If you're working harder, eating better, and doing worse, it sounds like you're overreaching/overtraining a bit. I would recommend getting your cortisol levels checked in addition to your T and Free T levels. You might want to have your cholesterol levels checked as well. Just to be safe. Sometimes when guys get really deep into overtraining the constant release of glucocorticoids starts to really screw with the LDL/HDL balance.
CF has a real tendency to burn people out. Intensity and constant variation put a demand on your CNS that over time can really dig a hole for you. You would not believe the number of CFers on steroids. It's like the dirty little secret that no one wants to talk about.
I'm guessing from your handle you train at CF Forged, so you're in or near San Diego. There are a ton of naturopaths in and around San Diego. Before you go turn to AAS, it would be good to get an idea of why you're not feeling dialed in. An adrenal stress index would likely be sufficient to establish your cortisol levels and circadian rhythm. Also, cortisol can really sap your T levels. You could have what many people wrongly refer to as "adrenal fatigue," but it's really just adrenal dysfunction because more likely, you're adrenals are overworking early in the day and underproducing later in the day.
In the meantime, you might try some carnitine-L-tartrate to help your CNS, and maybe take some phosphatidylserine pre workout. If you really think it's your T levels, you could try running Primordial Performance's TRS with their TCF-1, and if that makes you feel much better, then that would probably point to T problems. However, if you have insurance, it's probably cheaper to just see your doc and get the bloodwork done.
What's your diet like? Lots of fruit? How about sleep? Hard time falling asleep? Wake up multiple times? What's your training look like? Multiple WODs a day? Are you very active outside your workouts? Are most of your WODs over 10 minutes? When you do strength work, how often do you max out? All these things can really contribute to exactly what you're describing.