For the last couple of years, the United States men's national team has been in constant tinkering mode. Both current coach Gregg Berhalter and his longtime interim predecessor Dave Sarachan constantly experimented with lineups -- Sarachan played 56 different players in just over a year, while Berhalter has played 50 since taking over in Dec. 2019.

All that tinkering makes it difficult to glean much from the matches the U.S. has played, even more so when we look only at matches against solid competition. Under Berhalter, the U.S. has played just nine matches against teams in the top 50 of the FIFA rankings: two defeats to Mexico (1-0, 3-0), draws against Uruguay and Chile, a 3-0 pasting by Venezuela, two wins over Costa Rica (2-0, 1-0), and a win-loss split (0-1, 3-1) against Jamaica.

- Stream new episodes of ESPN FC Monday-Friday on ESPN+
- Stream every episode of 30 for 30: Soccer Stories on ESPN+

One thing we can take away from those nine matches, however, is that the USMNT's goal and Expected Goal totals were in stark disagreement.

The Expected Goals measure in soccer (xG) is both brilliant and, in very specific ways, limited. Goals are inherently random, and xG provides important context regarding the quality of your and your opponent's opportunities. In a process-versus-outcomes universe, xG can fill in the gaps regarding how your processes will likely affect your outcomes moving forward.

In a small-sample or single-player instance, however, xG leaves as many gaps as it fills. Looking at xG totals for a single match doesn't tell you all that much about how a team did or didn't control those 90 minutes. Meanwhile, a player who is an elite finisher will likely see his goal totals exceed his xG totals on an annual basis.

Lionel Messi, for instance, has exceeded his xG in nine of the last 10 La Liga campaigns -- the time he didn't, he was basically dead even (26 goals, 26.1 xG in 2015-16) -- and has produced 350 league goals, drastically exceeding his 278.2 xG. It's safe to say that wasn't simply luck.

Similarly, a particularly bad finisher will usually fall short of his xG totals. Hold that thought.

Goals and xG tell very different stories about the USMNT's recent performance

The enthusiasm level for the U.S. fan base has... been better. Average attendance at home matches since the start of 2018 is just 22,966, 19,073 if you take out three matches with attendance cheat code Mexico. Sure, these have been mostly meaningless matches, but from the end of the 2014 World Cup through 2015 -- a similarly meaningless period -- the U.S. averaged 36,689 (30,172 without Mexico) even after you exclude late-2015 WC qualifiers. That's a 37 percent drop overall.

The reasons, of course, are obvious: the USMNT failed to qualify for the 2018 World Cup and hasn't played particularly inspiring soccer in a while. Then there in the way the USSF has battled the U.S. women's team during its equal pay push, which has almost pitted the USMNT against the USWNT. Not great for the far less successful of those two teams.

Fan enthusiasm could only improve so much in 2019 no matter the performance, but some happy results would have helped immensely. And if you take xG at its face, the U.S. was unfortunate not to fare better in marquee opportunities. That 3-0 loss to Venezuela on June 9? The U.S. actually had 1.3 xG to Venezuela's 1.1. The losses to Mexico? The U.S. had a combined 2.8 xG to El Tri's 2.7. The 1-1 draw with Uruguay (currently fifth in the FIFA rankings) on Sept. 10? U.S. 1.8, Uruguay 1.1.

In total, the U.S. produced 14.9 xG in this nine-match sample compared to 8.3 for their opponents. In reality, these opponents outscored the U.S., 11-8. If this was bad luck, then progression to the mean could be awfully kind to this team in upcoming performances (whenever those upcoming performances might be).

But was it bad luck? Or are the U.S.'s finishers just bad at finishing?

Berhalter's style has taken effect

In his five-year stint with the Columbus Crew, Berhalter's intentions were obvious. He wanted to establish a higher level of ball dominance than the average MLS team, and while the Crew's overall possession rates (53 percent over the five years) was only slightly above average, most of their rate stats matched that of the average possession-style team.

The Crew built from the back and methodically worked the ball forward. It was an attempt at what you might call the modern possession style, and it worked pretty well. Columbus generally created more chances and shots than their opponents, they made the MLS playoffs four times in Berhalter's five years, they reached the conference finals twice, and they nearly won MLS Cup in 2015.

Whether it was down to philosophy or a lack of the right personnel, however, Columbus wasn't much for pressing. The Crew won fewer possessions than average in the opponent's half, and most of their defensive action stats (interceptions, tackles, clearances, etc.) were well below average. This lack of defensive pressure meant that they didn't create many easy goal-scoring opportunities, and they simply relied on their ability to play the possession game better than their opponents.

Focusing on the U.S.'s nine matches against top-50 competition, we see a lot of the same traits, especially when we compare their stats to what other opponents did against Uruguay, Mexico, Chile, Venezuela, Costa Rica and Jamaica in the same period.

You can see how Berhalter was enforcing his style of play. Even against this solid set of opponents, the U.S. averaged more passes (especially in their own half) and longer possessions than the Crew. You could make the case that they were too passive here -- that Berhalter wanted to establish the style even at the cost of aggressiveness and opportunity -- but that would make sense if you treat these matches as experiments and not matches of consequence.

Combined with decent set-piece play, they created pretty good chances compared to the other opponents, too. Their slow build-up play also meant that, more often than not, their defense wasn't caught in awkward positions when they lost the ball. Opponents averaged only 1.2 goals per match, after all, compared to 1.7 against other opponents.

In these nine matches, the 15 players above -- a list that includes many of the guys Berhalter will be relying on whenever World Cup qualifying actually takes place -- generated a combined xG of 14.2 but scored just eight goals. Take out Christian Pulisic, and the numbers are 12.6 and five.

Those are some horrible conversion rates. And if you go back and watch those matches, you certainly see plenty of mistakes. D.C. United's Paul Arriola made constant kamikaze charges into the box only to poke the ball right to the goalkeeper. Josh Sargent missed a couple of sitters in the box and had a tame penalty saved against Mexico. Jordan Morris scored against Uruguay and forced some good saves, but he also sent a couple of headers right at the keeper. Other forwards like Gyasi Zardes and Jozy Altidore barely created chances and did little with what service they got.

Watching in real-time, you could easily conclude the USMNT's problems here were based on a lack of finishing talent. But we're also talking about tiny samples. Let's look at what this same batch of players did in their league matches over the last year of action (minus Ulysses Llanez and Tyler Boyd, for whom we don't have league data).

Granted, Zardes has an iffy reputation as a finisher, while Arriola could stand to find a bit more poise in the box. But in MLS play both Morris and Altidore created more goals than their chances would have suggested, and with first Borussia Dortmund and then Chelsea in far tougher leagues, Pulisic, the U.S.'s only ace finisher in 2019, was anything but. Sample sizes can do a number on your perceptions.

Given the same chances, Morris could have easily scored a couple more goals and either flipped a result or made a match like Venezuela far less embarrassing. And given more of a run in World Cup qualifying, Altidore could still provide lots of value. The U.S. wasn't far from some encouraging results in 2019, and if certain players' conversion rates revert more toward their historic norm, we could get a clear, and encouraging, picture of how Berhalter's attack is supposed to work.

The U.S.'s other problem: random catastrophes

Indeed, opponents in these matches scored fewer goals than they did against other opponents. That's good. But among those 11 goals allowed were four that came directly from turnovers in the U.S.'s half of the field. Two -- the first against Venezuela and the second in the 3-0 loss to Mexico -- were inexcusable disasters.

At club level, with lots of matches and far more practice time, you can get pretty intricate in your tactics as you look for ways to build from the back and stretch defenses with long possessions. At the international level, however, that's much harder to pull off. The players aren't as familiar with either the manager or their teammates, and they can't build that familiarity in limited time. It's one of the reasons you often see coaches simplifying their methods in international play. Even Arrigo Sacchi, the mythical and accomplished AC Milan manager, had to simplify when he led the Italian national team. And if Sacchi struggled in some ways, lord knows Berhalter could too.

In these nine matches, U.S. opponents won the ball 4.9 times per match in the attacking third. That's not a terrible number, but combined with the U.S.'s own lack of defensive pressure -- they won just four possessions per match in the attacking third -- it created a deficit. In the last five World Cups, 66 teams have had a deficit of at least 0.5 possessions won per match in the attacking third; only 11 of them reached the quarterfinals or deeper.

In a tournament setting, with small sets of matches in group play before single-elimination knockout rounds, random, easy scoring opportunities are prime currency. And if the U.S. isn't going to create many of those -- which, again, really isn't part of the "Berhalter DNA" -- it must allow almost none. That means that Berhalter's play-from-the-back style has to grow far sharper. Maybe he can pull that off once the lineup changes dissipate and the practices get more focused in World Cup qualifying time. And maybe progression-to-the-mean smiles on the U.S. when it comes to finishing the chances Berhalter's style creates. Or maybe both he and the U.S. will soon find out why so many managers can't play a league style at the international level.