Hello coaches,
I’m working on a research idea where the goal is to create a protocol that tells you how long a conditioning game should last to reach a certain RPE.
For example, if I want a training session to feel like an RPE 8 on average, the formula would give me the right duration.
So coaches will be able to design an overload training or underload training more objectively to prevent fatigue etc.
Is this possible or do you need more details to answer that question?
Dear @Berk great to hear from you.
The goal you have in mind sounds pretty challenging, due to the variability of your main outcome variable, i.e. RPE.
You write about an “equation”, so I assume you already have an approach in mind. What kind of equation are we talking about and what are the other controllable variables involved.
To exemplify: RPE is the outcome “dependent” variable. What are the controlled “independent” variables?
And, perhaps the most important point: what is the precision you would like to get to? With the state-of-the-art methods, how close you will get to your expected RPE? It is important to evaluate this “target precision” with the intra-individual (differences in reported RPE for same individual-different sessions) and inter-individual (differences in reported RPE for different individual-same sessions) variability. If your target precision is lower than these variabilities, we’ll be in trouble.
Sorry, more questions than answers at this stage 
but these are the crucial points that need some thinking before even start drafting a model 
Dear @Andrea great to hear from you too. Thank you for your answer.
What I aim to test is the overload duration (corresponding to 7–10 RPE) of a 4v4 small-sided game. The players will play 4v4, and I will pause the game every 3 minutes to collect their RPE scores. If the average RPE is below 7, they will continue playing for another 3-minute block. Once the average RPE reaches the 7–10 range (e.g., after 6 minutes = 8 RPE), I will define that time point as the “overload duration.” But here’s the interesting part: What if I want to design a session targeting underload, say RPE 5, such as on match day -2? Could I use this protocol in reverse to estimate how long the players should train to stay within that RPE range?
Sorry I couldn’t mention the variables etc. Because I do not know 
Dear @Berk it sounds like the training duration is the only variable you manipulate? What kind of relationship did you observe in your experience between time and RPE? How much variability did you observe?
It seems like you can indeed find the estimated time that will get you to RPE 5, but it depends on how solid this relationship is. Do you have data to test these assumptions?
Hello Andrea, unfortunately, I don’t have any data at the moment. Initially, I just wanted to confirm whether it’s feasible. Assuming I do have the data, would it be appropriate to use a linear regression model to calculate it?
@Berk the linear regression is most likely your first shot, at least to set a benchmark and see how far you can get with the simples model out there. However, I suspect the responses will be highly individual. Plus, RPE will not be normally distributed, and you can treat it as a discrete variable. However, the short answer is yes: linear regression your first shot. Then individualised linear regression. Then it all depends on the kind of accuracy you can achieve with these models