What happened:

• U.S. unemployment peaked at 14.7% in April 2020 but had fallen to 6.9% by October 2020 and 6.7% in November 2020, well below the 16–22% range mentioned in the prediction. (bls.gov)
• In the November 3, 2020 election, Donald Trump lost to Joe Biden, who won 306 electoral votes to Trump’s 232. (en.wikipedia.org)

Why the prediction is ambiguous:

• Chamath’s claim was explicitly conditional: if unemployment were around 16–22% going into the election, then “you could field the prancing dog and he’ll beat Trump.”
• That condition did not occur; unemployment was under 7% by the election period, so the specific scenario he described never materialized.
• Because the antecedent of his conditional prediction was false, we cannot empirically test whether, under 16–22% unemployment, Trump would indeed have lost even to a very weak opponent. The fact that Trump lost to Biden at a much lower unemployment rate is consistent with his intuition that very high unemployment would doom Trump, but it does not confirm or falsify the exact prediction as stated.

Therefore, the appropriate evaluation is “ambiguous”: there isn’t enough real-world evidence to determine whether his conditional claim was right or wrong.