While a moderately high dosage of Ketamine does have anesthetic effects, through it's antagonism of the NMDA receptor at dosages (which suppresses activity required for signals to cross spine/brain), inhibition of nitric oxide production, and weak agonism of opioid receptors, it also has a slew of other interactions at various affinities.
Ketamine inhibits the reuptake of all three major monoamines. Thus it mimics to some degree both the mechanisms of action of SNRIs (serotonin/norepinephrine reuptake inhibitor) such as venlafaxine (effexor), used to treat depression/anxiety and DNRIs (dopamine/norepinephrine) such as bupropion (wellbutrin)/methylphenidate (ritalin), which are used to treat depression/ADHD. Ketamine also impacts the acetylcholine receptors, which have been linked to anti-depressant effects, and the less-well-researched sigma receptor.
While the above interactions may contribute to the reported anti-depressant effects, some of those same interactions make it undesirable as a medication. Dopamine reuptake inhibitors and mu-opioid antagonists are notoriously abused for their euphoric effects. However, those effects are contrasted with psychotomimetic effects of ketamine's interaction with NMDA, kappa-opioid, muscarinic acetylcholine, and simga receptors.
Ultimately, it's incorrect to simply categorize ketamine as a depressant or a sedative. The body is an exceptionally complex chemical network, and you're right to be skeptical. But I think it's important that we not dismiss the positive reported effects of the drug because of the side effects that occur at various points on the dose-response curve. We may find out that by controlling dosage we can tune a beneficial level of antidepressant response without psychosis, dissociation, or amnesic effects due to hormetic effects. Or it might work exceptionally well for some otherwise treatment-resistant cases. Even if not, we'll be able to use that data to tweak the binding affinities to investigate drugs that have a better therapeutic use.