Fish behaviour is one biological organisational level regularly used to assess the state of freshwater ecosystems and can be monitored using fish telemetry methods. The development of activity sensors incorporated into fish telemetered tags allows for non-spatial movement to be detected and is increasingly used to understand the energy budgets and response and fine-scale behaviour of fishes. In addition, detecting tagged fish remotely and in real time highlights the need to process fish activity data in near real time to make it relevant to managers in the water resource sector. Our study on Labeobarbus natalensis, a cyprinid, in the uMngeni River in KwaZulu-Natal, South Africa, adapted and then tested the exponentially weighted moving average (EWMA), as developed for financial predictive modelling, using activity data from fish. To determine changes in behaviour, we compared the EWMA-predicted fish behaviour against the present fish behaviour. We showed that the EWMA could adequately detect changes in behaviour on both individual and population levels. Changes in behaviour are potentially indicative of a change in environmental conditions and thus were developed into management alerts. We conducted further analyses using generalised additive mixed models (GAMM) to determine the relationship between fish activity and the environmental data collected. The GAMMs helped determine the potential drivers for changes in behaviour where the EWMA could detect these in real time. Detecting changes in behaviour in real time as a result of environmental variables can identify thresholds of potential concern influencing management decisions and allow managers to respond, contributing to improving effective freshwater management.