When to re-optimize or retire a system?

Even if the backtest is done perfectly with good optimization and walk forward results, it’s all about the past and the future is unknown.

After you have been trading with the same system for a while and the performance decline suddenly or gradually in an unexpected fashion, and you start to wonder if it still works. Are there any criteria or certain rules that determine the system is no longer working and you either need to re-optimize it or retire it?

Thank you .

This subject is covered in Howard Bandy's books AmiBroker Books - #2 by Tomasz