Friends, I want to share my experience with the new feature. Describe a strategy, and AI will test it in practice—or might even launch it.
Here’s what I got:
The first paper trade has been executed. The strategy went short on BTWUSDT: 861 coins at an entry price of around 1.1584. The price fell to 1.0723 — a 7.43% drop from the entry price, which is a gain for a short position. On paper, the account is up about 73 USDT (it started at 10,000 and is now at 10,073), or +0.73%.
And now, honestly, about what I don’t like. Between 16:27 and 16:33 Singapore time, the exchange started responding with “too many requests”—and the strategy couldn’t close the position: more than twenty consecutive rejections of the closing order, plus a log entry saying “order placed, but confirmation not received; manual reconciliation required.”
Bottom line: the strategy dashboard still shows an open short position of 861 BTW, while in its latest cycles the strategy itself is already reading the position as zero. Its picture of the world and reality have diverged. So that 73 USDT profit is still “on paper” for now, not realized profit. 😔
What this means, in plain terms: this is exactly the rough edge I told you about—“a partial close or UNKNOWN status isn’t brought down to zero.” It showed up with the very first real price movement. And thank goodness it happened on paper, not with your money—that’s exactly why a test drive is needed. 🙏
We can’t fix this with a code change: version r1 is already locked in, and any change means a new version with another audit and test run.
What should we do: stop it and look into the reconciliation issue, or let it keep running for now while we just monitor it?
Honestly, I really enjoyed our conversation—it was completely straightforward, with no hidden surprises. Of course, I didn’t run anything with real money.
But I’m really envious of all the IT folks who can just write a program like this.
If anyone’s interested in which strategy I was talking about, just ask ...💞