A number of things seem to be broken when calculating cart estimates, possibly due to the same underlying cause:
Matching estimates for USDT is calculated wrong (very high amounts)
Matching estimates for fractions of MATIC/DAI/USDC is higher than for whole amounts (i.e. 10x more matching for 0.1 MATIC than 1 MATIC)
Matching estimates for fractions of MATIC are lower than for equivalent amounts of DAI/USDC (despite the former being worth more than the latter), e.g. 0.1 MATIC shows lower match than 0.1 DAI
A number of things seem to be broken when calculating cart estimates, possibly due to the same underlying cause: