I will never forget how it felt to have television cameras in my face and a puzzled reporter asking why I wasn't celebrating after the networks announced my congressional candidate had won.
It felt awful. But the situation was certainly understandable.
After all, the role of pollsters and the process for predicting victors often confuses people, including members of the media, leading to false hopes and frustrations.
We certainly saw that this year, as surprise candidate Donald Trump, despite predictions to the contrary, vanquished one Republican primary opponent after another before shocking the nation by capturing key Rust Belt states and defeating Democrat Hillary Clinton.
How do upsets like this happen, and why don't professional journalists see them coming?
That's a big question with many possible answers, so let's tackle one piece _ the networks' Election Day coverage, during which outlets are focused on three main things: exit polls, how early numbers compare to forecast models, and what happens in bellwether precincts and counties.
That equation, like any other, of course, is susceptible to errors, which is how you wind up with a sure thing turning into a shocker.
In this election, when the expected surge of Hispanic voters failed to materialize for Clinton, pollsters scurried to create new models. No one wants to call a state wrong after the Florida fiasco of 2000, when some said we'd have a President Gore.
But they missed another element _ the curveball thrown by Trump supporters that wound up skewing forecasts.
Many people don't feel comfortable telling a stranger how they voted, particularly if religion or race is involved.
It's what politicos call the Bradley effect, named after former Los Angeles Mayor Tom Bradley, an African-American Democrat, who despite leading polls heading into the 1982 California gubernatorial race, wound up losing to George Deukmejian, a white Republican.
This year, Trump supporters, after enduring months of unflattering press portrayals and harsh critiques of their candidate _ no fan of the media himself _ had the last laugh, fooling exit pollsters and thereby not triggering alarms about flawed turnout models.
Networks could have predicted the outcome sooner and more accurately had the polls and forecasts been more sound.
Then there's the matter of how votes are tallied.
Officials in Detroit's Wayne County, for example, made it difficult for the networks by not divulging how many absentee ballots they received, choosing to only tally them at the end, causing some viewers to wonder why broadcasters could not declare a Michigan victor with 80 percent of the vote already in hand.
That constant attention to the percentage of votes already tallied is often misleading. The first 30 percent reported, for example, is often not at all reflective of the next 30 percent that will come in, so the apparent loser can suddenly surge ahead to win, bewildering partisan fans in the process.
Anchors also frequently place too much emphasis on who is allegedly winning each county or state.
Campaign organizers worth their salt know they must outperform the historical norm precinct by precinct, and if they don't, the networks should see that problem when they check the specific precincts that almost always go with the winners. This year, these bellwether precincts gave the pollsters early indigestion, which is why numerous states were "too early to call" for some time.
An ultimately disappointing ground game is part of the reason Clinton lost, with her team's efforts _ lauded as they were _ falling short, garnering about 3 million fewer votes than President Barack Obama received in 2012 and about 6.5 million fewer votes than he got in 2008.
For all these reasons, I knew better than to assume Clinton had it in the bag, just as I knew not to join the big celebration after my congressional candidate had supposedly won. I knew some of my opponent's strongest precincts had not yet been tallied and it was too early to join the conga line.
