You are viewing a single comment's thread from:

RE: I think I've built something INCREDIBLE!

in #bitcoin7 years ago

Nice initiative, would be great if you could release the buy and sell set up rules so that others could independanlty back test this trading strategy. Do you have the capability to back test for say the last 2 years? following that, do you ahve the capability to use a program to auto optimize the variables of the strategy in order to determine which are the optimal for this particular market? finally do you have the capability to forward test these variables in order to optimize the variables based on current market conditions? from what i can see you are running this to test over night, this way of testing probably just wont do it. need to run years of back data and several weeks of forward data at a minimum to confirm that this is indeed the holy grail. interested to see where this goes however....

Sort:  

Thank you for the support @starkerz. I appreciate you. I am not going to provide the source to the strategy. It has taken way too long to develope and create. Too much blood, sweat & tears. I hope you can understand.

I do wish I could auto optimize variables. I mean that would be awesome & fun. I would be fearful something like that would lead to curve fitting but I would love an AI program that could auto optimize to find the best take profit and stop levels. Most ideal time to move stop losses. I know with certainty I do not have the most optimal settings. But it is extremely profitable with extremely low risk.

Backtesting on MT4 is not ideal. With the ideal conditions (market data) metatrader 4 back tests are only 90% accurate. Which is way off. Again, back testing could also lead to curve fitting - so I never do it. However, this strategy is traded on the 8 hour time frame with orders updated every hour. This time frame does not come default on the platform making it impossible to back test if I wanted to. You are right, it is going to take some time to confirm that this is the holy grail, but I do think the closest to the holy grail we will ever come across. Thank you for your interest and being a part of this.

:) I back test, optimize and forward test with Ninja Trader. Its pretty damn good once you get the code correct. It certainly gets you past curve fitting. Would be good if you tested this on Ninja....

I was wondering first if you are for real.

With the ideal conditions (market data) metatrader 4 back tests are only 90% accurate. Which is way off.

90% accurate seem pretty darn good but the again I don't understand the context. I once did the Kaggle contest for the Titanic, just learning a tutorial. A model with 80-85% accuracy was very very good and more would just lead to overfitting.

Speaking of the Titanic reminds me of 1 in 60 rule in navigation. For each 1° off over a distance of 60 nautical miles (NM), it will result in 1 NM off course. Now imagine 10° off is 10 nautical miles. In Forex a 90% accuracy is not nearly sufficient for back testing. It is all about the forward test in real time.

OK, can't say I understood a 100% but say 90% ;) So I'll be following along then.

Hahahahahahaha

Coin Marketplace

STEEM 0.25
TRX 0.25
JST 0.039
BTC 94309.43
ETH 3309.61
USDT 1.00
SBD 3.28