Constructive ideas for TA use in Gunbot

[Gunthar]

Please if you have a REAL and tested strategy to implement in Gunbot using TA lib, post it here, we will discuss togheter and implement it.

Example: "SMA5 crossing SMA10 and SMA20 as buy/sell signals IF Stochastic RSI is between 30/70 and changes direction"


Thanks!
~Gun

[PHaRTnONu]

t want to give my stratigy away but i use conditional trading of 10 talib points (meet 9 out of 10 and meed 1 requirement also then issue a buy. sells are a bit easyer. 4 out of 5.) problem is if a rally happens (upward trend over 2% (also able to be flag by a tick counter of a talib (need to compare prior ticks to current ones) i know before a big rally 98% acuracy, problem is i have no tool capible of monitrior every pair at the same time so now i just eyeball it same sort of before a big dump i can normaly call a 10% sell of a minute or two before it happens again need talibs of around 7 with a conditional modifier..... anyways if i could have conditional set's that would be great LOL granted i dont even have a copy i have been exploring buying it because its realy teh only tool that could monitor all pairs on a exhange the way i want to....

[Gunthar]

~snip

by "talibs" you meant? Please explain exactly which one of the over 200 talib indicators you are using and how.

[PHaRTnONu]

all the talibs need to be configuable for customer settings

BBands
Balance of POWER
CCI
EMA *** multiple entances to depending on if selling.buying.long(rally).short(panicsellers)
macd hist signal and macd
Parabolic SAR
RSI
STOCH fast and slow
STOCHRSI fast and slow
Supertrend BF
**EDIT
SMA (again multiple config's)

all the data points are put in to a math formula for wighted descition's if the disction scale meet the goal it takes the action, the same stats will allso tell me when a rally or sell of is going to allow me to buy in at the botem and sell (near) the top or at the top... most of the time im in the ticker 1 minute behind the top....

[DaleyEdster]

Sounds very interesting! Care to share full details either on here on via PM so that I can test it out? Second opinion and all that. ;-)

[Gunthar]

all the talibs need to be configuable for customer settings

BBands
Balance of POWER
CCI
EMA *** multiple entances to depending on if selling.buying.long(rally).short(panicsellers)
macd hist signal and macd
Parabolic SAR
RSI
STOCH fast and slow
STOCHRSI fast and slow
Supertrend BF
**EDIT
SMA (again multiple config's)

all the data points are put in to a math formula for wighted descition's if the disction scale meet the goal it takes the action, the same stats will allso tell me when a rally or sell of is going to allow me to buy in at the botem and sell (near) the top or at the top... most of the time im in the ticker 1 minute behind the top....

aaaaaand ok, that would be cool and easy to do: a set of configurable variables in config.js and logs. But what we need is to give the bot entry points like: IF sma50 is > sma5 && RSI <= 70 && RSI >= 30 then setEntryPoint(sell)

dunno if you got the para-code

[PHaRTnONu]

right i have attempted coding it in coffee script while i can write the logic steps out in a understanable manner with my english (for the most part) i cant write code to save my LIFE rofl

[PHaRTnONu]

i would love to code it but i cant afford the bot LOL would be more then willing to give the whole stratigy but i would want to be able to use my key's i know that sounds dickish but its a sound scalping method to scalp in multiple pairs constantly jumping to the next 2 to 5% wave all day long i just ahve no tool to onitor allt he pairs for the signal's  more then willing to show screen shots of the method in action too

[jeffatkins]

Thanks for all the effort Gunthar... Is there a reason why Selling on Gain doesn't place a LIMIT SELL order immediately once the BUY order is filled? Seems like the timing delay could cause a missed opportunity... I'm new to the trading world so forgive my ignorance if this isn't a good idea.

[PHaRTnONu]

here is a tip guntar to add tot eh bots BB bounce. if Middle band is < then prior middle band (.02 tolrence (is if current is 55.89 proir 55.885)) do not trade  a good away to avoid automatic down trends....

[Gunthar]

here is a tip guntar to add tot eh bots BB bounce. if Middle band is < then prior middle band (.02 tolrence (is if current is 55.89 proir 55.885)) do not trade  a good away to avoid automatic down trends....

ok backtesting this

[allanster]

Whenever BB are constricted and then open, BB theory does NOT work and does the opposite of what you expect... https://youtu.be/gKUOVAwxxBY?t=306

Combine MACD 4C with BB to make BB more accurate... https://youtu.be/f2fSnuJ3Fbw?t=349

---

Combine WPR and RSI, only when both indicators signal (leave their zone), trigger action.

---

Use TA's to bring back something like currencies.js 2.0, grades coins by volatility and percentage swings, something that looks at volume and walls to identify pumps with option to auto activate pair with predefined setting, and perhaps a weather meter showing market condition, etc...

[vosechu]

I absolutely love this article about RSI and doing sales in the middle to cover half the risk. http://www.investopedia.com/articles/forex/08/rsi-rollercoaster.asp

For me I use MACD and Stochastic RSI primarily, but I double check them against EMA(,EMA(16), and BB(4). http://www.investopedia.com/articles/trading/08/macd-stochastic-double-cross.asp

[DaleyEdster]

here is a tip guntar to add tot eh bots BB bounce. if Middle band is < then prior middle band (.02 tolrence (is if current is 55.89 proir 55.885)) do not trade  a good away to avoid automatic down trends....

ok backtesting this

How did the backtesting go?

[PHaRTnONu]

Whenever BB are constricted and then open, BB theory does NOT work and does the opposite of what you expect... https://youtu.be/gKUOVAwxxBY?t=306

this is slightly untrue. there are indicators to the bb squeeze exit that is over 70% accurate.
but its almost 90% true the first indication outside the BB after a squeeze of les then 1%  is going to be the direction of hte new trend.

Combine MACD 4C with BB to make BB more accurate... https://youtu.be/f2fSnuJ3Fbw?t=349

this is a new idea to me and thank you for hte coment it looks fruitful i will look more into this my self...

Combine WPR and RSI, only when both indicators signal (leave their zone), trigger action.

have not tried this either thank you

Use TA's to bring back something like currencies.js 2.0, grades coins by volatility and percentage swings, something that looks at volume and walls to identify pumps with option to auto activate pair with predefined setting, and perhaps a weather meter showing market condition, etc...

but what ta's to use as a grading average? macd crosses with vol and chop? with volatility oscillator variance?

[allanster]

I'm not sure how to best handle this (which TA's), but it seems logical that in order to maximize profits we would want to select pairs with sufficient volume, highest volatility, and highest percentage swings of that volatility. These should be able to be determined with various TA. Something built in along the lines of what Chirag Mehta has done here... http://botgen.io/#/

This request is probably outside the scope of Gunthar's request for TA suggestions but what is needed is a way to identify pre pump conditions, steady or slightly increasing volumes over time, walls being put up, etc. If Gunbot could identify these "storms" before they hit, then an option to auto enable that pair could trigger with predefined pump settings.

The last "market condition" would be difficult I think to pull off, if Gunbot could see SMA 50 on chart of Global Market Cap then sentiment could be determined by if SMA 50 is underneath then trend is up and market is good, if SMA 50 is above then trend is down and market is bad. This is eye candy but would be a nice overall signal to have.

[PHaRTnONu]

I absolutely love this article about RSI and doing sales in the middle to cover half the risk. http://www.investopedia.com/articles/forex/08/rsi-rollercoaster.asp

For me I use MACD and Stochastic RSI primarily, but I double check them against EMA(,EMA(16), and BB(4). http://www.investopedia.com/articles/trading/08/macd-stochastic-double-cross.asp

a better tool is HMA just fyi rsi wont get you a sideways market BB bounce rsi can be misleading ina qucik 1% jumping in sideways action look back to the last month of monero youll see

[avi8r]

Please if you have a REAL and tested strategy to implement in Gunbot using TA lib, post it here, we will discuss togheter and implement it.

Example: "SMA5 crossing SMA10 and SMA20 as buy/sell signals IF Stochastic RSI is between 30/70 and changes direction"


Thanks!
~Gun

Here's something I'd like to see added into Gunbot in the not too far distant future... I like to trade Bollinger Bands, but ofttimes, a currency will not just ping pong, but does a spike up. I'd like some sort of Boolean parameter that allows Gunbot to buy and sell a certain currency as many times as it wants as long as it remains in "X" range, but once it does a buy at the current low, and then spikes up beyond X%, sell at the top once it crosses the upper Bollinger Band, and then disable the pair for any future trading unless and until I authorize it. That's the kind of manual trading I'm having to do and it just seems like Gunbot should have enough automated smarts to do that for me instead of leaving me holding yet another bag.

[allanster]

I would like to see something similar, just a simple entry and exit with option to take last ask. I'd most like to see dynamic trailing stop.

[dmains]

Does anyone know what time parameters poloniex uses for MACD? It seems EMA12-EMA26 is most common.

Looking at the charts on poloniex with 15 minute and 30 minute candlesticks, it looks like the local minima and local maxima of the MACD are more reliable BUY/SELL signals than crossovers.

Not sure if this is available in TAlib, but ideally:

IF Trend(MACD) goes from negative to positive, BUY.

(Could possibly be coded as IF 0>Trend(MACD)>x, where x is some empirically tested value. However, this would buy in some situations besides real 0 crossovers.)

IF Trend(MACD) goes from positive to negative & GAIN > y, SELL.

This would let us ride the pumps and MOST dumps very nicely

In the aggressive dump pictured below, we would probably get stuck with some bags, but that's why we have DOUBLE UP

XMR/BTC dump https://imgur.com/XPwV8OH