
Data Quality • 379
markets failed to respond to the triggering events of what we now know
was a jolting disaster in the making.
is discussion appears to lead us away from the topic of data sets, but
it is in fact directly relevant. If we detect no identiable precedent in the
data for an event, we will not catch it. is is not because our statistical
tools are not good enough, and it is probably not because we were care-
less in collecting data. It is endishly dicult to systematically collect the
needed data when the relevant precursors to the event to be forecasted are
not even known. Without the relevant data, we will not be able to identif ...