President Trump hosts a working luncheon with Baltic leaders Tuesday at the White House. (Kevin Lamarque/Reuters)

To hear President Trump tell it, his approval ratings are strong — and improving.

He pointed this out Monday on Twitter, quoting a Washington Examiner article that said his “approval rate among likely U.S. voters hit 50 percent,” giving him a better approval than President Barack Obama had at the same point in his first term.

On Tuesday, Trump put a fine point on it.

On Wednesday, even better news: His approval was “still rising,” as the Breitbart headline he tweeted said — hitting 51 percent! Everything’s coming up Trump.

Now, far be it from us to dampen Trump’s enthusiasm, but there are some important qualifications worth applying here. When he tweeted that thing about “Cheatin’ Obama” on Tuesday, we noted that the poll he kept citing, from Rasmussen Reports, was consistently much friendlier to him and more hostile to Obama than other polls.

At that point, Rasmussen’s poll numbers for Trump were higher than RealClearPolitics’ average of all polls 99 percent of the time. Through the same point in Obama’s presidency, Rasmussen was lower than the average 93 percent of the time.

It’s not a surprise that Trump cites Rasmussen all the time, given that it’s consistently better than most (all?) other polls. He cited it in early February, when his approval had “jumped to 49 percent.” He did so again in early March.

The problem is that even within the cherry-picked universe of Rasmussen’s polling, Trump’s picking cherries.

When he tweeted about his poll numbers jumping to 49 percent, that was correct. When he said that the polls were around 50 percent, they were actually at 45 percent. (Rasmussen uses a three-day rolling average; the dates we’re citing are the last day of each three-day period when Rasmussen updates its numbers.)

This week was particularly messy. The Examiner article saying he’d hit 50 percent came out on April 2, when he’d dropped back down to 49 percent. The day of his “Cheatin’ Obama” tweet, he was actually up at 51 percent. By the time he tweeted that number from Breitbart, he was back down to 47 percent. Not exactly “still rising.”

Poll numbers move around. There are margins of error and other variations that can mean that measuring the same audience can yield two different numbers even if no attitudes change. Pollsters look for statistical significance in polling changes, and a one-point movement between days isn’t that. But even in that context, saying that poll numbers that run 50-49-51-47 are “still rising” is nonsensical.

Trump highlights these numbers, of course, because he wants to present a particular picture of his presidency. He’s not historically unpopular, it’s just that other pollsters are not “honest” pollsters. (He bestowed that label on Rasmussen in the “Cheatin’ Obama” tweet.) If Rasmussen says he’s at 50 percent, he’ll announce that he’s at 50 percent even if no one else says it. If Rasmussen says he’s back down at 47 percent, well, he won’t announce anything. (For the record, the past seven poll averages released by Rasmussen average to about 47.7 percent.)

He does the same thing with the stock market. Over the course of 2017, he tweeted about the stock market incessantly as the indexes kept climbing. I made an index of my own, assigning points for each day that Trump tweeted about the markets (60 points) or the economy (30 points) and subtracting 10 points for each day he didn’t mention the markets.

Until January, it tracked closely to the changes in the Dow — a correlation of 0.95.

What happened in January? Well, shortly after this triumphant tweet on the anniversary of his inauguration, markets started to get rocky.

The first trigger was an announcement that wages increased quickly in January, raising concerns about inflation. Trump tweeted a lament about how, in the “old days,” good news like increased wages would send markets up, not down. Markets kept sinking after he tweeted that.

The Dow Jones industrial average, Nasdaq composite index and Standard & Poor’s 500-stock index began to recover from the February sell-off until last month, when Trump’s announced plan to add tariffs to products coming from China sent them downward again. Over the course of 2018, the Dow Jones and the S&P have lost value. The Nasdaq is about even.

But even in the midst of the recent volatility, Trump couldn’t help picking out a good number.

The Dow had its third-best one-day gain, prompting a celebration: “Great news,” he declared, following it with “#MAGA” — “Make America Great Again.”

But if America’s greatness is a function of sharp increases in the Dow, America’s greatness has had a rough year. Indexes are lower now than they were after that uptick; the trend in recent weeks has been downward, not upward.

Most presidents don’t obsess about poll numbers (at least publicly) and are wary of celebrating day-to-day fluctuations in the markets. Why? Because announcing loudly that things are going well because a number that changes often is indicating good news has the obvious downside of suggesting that things are going not well when the change is in the other direction.

Why Trump does this anyway isn’t clear. Does he just want to highlight good news for Americans to see? Does he assume that his base sees only his positive tweets and not the negative news in between them?

He keeps touting them. And we keep writing articles noting that his assessments are incomplete or misleading and that maybe he should stop doing cherry-picking numbers if he wants to avoid the inevitable switch from positive to negative.

Neither of us seems to be convincing the other.