Traders gain some knowledge, win, their emotions kick in, and they die.
What a great site! (Dailyfx.com rave) 4 replies
A Great Indicator needs a touch of a great coder 7 replies
Great Posts By Great Traders 79 replies
Average system + great discipline =great results? 44 replies
DislikedImproved Bollinger Bands indicator. In this comment PipMeUp discusses problems with the standard implementation of the Bands indicator. His objections are: 1) The number of samples for the deviation should be typically much bigger than the one for the MA. This implementation allows this, as you can specify the numbers separately. 2) The lag of the simple moving average is half of its period. If we subtract the moving average at position from the price at candle ...Ignored
Disliked{quote} After a second look, I am wrong in saying this, this would actually be a correct procedure in its world. The current SMA is of course the mean value over the past period so it does make sense, even though not taking trends into account is painfully imprecise. Sorry, kIgnored
QuoteDislikedTherefore there are three very important assumptions that are made when using parametric statistical techniques:
- All the treatment populations are normally distributed (the normality assumption).
- The normal distributions have the same standard deviation (or variance; the assumption of homogeneity of variance assumption)
- The data is taken from an interval or ratio scale (Since it is impossible to have normal distributions on any other kind of
...
Disliked{quote} I'm late to the debate but just wanted to add that if you perform time series analysis using the mean and standard deviation, you are implicitly assuming that the distribution is normal.Ignored
Disliked{quote} Hi Kprsa! I've two questions for you 1) Bollinger Bands: Since the samples for the deviation should be greater than the one for the MA, i would like to know if there is a "optimal" relation between the two samples... how much bigger the deviation samples should be??? 2) Recurrence Indicator: In your wonderfull indicator you give us the "Max width of the problem zone <k>" for the different kind of bars.... if possibile it would be usefull if the indicator state the date and time where the max problem has been measured. Just to give a complete...Ignored
DislikedImproved Bollinger Bands indicator. In this comment PipMeUp discusses problems with the standard implementation of the Bands indicator. His objections are: 1) The number of samples for the deviation should be typically much bigger than the one for the MA. This implementation allows this, as you can specify the numbers separately. 2) The lag of the simple moving average is half of its period. If we subtract the moving average at position from the price at candle ...Ignored
Disliked{quote} I admit, I am puzzled by this response. I had hopes that we could have an intelligent exchange of ideas in this thread. I was mistaken. I won't make that mistake again. Best regards, FXEZIgnored
Disliked{quote} if you perform time series analysis using the mean and standard deviation, you are implicitly assuming that the distribution is normal.Ignored
Disliked{quote} This is my view of indicators like standard deviation bands (BB) because you can't properly estimate the % of the data outside the bands based on the normal distribution.Ignored
Disliked{quote} It is also worth noting that the mean is referred to as the expected value. IMO the mean doesn't very well represent the expected value. That said, a shifted mean does a much better job, however you are late half the periodIgnored
Disliked{quote}and so for applicability to real-time trading you must predict the adjusted mean several periods into the future, which is essentially just a variation of predicting where price will be in 1/2 length in the future (a really hard problem).Ignored
Disliked{quote}Also worth noting is that implicit in the standard deviation calculation is the assumption that the mean is stationary. This is also not true for market data.Ignored
Disliked{quote}Also in the quote above, it assumes market data has a constant variance. Market data is heteroskedactic, or in other words has time varying variance.Ignored
Disliked{quote}IMO one of the most important features of "indicators" is timeliness. This is why in practical terms I've found more application of real-time indicators such as standard deviation bands (with the above defects) than of mean-shifted measures, due to the hard right edge of the market, which IMO seems to be the most important factor (timeliness).Ignored
Disliked{quote} I agree that it should be no difference when the price mean is flat with respect to time. Then also the lag of the moving average will be equivalent to zero, so in this case the lag correction is unimportant. But things do change in a trend. The blue line in the plot below is the standard SMA, the red is shifted back by a half of its period. Would you calculate the standard deviation based on subtracting the data off the first or the second line? Do you think it is the same in this case? {image} {quote} In principle one part of the problem...Ignored
QuoteDisliked30/20 | 30/30 => h_left/h_right
----------------------------------------------------------
73,259 | 77,615 => Prob of Trans bar resolved
82,1688 | 85,376 => Prob of Frac bar resolved
Disliked{quote} Hi mate, About the prob calc there ... using RS_V42 if i get the value like this below for h_left=30, h_right=30 & 20 : {quote} At current situation, we have a blue fractal arrow, and we are on the 3rd bars to the right of that blue arrow. a) If i want to know the probability of that arrow will be totally resolved in 20 bars after its position, must i calculate it like this ? (85,376 - 82,1688) / (100 - 82,1688) b) Does it mean that we don't care at which position we are now (after that blue arrow), we just need to know that h_right > current...Ignored
DislikedK I am sorry to request something on this forum but having read your statement on the shouts, I had to post it here. I have found something interesting. And I think you are the one who can do it easily. I also sent you a request on Stevehopewood's website. Kindly have a look at the attached pic. Look at GBP crossing the zero level, and look at the GBP pairs. Can we please make the attached indicator a multi symbol indicator as shown in the picture? The attached indi is a MTF and I wish you could add multi pairs option to the attached indicator....Ignored
DislikedDear K The one shown in the pic is the one created by Mladen, but it is not MTF.Ignored
Disliked{quote} I'm late to the debate but just wanted to add that if you perform time series analysis using the mean and standard deviation, you are implicitly assuming that the distribution is normal. This is the case with parametric statistical measures. The implication is that if the distribution is not normal, then you will draw false conclusions about the true nature of the distribution. This is my view of indicators like standard deviation bands (BB) because you can't properly estimate the % of the data outside the bands based on the normal distribution....Ignored