Ciao Ale,
non mi riferivo alla WF analisys o Montecarlo, questi non sono metodi di validazione validi,.
metodi validi di capire se quel TS è robusto sono altri.
faccio un esempio, ho postato nella sezione pro-buldier un topic riguardante un TS su Oro, dopo aver partecipato ad una giornata sulla validazione di Luca Giusti(non so, se lo conosci).
comunque in poche parole per validare un TS su Oro ad esempio , lui dopo aver fatto delle prove su un periodo di tempo di almeno 6/7 anni su oro , ha buttato giu un TS con delle variabili, la prima validazione che lui fa è discostarsi di poco da quelle variabili e vedere come si comporta il TS, esempio (se il valore della media mobile , che ti ha dato degli ottimi risultati, è di 10, devi provare a inserire come variabile una medi mobile a 7-8 -9 e 11-12-13 periodi , se il Ts ha delle basi solide non dovrebbe discostarsi troppo come performance da quella ottimale”questa è la prima validazione” quindi quando io vedo i valori di variabili tipo media mobile 9 periodi, atr 11, numeri non “tondi” già siamo in over-fitting.
per la robustezza va testato il TS su mercati simili a quello su cui dovrebbe lavorare il TS (cioè che rispondo anch’esse bene a logiche break-out), lui ad esempio ha provato su Platino ,e Benzina, e davano gli stessi risultati.
grazie per la tua risposta Ale
P.S:
per il TS che ti ho postato hai qualche idea come fare? 🙂
Hello Ale,
I did not refer to WF analisys or Monte Carlo, these are not valid validation methods.
Valid methods to understand if that TS is robust are others.
For example, I have posted a topic in a pro-buldier section about a TS on Gold after a training-day validation with Luca Giusti (I do not know if you know him).
However in a few words to validate a TS on Gold, for example, after testing over a period of at least 6/7 years on gold, he knocked down a TS with variables, the first validation he makes is to depart Little of the variables and see how the TS works, example (if the value of the mobile average, which has given you good results, is 10, you should try to include as a variable a moving medium at 7-8 -9 and 11-12-13 periods, if the Ts has solid basics should not deviate too much from the optimum performance “this is the first validation” so when I see the values of variables average mobile type 9 periods, atr 11, numbers not ” Round “we are already in over-fitting.
For robustness, the TS has to be tested on markets similar to that on which the TS should work (that is, I also say good at logical break-outs), for example, it has tried on Platinum, and gasoline, and they did the same results.
Thanks for your Ale response
P.S:
For the TS I posted to you have any idea how to do it? 🙂
ALEModerator
Master
Ok
I understand what do you mean. I usually use this sistem with daily time frame because if you use rsi, moving average slow and fast etc etc, as Giusti, I’m agee with him.
ALEModerator
Master
In this case you can cheak :
Different Trailing stop, DC, and CP. We have see that if use this strategy with the same parameters could work with similar asset. This could be a confirmation.
On daily time frame Giusti opinion it’s very correct, but in my experience if You lower the time frame, this test becomes more ineffective
I let you know about strategy
Thanks
ALEModerator
Master
Enzo,
could you open a new topic for your strategy to do it?
Thanks
Ale
Enzo, could you open a new topic for your strategy to do it? Thanks Ale
hi Ale, you talking about IDRN4 strategy ?
thanks
@enzo_52
Please speak English in english forums, don’t double post and use appropriate topic for your specific query. Thank you.
excuse me , Nicolas.,
P.S.
could you have a look to my new topic? :-=) dnr4 strategy
Dear All,
LONG Fractal EUR/USD open at 16.00
Regards
Confirmed. Little worrying as I am running it live and Yellen is gonna speak in 3 hours.
ALEModerator
Master
yes It’s danderous.. I hope that volatily touch take profit before stop
PereParticipant
Veteran
I tried to optimize the factor 2, and found that a factor 1.2 gives better results.
In this case, code should be changed as:
fac=1.2
if close[cp] >= highest[round(fac*cp+1)](close) then
LH = 1
else
LH=0
endif
if close[cp] <= lowest[round(fac*cp+1)](close) then
LL= -1
else
LL=0
endif
Could someone confirm this better results?
Nice out with profit before Yellen good result