Bitcoin Forum

Hey guys. We working on AI trading models, and have tested them for a while. We are still in R&D stage, so we just ask for feedback: do you think those live ttading results are any good (no leverage)?
https://docs.google.com/spreadsheets/d/e/2PACX-1vSoMdoqD_akKfxdpkjNWZ5mQ9ZDjUIz7vo6EvVclH6ElV0NQ2FY6qFTWLv5k1xvJN_xAjPJX8VoM8w9/pubhtml?gid=1245945501&single=true

What is the percentage accuracy on the average for the trades you entered. I checked through the link and was expecting to see such information and couldn't find that. Some figures are also different from some, why is it that way ? You need to let people who are interested on the accuracy of making profit and also its failures (because failure is also possible).

I will try this your bot with demo account and see how it works if you can give a free copy. I have used tradesanta and 3commas around January to may this year when the market was bullish and I  really made good profit but since end of May it has been on lose spread! Hope your both perform positive during the bullish and bearish market?

Most of those models are long only, so they have little chances to make profits in usd on the falling market. We will try leverage, thank you.

I have only trading logs for live models so I need to do the math to derive those numbers. What metrics are you interested in? Profit trades ratio, profit factor? I can calculate them after a weekend  if you want, please pick a model.

Good to know you want to try it out! Making money on bullish market is quite easy,  bearish is more tricky. We are still on testing stage, but we can connect test Binance account if you want to. Just pick a model from the list you like.

The data is only promising in terms of BTC "volume" but not the profits I guess.

Also may I ask what kind of factors does your bot use while predicting the trades? Moreover, if it's AI (Artificial Intelligence) then does it have the capability to understand the market by looking at the previous data, current indirect/direct factors affecting it? Because AI gives it really broad spectrum and it's mechanism is entirely different I guess.

It would be grateful if you can show us live working of the bot. When it's in action one can suggest more things about it.

I feel the results are good but you should still so a little more test run for accuracy sake but from the statistics I think your test run is ok for now. But you should also try other pairs not just BTC alone so we can conclude if it works perfectly for all market conditions

Thanks for reply!

>>  The data is only promising in terms of BTC "volume" but not the profits I guess.
It works like that. When market goes down BTC profits grow, and once it goes up models get some profits in USD as well. Those results mostly are for bearish period, "market_price_change_%" shows exactly how much BTC price went down since bot started trading.
 
>> Also may I ask what kind of factors does your bot use while predicting the trades? Moreover, if it's AI (Artificial Intelligence) then does it have the capability to understand the market by looking at the previous data, current indirect/direct factors affecting it? Because AI gives it really broad spectrum and it's mechanism is entirely different I guess.
We currently use candles since 2013 with some trading indicators and then let neural network (AI) find some patterns.

>> It would be grateful if you can show us live working of the bot. When it's in action one can suggest more things about it.
Those results are from live trading of the bots. Its not backtest. Or you mean getting live signals?

Have a nice day!

Hi,

I think this is quite difficult to get a good result with AI.

There's 2 ways to do this

1) Using Deep learning to figure out what's going on next

If you are using DL, you will need to write another scripts to control the trading logic.
And the addon script could be super complicated since you cannot know whether the model is giving you right answer.

Another thing is your the time scale of your training data. And what's the future you are asking the model to predict.
If you are going to feed ultra short time data in your training data. Then your predicting result couldn't be too far from current.

I am doubting this is useful for normal trader, unless you are trading with high volume.

And if you are feeding training model with long-time data, then you model may lack sufficient to get a promising result.


2) Using reinforcement learning to train a bot.

If you are using RL, there's a problem of learning spaces.

How can you let the bot learn if there's isn't such a history?

For example, if your bot is training with data till year 2019.

It won't be able to understand why the market can goes to 65k.

And depends on different learning model (or agent in RL)

it will do random things to verify itself.

Another thing is that, your RL (or DRL) learning process will cost you a lot of dollars.

The computing power will cost a lot more whenever you expand your environment.

And i think it is a must for you to contain the PNL status for the bot (otherwise it doesn't know if it's doing right)
Plus OHCLV, you got 6 dimensions. And i have already tried with this. There's no promising result i can say.

In conclusion , Please do it for fun, rather for profit....