There is no honest combined training load

Most training logs measure one thing well. Runs or lifts. Not both.

This looks like a product decision until you try to measure load for someone who does both. Then you find out it isn't a product decision. It's a literature decision. No validated formula exists that combines endurance and resistance training into a single number. Nobody has published one. I went looking. I read every review from the last ten years. The answer is the same everywhere. Don't try.

The closest thing is a method from 2001 that works by asking the athlete to rate the session on a scale of one to ten, then multiplying by duration. It's the only modality-agnostic load method in the literature that anyone takes seriously. It gets used because nothing else exists.

This seems like a small point until you sit with it. The apps that show a training load number every week claim to know something. If the training is only running, they might be approximately right. If the training is running and lifting, they are either ignoring half of it or fudging it.

Different systems, different fatigue

The honest reason there is no combined metric is that the two kinds of training stress are not the same thing. A long run at easy pace mostly stresses the cardiorespiratory and metabolic systems. Metabolic fatigue clears in hours. Neuromuscular fatigue from the eccentric loading and sustained effort can persist for a day or two, depending on duration and terrain. A heavy set of squats stresses the muscles, connective tissue, and nervous system in ways that heart-rate-based metrics don't capture. Fatigue from it can linger for two to three days. The adaptations are downstream of different signaling pathways. The recovery curves are different. The doses that produce adaptation are different.

Any attempt to add those two things together requires picking a currency. Pretend both are measured in heart-rate-weighted minutes and the lifting gets underestimated. Pretend both are measured in kilograms of mechanical work and everything aerobic disappears. Pretend both are measured in perceived effort and intensity profile information is lost. The perceived effort method is valid, but every attempt to unify the two loses something important.

The research community has been looking at this for a long time. There are reviews pointing out that the fundamental barrier is incommensurable units. Heart-rate-weighted time on one side. Kilograms of work on the other. Different recovery timecourses. Partially overlapping but distinct molecular signaling, including the AMPK-mTOR axis that early research identified as a key interference mechanism, though the clean dichotomy is less clear in human muscle than cell studies suggested. The interference effect is real but smaller than people thought twenty years ago. For recreational athletes, it is within noise.

So the question is: what do you do if you want to measure load for someone who does both?

Two loads, not one

The answer I arrived at is to measure them separately and show both. Not as a clever design decision. As the only defensible thing to do.

Duostride tracks endurance load from heart rate data when it is available and from perceived effort when it isn't. It tracks strength load from what gets lifted and how hard the sets are pushed, falling back to perceived effort when the session doesn't have that detail. Each number has its own fitness curve. Each has its own fatigue curve. Neither pretends to speak for the other.

This means the weekly view has two numbers where other apps have one. It makes scanning slightly harder. It's the honest trade. One number for two systems is a lie that asks for trust. Two numbers let the picture through.

The endurance load uses Banister's exponential TRIMP, rooted in the impulse-response model from the mid-1970s and formalized with sex-differentiated heart rate weighting, which accounts for different heart rate responses in men and women, in the early 1990s. It multiplies time by heart rate intensity raised to an exponential. It is validated across running, cycling, swimming, and team sports. It falls back to a power-based method for cycling with a power meter, and to a simpler heart-rate zone method when the data is less complete, and eventually to perceived effort times duration. There is always a path to a number.

The strength load is harder because there is no settled answer in the literature. What Duostride uses is effort-weighted volume load. Total reps times weight, adjusted by how close to failure the working sets were pushed. The adjustment factor is small: ten percent or so at the edges. Without logged sets, Duostride falls back to perceived effort times the portion of the session that wasn't rest. It is not perfect. It is the best defensible option given what the literature supports.

Both loads feed into a pair of fitness and fatigue curves. The shape of those curves is old news. The model is from the 1970s. An exponential moving average for fitness. A faster exponential moving average for fatigue. The difference between them is called form, a word that sounds fancy and means how fresh the athlete is today. The new thing in Duostride is that the time constants are different for the two sides. Endurance fitness is slower to rise and slower to fall. Strength fitness is faster on both counts. This isn't arbitrary. Strength adaptations show up in neural drive within weeks. Strength fatigue from heavy lifting clears in days. The literature is clear about this. What the literature hasn't done, as far as I can find, is run the two models in parallel for the same person. That's what Duostride does.

What this can and cannot tell you

I want to be specific about the edges of what this thing knows.

It can show when endurance fitness is climbing, flat, or falling. It can show when strength fitness is doing the same. It can show when this week's load is above or below what the body is used to. It can flag a single session that was significantly harder than recent training, which some research associates with elevated injury risk, though the evidence around acute load spikes and injury is still debated. It can flag when training has been very uniform for a while, which some models associate with elevated risk. It can show when short-term fatigue is climbing faster than fitness is building, which is usually the story of a hard training block.

It cannot diagnose overtraining syndrome. That is a clinical exclusion diagnosis that takes months of observation and rules out every other cause of fatigue. No app can do that. It cannot diagnose non-functional overreaching either. The patterns look identical in real time. The distinction can only be made in the rearview mirror.

It cannot estimate VO2 max. It cannot read hormonal status or immune function. It cannot predict a peak. It cannot prescribe tomorrow's workout. These are all things a coach can help with. They are not things a log can do from the data it has.

The readouts Duostride shows are observations, not prescriptions. If it says cardio fatigue is climbing fast, that is what it sees. Whether to do something about it depends on the training goal and what else is happening. That is the athlete's call.

The interference effect, or not

One more thing about concurrent training.

Anyone who reads about hybrid athletes will run into the interference effect. The original paper is from 1980. Strength gains are smaller when endurance training happens at the same time. The effect is real. The size of the effect has shrunk each decade as better studies were run. Recent meta-analyses place it smaller than early studies suggested. The effect on maximal strength appears minimal. Hypertrophy interference depends on endurance volume and modality. Explosive power and rate of force development remain the most reliably affected.

In recreational hybrid training, the effect is smaller than the noise in any individual's training response. Duostride does not apply an interference penalty to load numbers. It would be false precision. It does still notice when a heavy lift and a hard run land on the same day, because that's a useful observation to surface. But it does not invent a penalty factor in the load math.

For an elite athlete trying to squeeze out the last few percent of something, the interference effect matters and this log is not the right tool. For everyone else, the honest answer is that it is within the noise.

The first honest read

I built this for myself. I do a little of everything. I wanted to look at one page and know whether I was building, holding, or drifting. I wanted two numbers because I know my runs and my lifts are different, and I wanted to see them behave differently over time.

If it is useful, welcome. If something is wrong in it, please say so. This is the plan. I expect to update it as I learn more and as the research moves.

Nothing in it is going to tell anyone to skip their squat session. That is not its job.