A Sucky Crypto Trading Robot

Felt cute, might play around with the Discord SDKs later <:CS_PikaCute:745208927519768638>image

1 Like

I’m here for this

I don’t need your shit sir, I was half in the bag when I made this post

Very small initial balance on the robot, but I finally had a chance to pull out the holistic performance of the bot across the targeted currency pairs that it’s been trading. This is a 14 day snapshot in which it executed 773 trades ($6713 of trades with a $20 balance). All in all, it lost $13 but I extracted some really good information! I actually have consumed this information, made some changes, and reloaded the $13 it lost today. So far it’s up to $22.

The strategy that is informing the robot is pretty solid in at least one direction. With enough trades we can rule out certain types of trades algorithmically (I’m looking at you, Long DOGE). There are some that simply aren’t working - AVAX and MATIC are simply not working at all and have been promptly disabled entirely. There are also some that are working exceptionally well - ETH and ORDI. Some I witnessed consistently failing before doing this analysis and have already removed - JTO and ARB.

This type of incremental analysis allows me to tweak what types of entries I want to take when the algorithm is fired what it will choose to do with that information. I have stored this information in a DynamoDB which is being referenced by the Lambda function that’s invoked by the webhook.

Even though it hasn’t been successful by the strictest definition, I’m still fairly confident that the tweaks that have been made along the way are making big impacts to this dumb little crypto robot.image

1 Like

Also fun fact - every trade this thing makes is logged in my own personal discord which allows me to monitor each of these trades as they happen. As I quickly learned, I had to shut the notifications off because even though 773 in 14 days may not seem like alot holy god was it annoying to get those notifications constantlyimage

Today my algorithm did not correctly predict war but it did a pretty good job at recovering.

At its peak it was up 15% on the day. It then fell to -25%. It’s ending the day +5%.image

2 Likes

Oh yes I should give you guys a life updated:

The robot did not fair well throughout the Iran conflict and basically blew itself out playing the reversal in every possible direction. I temporarily took it offline while I work through some more backtesting on various approaches to volatility (namely, volatility recognition and standing the fuck down while it’s going on).

I’ll probably throw a few more dollars at it within the next week or so and get it up and running again while I still have my AWS credits

2 Likes