[SHARE] Simple RSI BULL/BEAR strategy
#91
(02-13-2018, 05:05 PM)tommiehansen Wrote: Unfortunately i don't know. The BTC-USD pair seem to be especially hard to get good results out of (maybe because of other bots?).
I haven't had the time to properly analyze why this is yet.

And please don't use any of these strategies with any larger sums of money. Gekko is still in beta after all. Smile

thanks for your quick answer

in the hope of success Smile

only 50$ Smile
  Reply
#92
(02-13-2018, 09:59 AM)tommiehansen Wrote: Frankly i find TradingView and pinescript to be a bit buggy so maybe it's me though. Smile

Coinigy allso seems a bit buggy but still very nice.
(Candlesize can be choose in minutes)

ONLY for the SMA, the result looks a bit weird.
Coinigy i can choose the SMA being sourced from the windows left side or right or none.
But the visual result is allways not as expected and the ammount of History pretty small.
so this Coinigy averaging becomes usless to look at.
Seems the  SMA_long is above and the short would never cross.

Can you recomend a way to look at the SMAs

By the way i found your RSI_Bull_Bear_ADX.
Thanks alot again!
everyone can copie/past some code but thats far away from a proper OPTIMIZED highend code.
Words are worthless so if i come up ill tip you and mike off, thats for shure

Just found your STRATGIES at
https://github.com/xFFFFF/Gekko-Strategies

luckly in the Strat it says c ThommieHansen because on the Github page it says nothing about the Author.
..So i thought xFFFFF would be the Author
  Reply
#93
(02-13-2018, 09:59 AM)tommiehansen Wrote: Cool, but doesn't seem to resemble the backtest results within Gekko?

Frankly i find TradingView and pinescript to be a bit buggy so maybe it's me though. Smile

If the strategy tester properties are setup right it comes near Gekkos result (with small deviations.. because of the candle-tick point of time i believe, since gekko just adds 15min to any point of time and tradingview basically starts at zero).

In the end it's just a nice helper to swiftly play around with strategy parameters (less than a second of processing time) and finally test them with gekko.
  Reply
#94
(02-13-2018, 11:06 PM)0mathew0 Wrote:
(02-13-2018, 09:59 AM)tommiehansen Wrote: Cool, but doesn't seem to resemble the backtest results within Gekko?

Frankly i find TradingView and pinescript to be a bit buggy so maybe it's me though. Smile

If the strategy tester properties are setup right it comes near Gekkos result (with small deviations.. because of the candle-tick point of time i believe, since gekko just adds 15min to any point of time and tradingview basically starts at zero).

In the end it's just a nice helper to swiftly play around with strategy parameters (less than a second of processing time) and finally test them with gekko.

Maybe, i just couldn't get anywhere near the Gekko results. I know TradingView and Gekko differ somewhat on the candles though.

And yes -- a nicer way. I usually use TradingView to get a sense of things on a larger scale. Unfortunately Gekko is very, very... slow in comparison to TradingView and to just visually get a sense of things.
  Reply
#95
Someone using this with succes and made profit?
I'm new to it and just did the complete setup and read this whole tread.

It looks promising, but if I went to the following Github https://github.com/xFFFFF/Gekko-Strategies and you open some strategies (including the one from Tommie) it almost always made loss instead off profit.

Will look further in this though :-)
  Reply
#96
Hi,

Great work! Thank you for sharing.


I'm having some issues repeating the backtrack results compared to the real world.
I had a strategy running yesterday, it gradually lost money so I stopped it.

When I backtrack that same segment with the same strategy it has way better results.

Server is running on a macbook pro.

Any one who has an idea on this?

Thank you so much!
  Reply
#97
(02-14-2018, 10:07 AM)dingus Wrote: Hi,

Great work! Thank you for sharing.


I'm having some issues repeating the backtrack results compared to the real world.
I had a strategy running yesterday, it gradually lost money so I stopped it.

When I backtrack that same segment with the same strategy it has way better results.

Server is running on a macbook pro.

Any one who has an idea on this?

Thank you so much!

Do better backtests with longer durations (at least 1 year) of data and also make sure to check different periods
to see how the strategy works in different market conditions (basically there are 3 major conditions: bull, bear and sideways).

That the strategy doesn't always profit is also to be expected and should be expected as well.
This is something you'll clearly be able to see if you check the 'Roundtrips' after a backtest.

(02-14-2018, 09:54 AM)Argon Wrote: Someone using this with succes and made profit?
I'm new to it and just did the complete setup and read this whole tread.

It looks promising, but if I went to the following Github https://github.com/xFFFFF/Gekko-Strategies and you open some strategies (including the one from Tommie) it almost always made loss instead off profit.

Will look further in this though :-)

Run your own backtests.
  Reply
#98
Anybody can help with correct importing these strategies (from git) to Gekko?.. thank's
  Reply
#99
Anyone knows what the problem is when installing talib?
Gekko is running fine, but when I want to install Talib I got the following:
Code:
C:\Users\nvw\Desktop\Gekko>npm install talib

> talib@1.0.4 install C:\Users\nvw\Desktop\Gekko\node_modules\talib
> node ./src/lib/build.js && node-gyp configure && node-gyp build

building talib functions...
fs.js:904
 return binding.readdir(pathModule._makeLong(path), options.encoding);
                ^

Error: ENOENT: no such file or directory, scandir 'C:\Program Files (x86)\Reference Assemblies\Microsoft\Framework\.NETFramework'
   at Object.fs.readdirSync (fs.js:904:18)
   at Object.<anonymous> (C:\Users\nvw\Desktop\Gekko\node_modules\talib\src\lib\build.js:22:30)
   at Module._compile (module.js:643:30)
   at Object.Module._extensions..js (module.js:654:10)
   at Module.load (module.js:556:32)
   at tryModuleLoad (module.js:499:12)
   at Function.Module._load (module.js:491:3)
   at Function.Module.runMain (module.js:684:10)
   at startup (bootstrap_node.js:187:16)
   at bootstrap_node.js:608:3
npm ERR! code ELIFECYCLE
npm ERR! errno 1
npm ERR! talib@1.0.4 install: `node ./src/lib/build.js && node-gyp configure && node-gyp build`
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the talib@1.0.4 install script.
npm ERR! This is probably not a problem with npm. There is likely additional logging output above.

npm ERR! A complete log of this run can be found in:
npm ERR!     C:\Users\nvw\AppData\Roaming\npm-cache\_logs\2018-02-14T14_18_20_543Z-debug.log

When seeing this error:
Code:
Error: ENOENT: no such file or directory, scandir 'C:\Program Files (x86)\Reference Assemblies\Microsoft\Framework\.NETFramework'


I installed the latest .net framework supported on my system (Windows 10). The only folders that I got in \microsoft\Framework are v3.0 and v3.5 . No ".NETFramework" folder to see...

Anyone an idea?
Thanks
  Reply
Hi Tommy,

Thank you for answering,

The strategy is tested on data for many months and in different time spans.
The tests that I've done include several roundtrips, my issue is that the roundtrips of the "real world" don't match the ones of the backtest, asif the "real world" buy and sell actions are somewhat delayed, I'm using real tight periods and 2 min candles to check its behaviour quickly.

might this be more of an overall Gekko/API issue maybe?
  Reply


Forum Jump:


Users browsing this thread: