Trader's self-deception: distrust of forwards. - page 15
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
From the set of runs from 1998 to 2008, the best one for 1998-2001 is taken and its result for 2002 is written into the resulting equities, then the best one for 1999-2002 is taken and its result for 2003 is appended to the previous one, and so on. A lot of runs are obtained in advance for the whole history. Essentially a trivial sliding window. There is no magic and repeted here.
A little more about R^2.
To me this is a very powerful indicator, but not enough. In practice I have encountered that some TCs, can produce very good and smooth equity upwards. Their R^2 is very high and their parameter set will crack any even ................................
Good day. Vasily. Give me the formula for R squared. Not familiar with...
You are wrong. Multiple runs are not obtained in advance, but sequentially, i.e. the same procedure is repeated, but at different intervals. You don't even need to read the text - you can see it quite clearly on the gif. I've implemented exactly the same algorithm in my auto-optimizer, but I run each variable separately at each step.
Karl, why optimise again if the optimisation cloud parameters are the same. Learn the basics, as they say.
With Walk-Forward there is no optimised cloud. The point of repeated optimization of intersecting areas is that each time they are part of another optimization segment taken with a shift. As a result, one and the same point of history gets optimized several times - first, then in the middle and then at the end of the optimized interval, but the corresponding forwards appear to be consecutive and non-intersecting. So you need to study not only math but English too, if a simple gif doesn't convince you.
No one tests the whole section at once, as you mistakenly think, if only because it would be impossible to isolate the optimised parameters for the section as part of that large one.
For those especially gifted, I want to say it once again: the optimization is performed only once during the entire testing period. Optimize your TS for the period from 2000 to 2015. Choose the best run for 2005 - 2008. Then optimize the same CU for 2005-2008. Pick the best run again. Go figure, the results with the parameters will match up to the penny. This is what is shown in the hyphoto. If you want to kill yourself, do over-optimization at each iteration.
he automatic Walk forward test is a system design and validation technique in which you optimize the parameter values on a past segment of market data ("in-sample"), then verify the performance of the system by testing it forward in time on data following the optimization segment ("out-of-sample"). You evaluate the system based on how well it performs on the test data ("out-of-sample"), not the data it was optimised on. The process can be repeated over subsequent time segments.
The hyphoto doesn't show the results at all, it only shows the optimization and forward segments. Please don't litter my thread anymore.
he automatic Walk forward test is a system design and validation technique in which you optimize the parameter values on a past segment of market data ("in-sample"), then verify the performance of the system by testing it forward in time on data following the optimization segment ("out-of-sample"). You evaluate the system based on how well it performs on the test data ("out-of-sample"), not the data it was optimised on. The process can be repeated over subsequent time segments.
The hyphoto doesn't show the results at all, it only shows the optimization and forward segments. Please don't litter my thread anymore.
Are you even trying to understand what I mean? Give me an example, step by step, of how you understand wft. That would make it easier to explain. Although you probably shouldn't, because the case is hard.