( in Quantitative Finance, AI, ML, ... ) A forward-testing ( as opposed to back-testing ) is a systematic approach to review & validate a quality of a System-under-Test [ SuT ], observed indirectly by it's ability to "generalise" ( in contrast to an otherwise un-avoidable overfitting artefact coming from a use of a trivial penalty-function minimiser being used in a back-testing phase )
( in Quantitative Finance, AI, ML, Technical Cybernetics, Modelling ... )
Forward-testing
is a systematic approach to review & validate a quality of a System-under-Test [ SuT ], observed indirectly, by it's ability to "generalise" ( in contrast to an otherwise un-avoidable overfitting artefact coming from a use of a trivial penalty-function minimiser being used in a back-testing phase ) so as to review SuT's behaviour in-almost-vivo ( as opposed to a back-testing ).
Strengths & Weaknesses
The concept of forward-testing relies on a metodical, systematic separation of available data set, that was used for the SuT design / back-testing optimisation, from the second part of data, being used for the forward-testing phase, be it by using "fresh" live-data, being generated from the external environment ( the outer ecosystem, surrounding the SuT, ) during the test, or by an a-priori split of the ( already static ) historical data.
