How to interpret perplexing qcovhsp result in blastx

Question

[samuel.himes]

I am using blastx to blast nucleotide sequences against a custom protein database.

Here is a snippet of my output. These are all the HSPs for seq1 against subject1

qseqid	sseqid	qlen	qstart	qend	sstart	send	qcovs	qcovhsp	pident
seq1	subject1	141	139	2	979	1024	98	33	86
seq1	subject1	141	48	74	44	52	98	6	55
seq1	subject1	141	95	112	650	655	98	4	50

I'm confused by the value of qcovhsp. Consider the first row. qlen = 141, qstart=139, and qend=2. I would think that the qcovhsp would be around 97%

(139 - 2) / 141 = .971

Why then is qcovhsp 33? In fact, scrolling through my results I don't see any qcovhsp values > 33. Does qcovhsp behave differently in blastx than it does in blastn?

FYI this was the command I used to run blast.

 blastx -task blastx \
        -db my_protein_db \
        -query my_fasta.fasta \
        -query_gencode 11 \
        -outfmt "6 qseqid sseqid qlen qstart qend sstart send pident qcovhsp qcovs qseq" \
        > output.txt