Jozefo: the forgotten BQX BTC bot

Beginning of this year I was collaborating with a member on an idea to design a bot to trade Voyager token.

Around February BQXBTC on Binance was consistently outperforming bitcoin. Actually in the crypto screener on trading view was displaying as the only token doing that while bitcoin was constantly pumping.

At that time they were moving in opposite directions making them perfect to act as a hedge fund.

So we created a bot for such a special occasion called Jozefo. I ran it for a while until I noticed that the observed behaviour was no longer valid.

That was in February. Since then one of our members kept it running without stop completes one deal every 7 to 10 days but when it does is like pay day.

As it uses such a large max deviation ~40% it fills the bags with discounted BQX and then uses the oscillating nature of this token to turn all that into profit in BTC.

This is the link to the actual Jozefo bot that was used non stop.

It's using around 900 dollars. Some deals are closing with 10% profit some with 5%, it all depends on how many SO manage to get filled and how long the cycle of BQX is taking.

In the TV chart above we can see that there's a volatility that takes around 15 to 30 days to cycle. The drops in price are between 40 to 50% right before there's a consecutive big pump which can easily allow a 10% TP. In some cases the pump after the dump is of 70% to 100%, so 10% is pretty conservative.

The bot is originally entering at RSI7 5m < 65 which is almost the same as using open deal ASAP due to the expected frequency of this condition.

Backtest improvements

Of course, I was itchy to see how the backtest script could make this even better. 

Back in February the idea of having a backtest was just a trading dream on 3 commas and the only pine scripts claiming to do this were very poor quality leading to bad decisions (and most out there still are!).

I limit the tests to only BB and RSI. The goal is to reduce that max deviation from 40% to around 20% so we can maximise our return and hopefully, still close more deals: A nice challenge.

DCA settings

After a few minutes I decided to go with two alternative settings based on the picture below:

BOT A: 2 SOs

BOT B: 3 SOs - this bot is the same as the picture but only adding an extra SO 

Deal start conditions

All results are for the period from 1/1/2021 until today.

So I tried loads of combinations but the ones that stood out were:

RSI7 1h < 30

• BOT A making 224% / 15 days longest deal
• BOT B making 130% / 15 days longest deal (extra SO didn't help!)

BB 4h 2std 

• BOT A making 133% / 12 days longest deal
• BOT B making 74% / 12 days longest deal (extra SO didn't help!)

Keltner Channels 45m (external indicator)

• BOT B making 148% / 12 days longest deal (extra SO didn't help!)

There are many more risky settings but they will get some deals to stay open for longer than 2 weeks. This bot should be seen as a set and forget not as a "let's check on my phone every 5 mins" :)