From theory to practice - page 1560
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
Stoploss is not a statistic, it's one specific realization of a process. It can be realised in one bar (or tick).
The CUSUM algorithm mentioned earlier is very similar to a stop loss exit. In your case the decision to exit a trade is made when a simple cumulative amount of increments (simply put, price increments over several bars) exceeds a certain threshold, and there the decision to exit is made when some (slightly more complex, but also determined by increments) cumulative amount is exceeded. In both cases everything can happen in a single bar, if it is large enough.
Here a single implementation of the process contains a sample of many increments or a few, but very large ones (even if the sample consists of only one number, but very large, it may be enough to refute the hypothesis about expectation equality to zero or about its sign)
The CUSUM algorithm mentioned earlier is very similar to a stop loss exit. In your case the decision to exit a trade is made when a simple cumulative amount of increments (simply put, price increments over several bars) exceeds a certain threshold, and there the decision to exit is made when some (slightly more complex, but also determined by increments) cumulative amount is exceeded. In both cases it can happen in a single bar, if it is large enough.
Yes, per bar or tick, so there is no amount, what matters is the magnitude of the price change.
Made an addition to the post above.
even if there is only one number in the sample, but a very large number, it may be enough to disprove the hypothesis of expectation equality to zero
How? I haven't seen even a hint of such technologies in the textbooks.
It depends on the nature of the phenomenon. If it is a physical measurement, it can be removed. Unfortunately, some outliers or gaps will not be deleted) Therefore, for example, we have to come up with models where these outliers are rare, but present - Maxim wrote about them recently.
Actually, the three sigma rule says that for a single number outside their limits we must reject the null hypothesis of the distribution, if we could not reject the very fact of obtaining that number.The CUSUM algorithm mentioned earlier is very similar to the stop-loss exit. Your decision to exit a trade is made by exceeding a simple cumulative sum of increments (simply speaking, price increments over several bars) a certain threshold
This is if the strategy is to hold on time, but almost everyone who looks for a TS with the help of indicators, order grids, channels, probably using graphical markers (- I don't know about that), does not use it
I do not know if it is correct or not - not to use position holding time, of course we may investigate it but in most cases we may profit only by price movements along the Y axis, the X axis (time) is used very rarely, or rather only for evaluation of working time of TS, in fact it is valatility. For the record it has now subsided, but from time to time the Renko chart on the forums is recalled en masse, there is no time axis, but it's like the same strategy where in the profit, and where not quite
If I remember the time to hold the position is used in High Frequency Trading, but the strategy there is definitely not built on bars - the ticks and liquidity are the main thing there
Here's another thought.
How do we calculate how quickly in the future the value of the distribution will return to the average?
After all, we want it to be fast, and not so fast that a trade is open and is hanging in a corridor.
This is if the strategy is time-holding, but almost everyone who searches for TS using indicators, order grids, channels, probably graphic markups (- I don't know about that) do not use it
I do not know if it is correct or not - not to use position holding time, of course we may investigate it but in most cases we may profit only by price movements along the Y axis, the X axis (time) is used very rarely, or rather only for evaluation of working time of TS, in fact it is valatility. For the record it has calmed down now, but at times the Renko chart on the forums is remembered en masse, there is no time axis, but it's like a strategy where you profit and where you don't.
SZZ If I remember correctly, the position holding time is used in High Frequency Trading, but the strategy there is definitely not built on bars - there are ticks and liquidity is the main thing
I don't think I had anything about exit by time. The price increments are summed up until they reach a certain value and we cannot know in advance how many bars will be needed.
Here's another thought.
How do you calculate how quickly the value of the distribution returns to the mean in the future?
After all, we want it to be fast, and not so fast that a trade is open and is in a corridor.
What is "distribution value"?