Cryptocurrency agent - testing rules

On this page I will share some of the Rules I have been testing to 'drive' my cryptocurrency trading agent and the results I have been obtaining.

Please note that these are just meant to be tests, I have been running them on an account with a small amount of money. There are by no means meant to guarantee any financial results.

First set: (workflow FastAgent1)

BTC - EUR, check every hour. If the price goes up (down) over 750 EUR since the last check, then sell (buy) 65% of current assets.

This set does not perform well. I believe this may be because the workflow compares the current price against the price one hour ago, regardless of if an order was placed or not one hour ago.

Second set: (workflow FastAgent2)

BTC - EUR, check every hour. If the price goes up (down) over750 EUR compared to the price paid in the last sell (buy) order, then sell (buy) 65% of current assets.

This set performs better. About 7% profit in 1 month. An issue is that it launches orders quite frequently - usually several times a day. This may be in part of the current geopolitical (shit) situation, which makes markets fluctuate a lot. Also I recently received a communication from Revolut that in the future they will be charging more commission on trades. This leads me to the next set.

Third set: (workflow FastAgent3)

BTC - EUR, check every 30 minutes.  If the price goes up (down) over 2000 EUR compared to the price paid in the last sell (buy), then sell (buy) 65% of current assets.

Notice these are small tweaks Vs. the previous set. Also note that the 65% would be probably too high for accounts with a higher amount of assets in them. This has been running for about one week. No results yet, because no orders have yet been triggered.

Below you can see part of the log output of this last workflow: the last 4 lines are showing a small pattern: BTC is losing value, so this would trigger Buy orders, but the orders are not actually placed by the workflow, because the price difference - decline from the last time we sold is still not over 2000€ (258€, 484€, 656€, 702€). If the trend continues, eventually a Buy order will be placed.

This last rule set is (again) not performing well. In stead of triggering too many orders like the previous one, this one triggered zero orders in about two weeks. I think this is because a combination of two factors: the thresholds are now too high  - 2000€ price difference and BTC price fluctuating less. This shows that ideally the rules should change according to the market and geopolitical situation.

In the next version of the workflow I will try a more 'hold on' type of approach: suspend buying for the moment and only sell if the current price goes above the maximum I ever paid when buying plus 2000€. As always, I will have to wait a few weeks at least for the results.

It is becoming clear that, more important than the actual workflow, is getting the rules 'right'