This one has been on the back burner for a few months now, been toying around with some ideas as well as getting ideas from Sean Williams and Mladen Jovanović, the latter put up a great tutorial video on his website with similar thoughts here.
Rather than talk through the research behind this latest update in the ‘acute:chronic journey’, here are links to the mathematical coupling paper by Lolli and colleagues, as well as a recent editorial by Windt and Gabbett.
I’ve played around with a few ideas on what to do with the load in the ‘lag’ days from when the acute average starts and the chronic average starts, in order to avoid coupling, two ideas I had were to use the days load for the lag period or average all the data in this period.
I decided to average the first 7 days of the chronic load and begun the EWMA calculations on day 8, with the acute starting on day 1, giving a 7 day period between chronic and acute.
On day 8 the chronic takes the 2nd days load, and puts in into the EWMA equation – 𝐸𝑊𝑀𝐴𝑡𝑜𝑑𝑎𝑦 = 𝐿𝑜𝑎𝑑𝑡𝑜𝑑𝑎𝑦 ∗ 𝜆𝑎 + ((1 − 𝜆𝑎) ∗ 𝐸𝑊𝑀𝐴𝑦𝑒𝑠𝑡𝑒𝑟𝑑𝑎𝑦), where the 2nd days load is the “Loadtoday”, and the 1st Days Load is the “EWMAyesterday”).
Meaning on the 8th day the chronic loads start a week behind, whilst the acute load still uses this original equation as in bold. So by the time we hit day 28, we will have a 21 day chronic and a 7 day acute completely separate of one another.
I’ve put in some charts to compare the ‘coupled’ and ‘uncoupled’ methods as well as a line chart showing how the acute and chronic data averages fluctuate, with bar columns showing how the a:c ratio reacts to these trends
I DO NOT KNOW IF THIS IS THE ‘RIGHT WAY’ OF DOING THINGS, MERELY JUST PLAYING AROUND WITH SOME IDEAS THAT MAKE SENSE IN MY HEAD, HOPING TO GENERATE SOME DISCUSSION AND OTHER IDEAS ON THIS TOPIC.
I’ve used some manual and automated formulas in this spreadsheet for comparison and to validate the automated formula I’ve used, as well as this I start the acute:chronic ratio on day 28, as there probably is nothing to be gained from using the ratio while the chronic load is still building.
As always keen for feedback/discussion!
See the download here