An accident or an unrecognised pattern? - page 7
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
Can only be refuted. Or not.
A rabid pessimist.
A rabid pessimist.
The hypothesis cannot be confirmed. It can only be disproved. I'm sorry. (chuckles)
They differ in the way the test results change when the input parameters change or the algorithm itself changes slightly. If the picture changes dramatically, then we are dealing with randomness, i.e. fitting with history. If even with significant changes there is an underlying tendency to increase - it means that it is quite possible that it is not a fluke. Although there is no guarantee anywhere - we are dealing in principle only with probability - but it is in our power to bend it in our favour.
I was thinking along the same lines as you and I gave the algorithm a lot of freedom on all the variables that affect the outcome, in the hope that they would..,
that they would throw the system off-balance and put everything in its place, but that didn't happen.
I only have three parameters that can affect the fit.
SL from 40-70
TP from 40-70
and Market entry point from 0-12 (conditional)
I seem to have given enough freedom, 12,494 possible outcomes. But the test for 10 years, still showed a steady positive result.
Accepting it as reliable doesn't allow me common sense, as it would mean
that there is an algorithm for guessing random events with a guaranteed positive outcome, this is controversial.
Did a test today on another broker's quotes and the result is comparable.
If I think further and assume that the algorithm is still positive and it doesn't conflict with the theorist...
Well, then we can only assume that the price movement in the market is of mixed nature, which has a part of regularity, asYousufkhodja Sultonov mentioned above:
and it is possible to catch it. It is often said as a logical sign, but no one has ever proven it.
Last but not least, not less probable conclusion is that I have an error in algorithm or conditions that leads to false results.
The problem is that I can't find it yet.
That would be the easiest and most understandable way out in this situation)
"... Well, if we have distinguished regularity from chance, then how should it be qualitatively different from it, except for the cause of the event itself. ...". A random event has no cause but has a consequence. Consequently, there is no causal relationship. There should be a comma after "Well".
"...We know about the Central Bank rate hike, but we don't know how the market will react to it. ...". You don't know, we're guessing. But there's no problem with punctuation.
"...The definition of a pattern as a distinctive feature,at least, requires definition. ...". On the basis of the above, I give the definition: the presence of cause and effect relationships indicates the non-accidental nature of the process, i.e., the presence of regularities in the process. By the way, two commas are missing.
Alexei, about the commas, I accept the comment. I have nothing to say, Russian has always been difficult for me to write).
If only you were as dashing in sorting out not only commas, but the essence of the subject matter.
How to distinguish the test results, whether they are the result of a fitting or catching some useful patterns.
You wouldn't be worth your while))))
Thought I could estimate the proportion of non-randomness in the market.
For my TS it is 5-10 trades per day - from 10 to 18, i.e. 8 hours = 480 min. Suppose the pattern for an entry exists for 2 min (a pattern is not necessarily some combination of candlesticks)).
Then the share of non-randomness in the market for my TS (5...10)*2/480 *100%= ~2-4%.
I'm trying to understand it... I intuitively feel that some logic is present here, but I am not able to estimate it yet. I need to sleep it off and think it over again sober)
The hypothesis cannot be confirmed. It can only be disproved. I'm sorry.
Interested, I read. It is possible to confirm a hypothesis. In scientific parlance, that's a normal turn of phrase. Say a hypothesis may or may not be confirmed(s), confirmed by something, etc.
Interested, I read. The hypothesis can, after all, be confirmed. In scientific parlance, this is a normal turn of phrase. Say, a hypothesis may or may not be confirmed(s), confirmed by something, etc.
Well, if taqi, then sure.
I reasoned the same way as you, and in testing, I gave the algorithm enough freedom on all the variables that affect the outcomes, hoping that they would,
that they would throw the system off-balance and put everything in its place, but it didn't.
In addition to fitting, there are many traps in testing, when you do not correctly assess what the tester produces and the correlation of this process with the process of real trading, due to lack of information. Put the algorithm on a demo for a couple of months and everything will become clear.
Well, that's for sure. In any case I will do tests on demo, but this is long and will not ensure reliability either. I would like to find a faster proving method.