Discussion of article "Money-Making Algorithms Employing Trailing Stop" - page 4
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
My posts were directed at the author of the article due to the fact that there is no comparison in the article between systems with trawls and the same systems without trawls, if those systems have SL and TP. I didn't say anything about the usefulness or uselessness of trawls.
And this is already yours, IMHO!
Google: "Separate testing of market entries/exits." There is a lot of literature and references. I have no reason not to trust them, especially when you write, test and analyse them yourself.....
And this is already yours, IMHO!
Google: "Separate testing of market entries/exits." There is a lot of literature and references. I have no reason not to trust them, especially when you write, test and analyse them yourself.....
Googled it. Found only many copies of the same article. "Separate testing of trading system entries and exits".
http://wellforex.ru/publ/forvard_testy/stati_forex/razdelnoe_testirovanie_vkhodov_i_vykhodov_torgovoj_sistemy/7-1-0-76
This article confirms my thesis. For testing, they tear off an input(output) and sew it to a "simple" output(input). And they compare not the tested inputs, but the whole system with tested inputs and simple output. the simple input cannot be evaluated in any way.
Indeed, how can we compare a random entry and a reverse entry (an entry in the opposite direction to the previous trade)? Everything will depend on the output stitched to them.
By the way, if google gives you something more valuable, share the links at once. Google's output changes over time and valuable links can simply disappear from the output.
This is the first time I meet trailing profit, there are many written about trailing stop, but I meet trailing profit for the first time.
I have a question about the trailing profit algorithm, in this line:
multiplication 1.01 is it you add another 1% to the main TP? Or what is it?
If you don't mind, please comment on this line.
This is the first time I meet trailing profit, there are many written about trailing stop, but I meet trailing profit for the first time.
I have a question about the trailing profit algorithm, in this line:
multiplication 1.01 is it you add another 1% to the main TP? Or what is it?
If you don't mind, please comment on this line.
This is the remnants of my debugging agony. This is also where MathAbs came from. It seems like everything should work without them, but sometimes I get an error like wrong stops in the query. And somewhere in 120 trades that you can't get to with your hands, and even if you do, it's not clear what to do.
Let's say we set the TR in this tick. In the next tick the price has moved a little, the condition of TR shift is triggered, and a request to the server with a new TR is sent again. But because of NormalizeDouble the new TR is equal to the old TR and the order has no sense. This was done in 1.01 to make the server less often and get an error.
Although I may be wrong here. And 1.01 should have been somehow linked to digits in NormalizeDouble, but then I just lost my temper and forgot about it after 1.01. "Hackwork, Sir".
At the algorithm level, it's like increasing the TR by a percentage. But the algorithm was debugged with 1.01 and the TR optimum is so flat that it will still work.
I have already written about the transfer to breakeven before, it is put there because of the fear that the bet will not play, make more flexible systems and BU will not be needed.
As for trawling in general, it was already said above, when trawling stoploss is reached more often than take profit (according to statistics), and this respectively cuts the profit and increases losses. Even if sl is moved to BU you still get less than when tp is reached. Hence mat. expectation drops.
It is already difficult to win back your money from the market, and here we play a giveaway with our own hands.
In general, IMHO, sl and tp are needed for emergency (force majeure) depo protection. In a normal situation it is necessary to exit in accordance with the forecast about changes in market behaviour. By placing sl and tp or trawling them you pull patterns on the market, but the market quickly calculates the most significant patterns and works them out effectively.
I completely agree with you and use the same in practice. But there is a nuance on the topic of the article.
TC is not supposed here. Since we cannot distinguish a reversal into a new trend from a correction, we do it simply: enter the market, if stop, then reverse and trawl. If we draw a ZZ, it's an attempt to ride these zigzags that we can't predict and don't try.
Why it will work, who knows.
The number of trades the author of the article has is a bit small. It looks like "buy and hold" and accidentally found a ZZ with a large distance between reversals.
So, there is something, although I have not figured out the validity.
I completely agree with you and use the same in practice. But on the topic of the article there is a nuance.
Here we do not assume a TS. Since we cannot distinguish a reversal into a new trend from a correction, we do it simply: enter the market, if stop, reverse and trawl. If we draw a ZZ, it's an attempt to ride these zigzags that we can't predict and don't try.
Why it will work, who knows.
The number of trades the author of the article has is a bit small. It looks like "buy and hold" and accidentally found a ZZ with a large distance between reversals.
So, there is something, although I haven't understood the validity of it.
The algorithm works only because sometimes (during crises or when central banks regulate) the exchange rate stops being chaotic.
The number of trades is really small. It bothers me myself. The number of trades can be increased at most twice by reducing TS or TP to 300. But I have not understood this area of TS and TP values yet. Profitability, spread loss and adjusting to history are mixed here.
The trailing stop "looks like" a "buy and hold", but the trailing take is not. The algorithm thinks differently than a human trader and it can be felt. That's why trailing take is not as popular as trailing stop? It doesn't even have a name. Trailing takeout is a great pseudo-profiteer. And traders swear at trailing stop because it is a pseudo-profiteer. I guess it's all a matter of psychology or history.
But the algorithm does not care about trailing stop or trailing take. It makes no difference.
the algorithm works only because sometimes (during crises or regulation by central banks) the exchange rate stops being chaotic.
The number of trades is really small. It bothers me myself. The number of trades can be increased at most twice by reducing TS or TP to 300. But I have not understood this area of TS and TP values yet. Profitability, spread loss and adjusting to history are mixed here.
The trailing stop "looks like" a "buy and hold", but the trailing take is not. The algorithm thinks differently than a human trader and it can be felt. That's why trailing take is not as popular as trailing stop? It doesn't even have a name. Trailing takeout is a great pseudo-profiteer. And traders swear at trailing stop because it is a pseudo-profiteer. I guess it's all a matter of psychology or history.
But the algorithm does not care about trailing stop or trailing take. It doesn't make any difference.
That's right. A person's non-symmetrical attitude to profit and loss is a feature of psychology, a lot of literature has been written on this topic.
But the machine in particular and statistics in general do not care. And this is the power of autotrading.
Googled it. Only found many copies of the same article. "Separate testing of trading system inputs and outputs"
http://wellforex.ru/publ/forvard_testy/stati_forex/razdelnoe_testirovanie_vkhodov_i_vykhodov_torgovoj_sistemy/7-1-0-76
This article confirms my thesis. For testing, they tear off an input(output) and sew it to a "simple" output(input). And they compare not the tested inputs, but the whole system with tested inputs and simple output. the simple input cannot be evaluated in any way.
Indeed, how can we compare a random entry and a reverse entry (an entry in the opposite direction to the previous trade)? Everything will depend on the output stitched to them.
By the way, if google gives you something more valuable, share the links at once. Google output changes over time and valuable links can simply disappear from the output.
I already remembered: "D. Katz, D. McCormick. Encyclopedia of trading strategies". See the discussion on the topic on foursquare - in this thread.
...
At the algorithm level, it looks like increasing TR by a percentage. But the algorithm was debugged with 1.01 and the TR optimum is so flat that it will still work.