Bitcoin Forum

Bitcoin => Project Development => Topic started by: mrkfdr on April 19, 2020, 08:46:22 PM



Title: My new trading bot is up&running and i came up with an interesting question
Post by: mrkfdr on April 19, 2020, 08:46:22 PM
Just completed to develop my first trading bot and put it into a test run. ;D
The trading model is far from being perfect, but so far (the last couple of days), does the job , didn't loose any money and earned a few bucks.

During the development a came up with a question, to use one bot or several bots.

If using one bot
The architecture is to fill up the code with endless number of restrictions and fallback parameters to avoid falls and gain maximum profit.
doing this the maintenance is hell, slows down reaction time, and there is always a little tweak to add that will probably brake the whole logic...
but
you have one bot for the job. and it will make maximum profit. (if there is such a thing...)

Or, use few bots
based on different methodologies like, take lower risks or higher risks, of when to buy or sell based on different models.
so some will perform better then the other
but,
compare to the first one it's hard to tell how it will perform..

Any toughs on this ?
thnx
/\/\rk





Title: Re: My new trading bot is up&running and i came up with an interesting question
Post by: pooya87 on April 20, 2020, 04:48:16 AM
If using one bot
The architecture is to fill up the code with endless number of restrictions and fallback parameters to avoid falls and gain maximum profit.
doing this the maintenance is hell, slows down reaction time, and there is always a little tweak to add that will probably brake the whole logic...

why not split all the different parts into smaller ones and then have then run separately but still inside one bot. for example at startup you could implement a selection process that chooses which options to run and which not to.

additionally if you feel maintenance of your code is hell then it means your code is bad, try using IoC principle in whatever language you use and have loosely coupled parts that don'y depend on each other this much and only communicate with each other in a "loose" way.


Title: Re: My new trading bot is up&running and i came up with an interesting question
Post by: mrkfdr on April 20, 2020, 08:27:35 AM
If using one bot
The architecture is to fill up the code with endless number of restrictions and fallback parameters to avoid falls and gain maximum profit.
doing this the maintenance is hell, slows down reaction time, and there is always a little tweak to add that will probably brake the whole logic...

why not split all the different parts into smaller ones and then have then run separately but still inside one bot. for example at startup you could implement a selection process that chooses which options to run and which not to.

additionally if you feel maintenance of your code is hell then it means your code is bad, try using IoC principle in whatever language you use and have loosely coupled parts that don'y depend on each other this much and only communicate with each other in a "loose" way.

Architecture is fine, been writing code and mentoring for a long time.

The thoughts were about one bot doing all VS several bots doing different tasks and how the sum of each end up.


Title: Re: My new trading bot is up&running and i came up with an interesting question
Post by: buwaytress on April 20, 2020, 02:42:07 PM
I always assumed people used 1 bot and then tested out multiple scripts -- the only reason you'd use more than 1 bot was to see which bot was capable of the most options, which executed strategies the best, which connected to trading platforms best.

Not a veteran bot user, but when I tried in the past, each was perfectly capable of running many strategies concurrently, each keeping its own records, easy to compare side by side in the same bot. And this was years ago. Pretty sure things have advanced by now.


Title: Re: My new trading bot is up&running and i came up with an interesting question
Post by: Lakai01 on April 21, 2020, 07:28:15 PM
Not a veteran bot user, but when I tried in the past, each was perfectly capable of running many strategies concurrently, each keeping its own records, easy to compare side by side in the same bot. And this was years ago. Pretty sure things have advanced by now.

At the height of my trading career mid to late 2017 (well, in a bull market, everyone can claim to be a trader ... ;) ) I set up an open source bot based on Gecko. This bot already does exactly what you describe and can, for example, carry out several different runs with different bot strategies concurrently. I liked the paper trading/backtracking feature best at that time. You can try out your trading strategies in Gecko based on historical data and thus determine whether you would have made a profit or not.

Maybe that also would be an opportunity for you OP to determine which strategy would be the most suitable here: Download an open source trading bot and use paper trades/backtracking to test which approach would have brought you the most profit (or, as in my case, the least loss).

You can find information about Gecko here (https://gekko.wizb.it/). The project is no longer being developed, but that should not be a problem for this application.


Title: Re: My new trading bot is up&running and i came up with an interesting question
Post by: buwaytress on April 22, 2020, 11:50:08 AM
Not a veteran bot user, but when I tried in the past, each was perfectly capable of running many strategies concurrently, each keeping its own records, easy to compare side by side in the same bot. And this was years ago. Pretty sure things have advanced by now.

At the height of my trading career mid to late 2017 (well, in a bull market, everyone can claim to be a trader ... ;) ) I set up an open source bot based on Gecko. This bot already does exactly what you describe and can, for example, carry out several different runs with different bot strategies concurrently. I liked the paper trading/backtracking feature best at that time. You can try out your trading strategies in Gecko based on historical data and thus determine whether you would have made a profit or not.

Maybe that also would be an opportunity for you OP to determine which strategy would be the most suitable here: Download an open source trading bot and use paper trades/backtracking to test which approach would have brought you the most profit (or, as in my case, the least loss).

You can find information about Gecko here (https://gekko.wizb.it/). The project is no longer being developed, but that should not be a problem for this application.

Exactly. And I've stayed well out of trading, let alone using bots for years now, but can't help but see during 2017/18 even Gunbot tout these functionalities. Think more recently HaasOnline got a major upgrade that lets you do all those things too -- historical backtracking sounds pretty cool, but strategies that worked well in past don't tend to in current/future markets no?

And yeah, everyone can claim to be so smart in a bull market, I certainly recall all the proud posts of winning strategies being peddled to this day;)