I suggest that rather than introduce additional elements to the decision-making process, each of which would have to be backtested and forwardtested in isolation in order to control the variable, a more efficient and expedient alternative would be to follow the rules more closely.
In this case, you do have an SL break followed by a DL. A short DL but a DL nonetheless. The DL is broken by the 1535 bar. The focus then turns to looking for a retracement for a short, not for a long. Therefore, the retracement that occurred two bars later provides the opportunity for that short.
If the 1534 bar had merely touched the DL before continuing upward, that would have been a good opportunity for a long. But that's not what happened. The next bar instead plunged five points, and the bar thereafter, after trying to make a higher high, instead fell back to "close" lower than its high. This was observable in real time. But even if all this had been misinterpreted or ignored, the long would not have been triggered if the entry had not been so tight.
Keep it simple. Don't try to fix what ain't broke.