Manage through fear and intimidation, love, role modeling, ... or if you can't measure something, it might be the very most aspect of the problem?
Take the valuation and risk analytics of a moderately complex financial instrument, say, a callable reverse floater type with special schedules and conditions, like range accrual, snowball, .... Yes, if models get too complicated, there is a danger of spurious accuracy , which may lead to loss of criticism.
But if you valuate the reverse floater across different models, like short rate and market models, and get significantly different prices and risk spectra, especially with market data in different times, you should look into the construction of the instrument itself (all provided, that you are sure that the model-solvers are adequate, accurate and robust)?
Spreadsheet or data base, how can you come up with a trading and risk strategy, if you do not have this "soft measurement"?
At the other hand, I increasingly read questions on the-best-trading patterns. What will happen, if we all apply the best-trading-patterns?
Yes, data and equations shall not become think-stoppers, but .......