Thinking about it, I have to disagree. It seems likely that a majority of cases are ones where the first runner-up lost by calendar year MQD. It's fairly clearly the sort step most likely to create enough buckets that the top N buckets have exactly as many fliers as upgrades remaining.

Consider an A321. 20 F seats. 17 are sold (incl. awards and FCM), leaving 3 for upgrades. Let's say there are 12 DMs who aren't in F yet.

"Cabin purchased". Deliberate ambiguity now around C+, but that doesn't matter here: the modal number of C+ purchasers out of 12 DMs is almost certainly zero, so we're still at 12 chasing 3.

MM Status. Out of 12, what are the chances of exactly 3 being MM? If it's 2 or fewer, than we're going to have 10+ 0MMs fighting for 1-3 seats. If it's 4+, then they might get bucketed into finer grained MM statuses (DL has removed the verbiage suggesting that raw lifetime miles is used to sort, but added verbiage suggesting that there are multiple MM statuses), but the odds are against there being 3 2+MM, so we almost surely need more tiebreaks among the MMs. Let's say that there are 2 MMs out of the 12, leaving 10 0MMs for 1 seat.

Out of those 10, say 5 have the Reserve card. 5 chasing 1.

Number of corporate flyers probably varies wildly by flight. There's unlikely to be exactly 1 out of our 5 DM 0MM Reserve cohort: 0 or 2 are more likely, so either top MQD from 5 or top MQD from 2 gets the upgrade (meaning that the runner-up "could have had class" by getting more MQD since January 1).

If you're normally 12th runner-up for the last upgrade, then it's unlikely that a relevant number of CY MQD helps you. But conditional on being 1st runner-up, you probably missed because the winner had more MQD.