Suppose you calculate a Theil's U value for some forecasting model $($call it "Model $1"),$ and you find it to be $0\mathrm{.}67.$ What does this definitely tell you about the accuracy of your forecasting model?

The average percentage error of Model $1$ is $67\%.$

Model $1$ has a larger RMSE than the naive $1$ model.

Model $1$ has a smaller RMSE than the naive $1$ model.

The cost of implementing Model $1$ as a forecasting tool will be less than any other nfmimel.