The two ways to underwrite land appreciation gave answers 40% apart

I tried to put a number on expected appreciation for a 26 acre parcel in an exurban county and got two answers I can't reconcile.

Method one, county trend. I pulled fifteen years of vacant land sales the county publishes, threw out anything under 2 acres and anything over 200, and got a median price per acre compounding around 5.1% a year, with two flat stretches and one 2021 spike that distorts everything. Applied forward over a ten year hold, the parcel goes from $4,100 an acre to roughly $6,750. Note that about a dozen states don't make sale prices public at all, so this method just isn't available in some places.

Method two, land residual from the developed ring. Nine miles closer in, finished lots trade around $78k. Builders I've talked to say they can pay somewhere near 20 to 22% of finished lot price for raw ground before entitlement and infrastructure. That implies my land is worth about $15,600 per lot equivalent whenever the ring reaches it, which at roughly 2.2 lots an acre net of roads and detention is a much bigger number than $6,750. It just says nothing about when.

So the trend method prices continuity and the residual method prices arrival. My carrying cost is $1,180 a year and doesn't care which one is right. Everything hinges on whether the ring actually reaches nine miles in a decade, and I have no defensible way to estimate that.

Which basis do you actually underwrite on, and if you use both, how do you resolve them?