Backtesting
Backtesting and backtesting referred to evaluate a strategy, theory or a model by the strategy and the theory and the model is applied to historical data the process. Typical applications for backtesting include examining the following questions:
- What results would a trading strategy have had on a stock market in the past?
- How well did the predictions of a climate model fit together with the actual weather?
A key difference between backtesting and other historical tests is that backtesting calculates how a strategy would have behaved if it had actually been executed. Therefore, for a correct backtesting result, it is necessary to correctly replicate the relevant historical conditions. Assuming that future events are very similar to past events, backtesting can be a useful tool for analysis and forecasting.
Backtesting in finance and economics
Backtesting in trading strategies
New investment and trading strategies in capital or stock markets are tested by observing what results they would have produced in the past if they had actually been applied. This type of backtesting tries to learn from the “mistakes of the past” without losing money. Since real data serve as the basis, it is hoped that the weaknesses of a strategy will be better recognized than when using synthetic data. Backtesting was first used in 1983 by Louis B. Mendelsohn in commercial software for the PC.
Backtesting as quality assurance in risk management
Backtesting is used in the risk management of banks to check the quality of the risk measure value at risk . It compares how often the forecast loss limit is exceeded. If there are too many so-called outliers, the value at risk model must be modified. Banks must also carry out regular evaluations for rating models. Backtesting the probability of failure plays a major role here. The extent to which the predicted number of failures corresponds to the number of failures actually occurred is checked, the binomial test often being used for this in practice .
Backtesting on climate models
Backtesting plays a major role in the evaluation of weather models and climate models . It is tested whether new models with historical input data can also reproduce the historical weather or climate.
Individual evidence
- ↑ Stocks, Futures and Options Magazine, August 2010
- ↑ Philippe Jorion: Value at Risk, the new benchmark for managing financial risk. McGraw-Hill, 2007, ISBN 978-0-07-146495-6 , pp. 139 ff.
- ↑ Regulatory framework for backtesting (back-testing) when calculating the capital requirement to back up the market risk with the bank's own models, Basel Committee on Banking Supervision, January 1996
- ↑ ÖNB guidelines on market risk. Volume 3, Assessment of a Value at Risk Model, p. 27, archive link ( Memento from March 18, 2013 in the Internet Archive )
- ↑ ÖNB series of guidelines on credit risk: Rating models and validation, April 2004, archive link ( Memento from March 18, 2013 in the Internet Archive )