It seemed obvious that -S has to be best, and a cursory check of Zyzzyva, showing that AELNST produces more bingos than any other six letter leave here confirms that.

But wait...

AELNST? does produce 36 bingos, but it only produces 15 different bingo racks. A total of 71 tiles in the remaining pool bingo with it.

The right play is -N, leaving AELSST?

AELSST? produces 31 bingos from 19 different bingo racks. It works with 80 of the remaining tiles.

Didn't do as extensive an analysis on the eights, but AELSST makes 105 eights, whereas AELNST makes only 93.

Great one, thanks for sharing it. Sorry if I goofed on any of my tile counts.

Quackle says L. I think I agree.

My Quackle is showing N > S > L, with 2-4 points separation. Two-play seems to amplify the differences. Hmm, I wonder why our results are so different.

Two-PLY, that is.

That's what my long 2-ply sims have, -N about 2 better than -S and 4 better than -L. Surprised the differences are even that large. (I picked -S before the sim.)

-N

Well, I guessed the second best exchange. Unless you encounter these, I think finding them in a short amount of time is basically impossible.
It's pretty easy if you are willing to waste 6 minutes on the opening rack, but I'm generally okay with making the wrong play here.
One of the advantages to studying by stem probability is that you should know the answer to this rack if you go deep enough in stems. To me, the costs outweigh the benefits, but herein lies the main benefit.
I have a similar opening rack in my book.

It's nice when even the worst choice is good.

My first choice was L, which a long 4-play Quackle sim believes is the worst choice.

I think like most people, I assumed LATENS? would yield the most bingos, followed by SANEST?, and then TASSEL?. But I chose exchange L because I assigned some value to not strengthening the unseen pool by adding an S to it. In hindsight, I wonder if this kind of logic is even worth taking into account this early in the game with 86 tiles still in the bag?

But yeah, I definitely didn't predict that TASSEL? would be the strongest leave.