First of all I would like to remark that the way Gunbot calculates PNL would have shown much better results than Binance PNL does as Gunbot only takes into account the closed trades. Ok, the thing is, are the results good? Are they bad? Are what I was expecting?
In my opinion, obviously we need longer periods to make any conclusion, but for me it’s a very good result despite Binance shows +10$ profit. Why do I think it? Because I have been trading with coins whose price had a huge drop and despite of that, it managed to get a little profit and get a much better entry price than the last one I got when last cycle started.
If you take a look to USDT-BTC chart, you can see the last cycle started at around 58k$ and the new cycle started around 55k$. And we know that during the trip from 55k to 58k it will keep doing profitable trades, don’t we?
In my opinion, this is something like I described for scenario 3:
When BTC price goes down slowly, it gets a better average bought price and it gets rid of the bag with profit when the price bounces a bit, and at the same time starts a new cycle with a better starting point than the previous one. Note that the price never goes straight down or straight up, I mean, regardless the trend is upwards or downwards, the price is always bouncing up and down.
We can see a similar behavior for ETH and BNB pairs: