Not to contradict the other answers but to add:
Along with the return to the mean James K. mentions in his answer, there are, maybe surprisingly, legitimate reasons a person, of any ethnicity, could have decided to support Trump by seeing his actions in office (I'm hard pressed to say many would have changed opinion by listening to him). I say this as there are still many single issue, or "few issue" voters. Below I outline a couple specific points that likely explain at least a portion of the increase in the Latino vote for Trump in 2020 compared to 2016.
According to this report in the National Archives, written by Trump's staff while leaving office1, so please be warned of the bias, here is a list of accomplishments that could have swayed Latino voters:
Before [the pandemic], we built the world’s most prosperous economy.
America gained 7 million new jobs – more than three times government
Middle-Class family income increased nearly
$6,000 – more than five times the gains during the entire previous
The unemployment rate reached 3.5 percent, the
lowest in a half-century.*
Achieved 40 months in a row with more
job openings than job-hirings.
More Americans reported being
employed than ever before – nearly 160 million.
Jobless claims hit
a nearly 50-year low.
The number of people claiming unemployment
insurance as a share of the population hit its lowest on record.
Incomes rose in every single metro area in the United States for the first time in nearly 3 decades.
Explanation: Most of these statistics also are disproportionately beneficial to working-class and lower income families. As is, regrettably, a well known issue in our country, Latinos still make up a disproportionate portion of that group. (unemployment affects those with less wealth more greatly than those with more wealth)
And most specifically to the demographic mentioned:
Delivered a future of greater promise and opportunity for citizens of all backgrounds.
Unemployment rates for African Americans, Hispanic Americans, Asian Americans, Native Americans, veterans, individuals with
disabilities, and those without a high school diploma all reached
Unemployment for women hit its lowest rate in
nearly 70 years.
Lifted nearly 7 million people off of food
Poverty rates for African Americans and Hispanic Americans reached record lows.
Income inequality fell for two straight
years, and by the largest amount in over a decade.
The bottom 50
percent of American households saw a 40 percent increase in net
Wages rose fastest for low-income and blue collar workers –
a 16 percent pay increase.
African American homeownership
increased from 41.7 percent to 46.4 percent.
The point is there were numerous, legitimate success stories for the US during the Trump administration in specific areas. Many people are still 1-3 issue voters. Regardless of one's overall feeling towards the Trump administration, it's easy to see how another's vote could be swayed by specific positives that are more prevalent than other negatives in their lives. The above are all specific reasons why the Latino vote may have swayed more drastically than other demographics - but is by no means a comprehensive list.
Whether it is fair to attribute the economic state at that time to Trump or his policies, these improvements did happen during his presidency, which certainly impacts voter opinion - even if all the economic improvements were, i.e. due to consistent quantitative easing since '09, or pure speculation on advanced deregulation further than anything that occurred from 2016-2020, or anything else outside any current president's control.
1 "and later reviewed for accuracy by historians/archivists (but maintaining biased, "fluffy" language)" this is a statement I thought this morning to be accurate, but I'm not sure is, I will be researching, but divisban has made very good points about this specific type of entry to the archives. I'm hard-pressed to believe the core-data would be inaccurate despite clear political bias in the phrasing, but I cannot verify at this time. Suggestions are welcome. - also alternative sources for some of this data would be much appreciated (or better yet for my laziness, primary sources cited by Trump's team for this doc)