You raised your dose on Monday. You also started creatine that week, slept badly Wednesday and Thursday, and had a brutal stretch at work. By Sunday you feel different. Better, maybe. Or worse.
So what caused it?
You can't know. Not because you're not paying attention, but because four things changed at once and they're now tangled together in a way no amount of staring at your notes will untangle. That's a confounding variable problem, and it's probably the single biggest reason people spend years on TRT without ever figuring out their own protocol.
What a confound actually is
A confounding variable is something that moves at the same time as the thing you're testing, so you can't separate their effects. You changed your dose. But you also changed your sleep, your training, your stress, and your supplement stack in the same window. Each of those independently affects energy, mood, libido, and recovery. When the outcome shifts, every one of them is a suspect, and they all have an alibi.
The cruel part is that the explanation that feels most obvious is usually just the one you were thinking about most. You raised your dose hoping to feel better, so when you feel better you credit the dose. The two bad nights of sleep get quietly written out of the story. That's not analysis. That's confirmation bias wearing a lab coat.
Why TRT is unusually prone to this
Some systems are easy to isolate. TRT is not one of them. The symptoms people track on a protocol, energy, mood, libido, sleep quality, recovery, are exactly the symptoms that everything else in life also moves. A fight with your partner tanks your mood. A late espresso wrecks your sleep. A deload week changes your recovery. None of that is hormonal, but all of it shows up in the same numbers you're using to judge your hormones.
So the hormonal signal you're trying to read is sitting underneath a pile of non-hormonal noise that happens to look identical. This is the same reason single readings mislead. A confound is just a single reading's problem stretched across a whole week.
The two ways out
There are only two real strategies, and most people use neither.
The first is to change one thing at a time. Hold everything else as steady as you reasonably can, move a single variable, and wait. This is the disciplined approach, and it's covered in depth in why changing multiple variables breaks your protocol. It works, but it's slow and it requires patience most people don't have when they feel bad and want to fix it now.
The second is to log the confounds. You can't always stop life from changing four things at once. What you can do is record the other three so they're visible when you go back to interpret the week. If your dose change happened during two terrible nights of sleep and a stressful work stretch, that context is the difference between "the dose made me feel worse" and "I felt worse during a week where I also barely slept." Same data. Completely different conclusion.
This is what event logging is for
Recording context separately from your daily check-in isn't busywork. It's the mechanism that makes confounds visible instead of invisible. A poor night of sleep, a stressful day, a missed workout, a few drinks, these are events, and the entire reason they're kept separate from your daily state is so that later, when you're trying to read a week, you can see what else was going on underneath it.
Without that context, every week is a black box. You changed something, you felt something, and the wire connecting the two is invisible. With it, you can look back at a stretch where you felt unusually good and notice that it also happened to be a low-stress week with great sleep, which means it might have nothing to do with the protocol at all.
Why this gets easier with time
A single confounded week tells you almost nothing. But confounds tend to be random, and randomness averages out. The stressful weeks and the easy weeks, the good sleep and the bad, they scatter across months in no particular pattern. The protocol change you made, if it actually did something, doesn't scatter. It's there in every week after you made it.
That's the quiet advantage of tracking over a long enough window. Confounds are loud in any single week and nearly silent across three months. The thing that persists through all the noise is usually the thing that's real. You just have to log long enough, and honestly enough, to let the noise cancel itself out.