Quiet, Careful or Keen?

A card issuer tests three fraud models on the same 8,000 transactions, of which 40 are fraud.

As the Choosing the Right Metric lesson warned, accuracy is easy to fool when one class is rare. So the team ranks the models by the Matthews correlation coefficient (MCC), a single score that uses all four cells of the confusion matrix, with fraud as the Positive class:

$$\text{MCC} = \frac{TP \cdot TN - FP \cdot FN}{\sqrt{(TP + FP)(TP + FN)(TN + FP)(TN + FN)}}$$

MCC runs from $-1$ (every call wrong) through $0$ (the predictions tell you nothing about the truth) to $+1$ (every call right). If any of the four totals under the square root is $0$, MCC is defined as $0$.

What is the MCC of the model that ranks first? Round your answer to 3 decimal places.