} repeat ad nauseum is not "perfectly sideways", it is in fact a relentless trend.
You gave a specific example that not only did the opposite of what you said it did, it also did exactly what you said it didn't do.
It was a mistake, happens to everyone, suck it up.
wrong again. The sequence is perfectly sideways (since you can't seem to agree with math, your eyes should at least be able to ascertain the direction from the plot I posted), and since you want to be a stickler about it, let me educate you a bit. The two common forms of financial price models are product of simple returns, and linear sum of log returns; notice I expressed neither preference in my vectors. The majority of financial literature implicitly assumes the latter form. So you see, it is perfectly consistent (although a trivial point) with my illustration.
The example made the point exactly; either 1) you wrongly assumed the returns model or 2) you are not competent enough in financial math to understand the difference. Either way-- your comment was not only incorrect, it was a waste of time. Nice try, though.
And obviously, you didn't comprehend the flaw of using hamming distance as a trend metric, which was the point of my post. But seeing as you have problems discerning the definition of sideways, it makes sense.
Class dismissed, I'm done with you.