From theory to practice - page 604
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
And you have the Grail of pure gold in your picture.
I looked at the 100% quantiles of the increment sum distributions and compared them to the quantiles of the price itself for the same historical plot:
Black - 100% quantile for the sum of the increments in the sliding window=8 hours
Purple - 100% quantile for the price in the same window.
Taki are two different processes generating different probability distributions.
If we use the dynamic 100% quantile for the sum of the increments, we get the following picture:
I don't exclude that if you work in a sliding window = 24 hours (on the advice of my friend Bas'a) and use a dynamic 99% quantile, you can get a picture similar to yours.
Grail project #100500.
2. Profitability graphs of the trading system using tick MAs for specifying the deal entry.
2.1. ALPARI_MT5
2.2. ROBOFOREX_ECN
2.3. ROBOFOREX_CENT
2.4. ROBOFOREX_CENT longer period
Conclusions:
1. We can see traces of results grouping in some structures (there is a fish!).
2. We can see some similarity of formed structures on accounts of different brokers and types.
3. Nevertheless on accounts of different brokers and types even in one and the same time period (2.1-2.3) significant difference of quantitative characteristics is observed.
4. As the testing interval changes (2.3-2.4), the quantitative characteristics float, but the essential part of the general features remains unchanged.
Reference.
It took about 5000 core-hours (4 tests X 2 days X 50 cores) of tester to get only these results.
Grail project #100500.
....
1.10. Since strong disturbances break the channel trading model, the area of applicability of this model should not approach the areas of strong disturbances predicted.
etc.
All the 600500 pages cited here are an attempt to look at 1.9 alone with complete disregard for 1.10, which makes 1.9 meaningless at all.
Nah... I can't do this anymore. I'm going to go get a stopper. (с)
By the way, until I've had a shot, point 1.10 is correct only in the highlighted stating part. The following is incorrect.
Conclusions:
1. You can see traces of grouping results into some structures (there are fish!).
...
It took about 5000 kernel-hours (4 tests X 2 days X 50 kernels) of tester time to get these results alone.
Conclusions:
1. Traces of grouping results into some structures can be seen (there are fish!).
These structures will float over time. It is the same as in any constrained plot of SB - there will be "patterns", but they will change in the future.
For a qualitative check, it is better to take a yield plot (will show stability over time), and not on the fitting plot, but on the OOS.
These patterns will float over time. It is the same as on any constrained SB plot - there will be "patterns", but they will change in the future.
For a qualitative check, it is better to take a yield graph (will see stability over time), and not on the fit plot, but on the OOS.
Yield graph October 2015 - September 2018
Fitting interval March 2018 to September 2018.
Nah... I can't take it anymore. I'm gonna go get a shot. (с)
By the way, before I get drunk, point 1.10 is only true in the highlighted statement part. The subsequent is incorrect.
Horror. On an old laptop in a rudimentary SciLab, it only takes a few hours to figure out what fish to eat.(So you would share the methodology.
Otherwise it's kind of barefaced.
And in regard to point 1.10 a little later there will be pictures to illustrate the effect of the provision: "The scope of the model should not approach the areas of projected strong disturbances.
P.S.
Without trading in the area of strong disturbances (news).
Trading without restrictions on the area of strong disturbances.
The result is obvious. A rebound (inward) channel strategy is incompatible with areas of sharp price movements.
Well, you could share the methodology.
Otherwise it's a bit unsubstantiated.
And on point 1.10, there will be pictures later to illustrate the impact of the provision: "The applicability area of this model should not approach the areas of strong predicted disturbances.
Regarding 1.10. If the news etc. go to the border of the channel, it makes no difference to us. If from the border towards the centre, we work them out. In either case, we do not care if the channel shifts (collapses, in your terminology) or not.
Yield schedule October 2015 - September 2018
The fitting interval March 2018 to September 2018.
Another thing) good result. More averaging to be thrown out?
Even getting a random process from a random number generator will give some deviation from the ideal process.
Computerised RNGs are pseudo-random.
Another thing) a good result. More averaging to be thrown out?