Major Model Improvement?

I noticed today that the standard slow(5,3) stochastic began to decline today while my preferred full(5,5,3) stochastic increased steeply. The existing model forecast predicts a decline in the market starting on Wednesday. But with the slow stoch already declining I would feel uncomfortable about being long. I did some trading simulations on a spreadsheet and found that the slow(5,3) stoch is often too early in determining changes in trend while the full(5,5,3) stoch is too late. Is this why I find it hard to stick to trading the model? Averaging the two stochastics outperforms either of the two stochastics. So now I have applied the forecasting models to this averaged stochastic. All three indicators estimated using the new model say the averaged stoch will fall on Tuesday. To do that the NDX must fall at least 10 points. Therefore, the model's decision is to get short at today's close.