Splitting 10,10 v. 9(!), negative expectation?

In bjmath.com there are some very useful tables for expected values from Cacarulo. However there are a couple of them that puzzle me.
Why splitting 10,10 vs. 9 has a negative expectation (-0.29567) since, from the same tables, hitting a ten vs. 9 (6,4 or 7,3 or 8,2) has a positive one (roughly 0.116)?

This is just a guess and off the top of my head, but consider the hand compositions. With a 10,10 vs a nine, you have two big cards gone, while with the two-card tens, you have two small cards gone and the big cards still left in the deck, giving you two more shots at a twenty that you won't have with the split tens. In addition, you can double down on those two card hands and you can't double down on that single ten. In fact, you aren't going to get a doubling hand at all with one ten in the hand. When you factor in those things, you can see that the expectation won't be the same with a one-card split ten as opposed to a two-card ten.

Bootlegger is partially correct about hand composition. The other thing is that you're thinking of SPL1. The tables are SPL3.

If you splitted only once then your assumptions would be correct:

Splitting =  22.48167394228079% (SPL1)

Splitting =  -9.93442303338956% (SPL2)

Splitting = -29.56647317585910% (SPL3)

Here is my take on this question.

If you stand on your two tens against the 9, according to Professional Blackjack, page 332, your expected value is 0.756...If you split, you use the hit column and get an expected value of 0.117 for each hand...It looks to me like they have assumed that you might get 4 hands which would give you a total of 0.117 X 4 = 0.468.

Since you could have gotten 0.756 by standing, but by splitting, you could get at the most, 0.468, you have thrown away, 756 - 468 = 0.288.

That's pretty close to the -0.295 figure that they quote.

...but not the way you're talking. I can't say anything more... I just play in Europe. This might help you figure it out why I made such a question. ;-)