AI is taking over many jobs, starting from cashiers all the way to doctors. I wanted to see if AI could replace traders by running a small (and maybe a costly) experiment.
Rules are simple. For the next week, I’ll long/short Bitcoin based on an AI twitter bot‘s forecasts. Every position will be opened at 00:00 UTC and closed at 23:59 UTC.
Our start balance is $1000. And we won’t use leverage in our trades.
Day 1/7: Great start?
The bot’s forecast was barely bearish. We opened a short position at $29.5K (when the forecast was tweeted) and closed it at $29K (at the end of the day). We ended up with a profit of $17. This brings our final balance to $1017. Probably luck. We will see by the end of this experiment.
Day 2/7: Sometime it happens
This time I couldn’t really tell if the bot was bullish or bearish, so I didn’t open any positions that day.
Day 3/7: Two wins in a row?
Here, the forecast was clearly bullish. Time to long. We opened a long position at $28.6K (when the forecast was tweeted) and closed it at $29K (at the end of the day). We ended up with a profit of $14. This brings our final balance to $1031. Two wins in a row baby!
Day 4/7: THREE wins in a row?
This is our luckiest day yet (I really hope this continues). We longed at $29K and closed the position at $30.3K, ending up with a profit of $46. This brings our final balance to $1077.
Day 5/7: I knew it’d happen
Here, the bot seemed neutral and bearish at the same time, so short it is. We opened a short position at $29.5K and closed it at $31.8K (at the end of the day). We ended up with a loss of $84. We’ve successfully wiped all of our previous profits, bringing our final balance to $993. Yay!
Here, the bot seemed neutral, and its confidence was slightly high, so we didn’t end up opening any positions that day.
Day 7/7: Comeback or Horrible Fail?
Ah yes. Nothing goes up forever. The bot here was barely bullish, so I ended up longing. Big mistake. We longed at $31.8K (when the forecast was tweeted) and closed it at $29.7K (at the end of the day). We ended up with a loss of $66.
We won three trades and lost two. Our win rate seems to be 60%. We won a total of $77 and lost $150. Finishing up with a total loss of $73.
Of course, five trades aren’t enough to evaluate any trading strategy. The bot seemed to do well initially, but then the market started crashing.
This small experiment shows that the crypto market is genuinely unpredictable. Not even by a shitty AI Twitter bot that made me lose $73, which probably flips a coin to generate random predictions.