Article Text
Statistics from Altmetric.com
In replying to a PostScript on the use of rolling averages,1 Drew and colleagues failed to respond to the main point of the letter. In fact, the original commentary raised concern on the specific mathematical model used for quantifying acute and chronic training load (TL), not on how TL can be subsequently analysed and interpreted. This could be of some concern, as it seems to suggest the authors may be confusing a measure (eg, TL) with analysis of data (eg, acute:chronic workload ratio).
Drew and colleagues2 also reported the use of sophistry. Interestingly, in a recent letter, Williams and colleagues3 expanded on the rolling averages topic further clarifying the limitations and clearly proving that there was no sophistry in the initial letter.
Forty years of research,4-7 as well as popular online systems used in professional sports to monitor TL,8 9 have implemented and fine-tuned different kinds of exponentially weighted moving averages (ie, non-linear approach) to model the declining effect of training stimulus over time. Mathematically, rolling averages do not account for this basic principle of human physiology, hence their use in this context seems to be far from ideal.
To avoid the risk of building evidence with flawed data, only the use of valid data should be considered in research. For instance, only TL quantified with appropriate modelling should be used in TL analysis.3 Given that the use of rolling averages for the quantification of TL seems to be lacking ecological validity and validation, its use for quantifying TL should be interrogated before being applied broadly.
Footnotes
Competing interests As previously stated I have declared to the editor I may have a potential conflict of interest, however I was told it was not an issue.
Provenance and peer review Not commissioned; externally peer reviewed.