I would like to correct myself.

The strategy is probably not overfitted. I made Walk Forward Optimization and it looks that the strategy is quite solid. The parameter is needed, because otherwise there are too many entry signals. It really improves the result, but I don't think it is overfitting.

The problem is, that there are some loosing years. The strategy is not good enough for trading, but it is a nice experiment anyway. It leads to other ideas for using VIX for trading.