Discussion of article "Exploring Seasonal Patterns of Financial Time Series with Boxplot" - page 28
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Ran not by increments, but by signs from the beginning of 2014. The best result
No connection by signs on H1.
By increments I have a different result
You can use the sliding window to measure correlation. And then look at its statistical characteristics. I think it will float a lot.
You can, but you can see it in the scatter plot above. There are emissions, but you're looking at the average.
I'm not saying anything, it's just research, I'll check it out later.
I'm getting a different reading on the increments
I have honest, non-overlapping intervals.
I have honest non overlapping intervals.
You know what the problem is - sigs + spread may not cover the spread, especially in 0-1 hours. Doc (coconut) has already made a similar TS in the MO branch, he traded at 0.
Otherwise, let them overlap, everything is normal, how else? Predict the closing price, trade deviations from it.
Let them overlap, it's fine, how else?
With overlaps, any random will show high correlation.
With overlaps, any random will show high correlation.
I'll split it into 2 independent pieces and check it then, I'll upload it later.
But no... it's nonsense... the sequence of cloze will not be preserved, the curves will be different. What's the point of looking at it like that? It's like taking two left curves.But no. That's nonsense. The sequence of cloze will not be preserved, the curves will be different. What's the point of looking at it like that? It's like taking two left curves
My bottom line.
Forum on trading, automated trading systems and testing trading strategies.
Discussion of the article "Investigation of seasonal characteristics of financial time series using Boxplot diagrams"
fxsaber, 2019.12.16 07:52 AM.
I ran not by increments, but by signs from the beginning of 2014. Best result
There is no connection by signs on H1.
My bottom line.
the point is the correlations are different at different hours at my place.
The difference is obvious, how do you reconcile that with the hypothetical results on the SB? I don't know:
and for each lag the correlation varies, one pair of clocks is dominant, the other.
In the second picture, it's all sliding into the positive area.
lag = lagged closing price, e.g. 0 bar divided by the 10th bar. 1-close[0]/close[10]
The increment of the current hour is highly correlated with the increment of the previous hour, it turns out. The greater the lag the greater the correlation, for certain hours 1-close[0]/close[10] correlates with 1-close[1]/close[11]
I agree with fxsaber that the correlation here will be large because of the overlapping intervals:
1-close[0]/close[10] = (close[10]-close[0])/close[10] =(close[10]-close[1]+close[1]-close[0])/close[10]
1-close[1]/close[11] = (close[11]-close[1])/close[11] =(close[11]-close[10]+close[10]-close[1])/close[11]
The selected parts match, which will lead to a correlation
I agree with fxsaber that the correlation here will be large because of the overlapping intervals:
1-close[0]/close[10] = (close[10]-close[0])/close[10] =(close[10]-close[1]+close[1]-close[0])/close[10]
1-close[1]/close[11] = (close[11]-close[1])/close[11] =(close[11]-close[10]+close[10]-close[1])/close[11]
The selected parts match, which will lead to a correlation
there are pairs of clocks for which the correlation is 0.
what does this mean? I can't speak for maths and great scientists right now, I'm just talking about what I see
For example, I don't understand at all what Saber is looking for on non-intersecting intervals, because there will always be 0 there and only on occasion not 0
So he's looking for correlation in the mean/median increment bias, in my opinion, which is already there in that boxplots article