Hello, Krzysztof

You analyze 300 days cycle so the amount of available price history is critical issue. I have downloaded gold price since 1975, this fact can bring some difference.

I've downloaded  price history from eSignal and from TS and got exactly the same cycles and the same projection lines:

The next source of instability is autoselection.

For example WFA chart for 298 cycle looks this way:

not good at all.

138 days cycle looks a little better, but still...

I also have downloaded gold from Yahoo (all three sources show different price) and got three WFA charts for 138 days cycle TS Datafeed, eSignal and Yahoo:

Resume: this model does not look stable and this factor dominates here. Difference in price data brings some uncertainty  but this fact does not destroy stable model.

This is what we have....

Best regards.

Sergey.