Am I being trained properly?

[Long Walk]

I bet they do if he claims it.
However, I recall past examples where he claims his regression methods will result all comparable sales adjusting to the same exact amount. - of price ( not value )
But even if you swap out value for price, arriving at such a precise exact sale is itself a problem because it does not reflect natural market variances. He also ( like many statistical programs ) feeds in large amounts of data to get a result - for example 40 sales or 90 sales -l how many of the 90, or 40 etc sales are comps for the subject ? 5? 10? 20? -it means we are relying on data from less similar to subject properties. While that can work for broader trends, such as a market condition adjustment or a location adjustment, it can skew results in other aspects From a past example he gave his program came out with a adjustment for a fireplace. Cleary that result was preposterous, yet it was "proven" by the math. But it needs to be proven by the market.

Large municipalities with lots of data to use for regression can't achieve better than 90%± accuracy. IOW, if the assessment is within 10% of market value, high or low, the property is fairly assessed. Courts have ruled on this repeatedly in different places.

At best, all valuation by anyone comes down to judgment based on the best data available. Here is what the NH Supreme Court said about market value in 2018 citing a 1974 decision: