Econometrics: let's discuss the CU balance sheet. - page 8
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
There, at last. This is exactly the topic of the thread.
You can't compare a 240 with a 52. For me, it is a matter of principle. The point is that 240 and 52 are realisations of random variables. That's how it fell out. And the basic question is: what is the probability that in the future it will be the same, or almost the same, or not at all the same?
Residual analysis should answer this question, and it is far more important than the size of the test profit.
What idiot would throw out a profitable TS if it has or does not have fat tails in the distribution?
There is a profitable TS giving, for example. 30% profit per month for a year, but it has thick tails in the balance chart model residuals distribution. So?
well you sure are dumb)) I've already explained that this is a sign that past results cannot be trusted. Including "30% profit per month for a year". You can do the same thing with Martin and there will be equity drawdowns, which will actually be thick tails in the negative zone. However, some people are smart enough not to trade
well you are definitely dumb)) I have already explained that it is a sign that past results cannot be trusted. Including "30% profit per month for a year". You can do the same with Martin and there will be equity drawdowns, which will actually be thick tails in the negative zone. However, some people are smart enough not to trade that
Stupid, forgive me!
Reread the thread eight times - there's no equity, just a balance chart..... Maybe it's you who's stupid?
Try again - you have a profitable TS, built (for some reason) a regression model using the balance chart. The residuals are normally distributed, but with thick tails. And what does this mean?
P.S. I'll write the correct answer below...
You have gone out of your way to answer this. DEMI gave an answer from a textbook on regression analysis, but there is little regression analysis when modelling kotir. And it doesn't show normality anywhere.
So it doesn't matter if it' s regression analysis or whatever. The sum of a large number of independent distributions tends to be normal.
What idiot will throw out a profitable TS if it has or does not have fat tails in the distribution?
There is a profitable TS giving, for example. 30% profit per month for a year, but it has thick tails in the distribution of balance chart model balances. So?
My understanding is that fat tails are variable variance, outliers in the quotient. And this is not a good thing. The previous TC result cannot be extrapolated into the future. This suggests that the TS will "go stale". This is a characteristic of the TS built on TA, where the statistics of the TS are not controlled. You make it - it works for a month, three months, my record is 6 months. That's why I started to think about the topic of this topic.
If the residual is stationary, then the profitability of the TS will be unchanged with one specification: it will fluctuate within the variance. If you subtract the variance from profitability and still get a plus, then it's perfect.
If the residue is normal, the sigma rule will work.
.
That's the way it should be.
Stupid, forgive me!
Reread the thread eight times - there's no equity, just a balance chart..... Maybe you're the dumb one?
Try again - you have a profitable TS, built (for some reason) a regression model on the balance graph. The residuals are normally distributed, but with thick tails. And what does this mean?
P.S. I'll write the correct answer below...
only equity, if faa balance has analysed it is incorrect as it hides the drawdowns within trades. But these are purely MT terms.
if the MOI is unchanging, that's good. The system can be traded - robustness. If not, there is no confidence in past results and the system cannot be traded.
My understanding is that thick tails are variable variance, outliers in the quotient. And this is not good. The previous TC result cannot be extrapolated into the future. This suggests that the TS will "go stale". This is a characteristic of the TS built on TA, where the statistics of the TS are not controlled. You do it - it works for a month, three months, my record is 6 months. That's why I started to think about the topic of this topic.
If the residual is stationary, the profitability of the TS will be unchanged with one specification: it will fluctuate within the variance. If the variance is subtracted from the profitability and it is still positive, then it is ideal.
If the residue is normal, the sigma rule works.
.
It should be so.
I will publish the answer as promised above - we need normality of residuals only for reliable interval estimates (to calculate the width of the confidence interval) - an important procedure for applied problems. But if interval estimates are not needed, we can build a regression for any distribution of both the observed value in the sample and the residuals.
So normally distributed residuals, abnormally, with thick tails, with thin tails, without tails - all the same..................................................
By dropping the trend, you are dropping the main thing, because it is the trend that determines the direction and slope of the process in question. To make it clearer, consider two realizations that are similar to the previous ones, but the first is +240% and the second is -52%. Take out the trends and you get noise. In this case, what will the analysis of this noise yield? In this way the "baby" is spilled out and only "water" is left.
I think it's not the first time I've explained it to you personally - no one is throwing out a trend anywhere.
Wrote above: the trend is "clogging up" the statistics and gave the proof on the calculations. That is why I highlight the trend in order to have real figures on the noise.
As far as I understand, fat tails are variable variance, outliers in the quotient. And this is not good. The previous TC result cannot be extrapolated into the future. This suggests that the TS will "go stale". This is a characteristic of the TS built on TA, where the statistics of the TS are not controlled. You make it - it works for a month, three months, my record is 6 months. That's why I started to think about the topic of this topic.
If the residual is stationary, the profitability of the TS will be unchanged with one specification: it will fluctuate within the variance. If the variance is subtracted from the profitability and it is still positive, then it is ideal.
If the residue is normal, the sigma rule will work.
.
That should be how it works.
so if the profitability of the TS is assumed to be unchanged, i.e. mo=const, why some complicated detrending instead of just subtracting the linear trend from equity? I.e. trend model y=kx, where k=mo, x-transactions, y-equity