"Low side/high side overrides" son correctivos de las decisiones de conceder o no préstamos (decisiones crediticias) utilizando los modelos de puntuación crediticia.

Es decir, si el solicitante del préstamo tiene una puntuación crediticia favorable, de acuerdo a los parámetros de riesgo (crediticio) aceptado por la entidad prestamista, pero esta última decide finalmente no otorgar el préstamo sobre la base de otros criterios, hablamos de "HIGH SIDE OVERRIDES".

En cambio, si la puntuación crediticia del solicitante no es favorable (en función de los parámetros de riesgo asumidos por la entidad prestamista), pero esta última otorga el préstamo sobre la base de otros criterios alternativos, estaríamos hablando de "LOW SIDE OVERRIDES".

"If a consumer has a credit score that is not within the range of risk acceptable to the lender, a lender may examine other characteristics of the borrower, or consider extenuating circumstances that adversely affected the score, and decide to extend credit to the consumer despite the score. This is known as a "low side override." Conversely, if a borrower has a credit score that falls within the range of risk acceptable to the borrower, other factors may lead the lender to nonetheless decline the loan. This is known as a "high side override."

http://www.ftc.gov/os/2000/09/creditscoring.htm

"For automated systems, underwriting exceptions are termed “overrides.” An override is a decision to accept or reject
an applicant when the scorecard says otherwise. High-side overrides and low-side overrides exist. With a high-side
override, the bank rejects an applicant which the scorecard would approve. A high-side override can occur when the
bank considers variables or characteristics which were excluded from the model. With a low-side override, the bank
approves an applicant which the scorecard would reject. An excessive level of overrides, either high-side or low-side,
negates the use of scoring models. If the scoring model properly reflects the bank’s risk parameters, overrides should
be used with considerable caution.
The three types of overrides are policy, informational, and intuitive. Policy overrides occur when management sets up
special rules for some kinds of applications. Informational overrides are based on information not available in the
scorecard or only available in small quantities, like other existing customer accounts. Intuitive overrides are based on
“gut feeling.” Management should have the ability to identify the volume and track the effectiveness of overrides.
Typically, the override rate is calculated by comparing the number of overrides, either high-side or low-side, to the
number of applications in that population. For the high-side override rate, compare the number of applicants over the
cutoff score who were denied to the number of applicants over the cutoff score. For the low-side override rate, compare
the number of applicants below the cutoff score who were accepted to the number of applications below the cutoff score.
A large number of overrides is equivalent to having no cutoff at all and jeopardizes the bank’s ability to measure the
predictability of the scoring system. When override rates are more than what the scoring manual provided by the
vendor or developed by the bank deems reasonable, the examiner should evaluate the override policy and discuss any
concerns with management."

http://www.occ.gov/publications/publications-by-type/comptro...

"OVERRIDE RATE"
"Typically, the override rate is calculated by comparing the number of overrides, either high-side or low-side, to the
number of applications in that population."

Proporción/índice/relación anulaciones-solicitudes de préstamos

"Credit Risk Scorecards: Developing and Implementing Intelligent Credit Scoring"

http://books.google.es/books?id=onYzC_EGgO0C&pg=PA158&lpg=PA...