I've been reading a few posts, and I can see that a lot of
people think that
1. You need a large betting spread to win big at SP21, and
2. It will give an unacceptable ROR.

This is not true. (An explanation follows below).

In my book, I use proportional betting.
Proportional betting will always give the highest SCORE game.
Unfortunately, in Blackjack, you can't play that way. That's why nobody does. But I always do in SP21.

For example, if you were playing a Blackjack game with HE 0.40%, your advantages at each count would be:
(very roughly, to the theorists out there. The advantages are very rough estimates, as you can see. I assume that adv increases by 0.5% per TC increment which is an average, and a very bad one at that. )

TC adv A's_bets B's_bets
0 -0.4% $0 $0
1 +0.1% $100 $20
2 0.6% $200 $120
3 1.1% $300 $220
4 1.6% $400 $320
5 2.1% $400 $420
6 2.6% $400 $520
7 3.1% $400 $620
8 3.6% $400 $720
9 4.1% $400 $820
10 4.6% $400 $920

(Sorry about the formatting; I spent ages making a nice table, but it doesn't come out that way after you click "Post". There should be 4 columns: TC, adv, A's betting strategy, and B's betting strategy)
Let's call A a typical BJ player, spreading from $100 to $400.
Let's call B a proportional bettor (my style of playing. He is betting proportional to his advantage. (For simplicity, we are not
doing inversely to variance, as variance doesn't change that much.)

Let's say the penetration is 80%, and both players are only playing the advantageous hands (WIWOWIWO...).
We'll approximate the variance as 1.3, increasing by 0.01 for each increase of +1 in the true count (playing with indices, obviously).

The results are:
(note these results assume a bet of $0 on advantageous hands, and these "non-playing" hands are counted.)
w = win rate per hand
SD = std dev per hand
ROR = risk of ruin with a 30K bankroll (calculated using George C's formula)

Player w SD ROR
_____________________________________________
A $0.88 $164 14%
B $0.96 $156 9%

(formatting didn't come out; there should be 4 columns: player, w, sd, and ROR.)

************************************************** ******
Notice how player B has more than double the spread of player A, but only two-thirds of his ROR!
************************************************** *******

(Once again, the the theorists out there, these calcs are
super rough. No simulations done, just a spreadsheet using the approximate frequency distribution of an 80% pen shoe).

So player A has a lower win rate, a higher std dev, and a much higher ROR than player B.
But Player B has more than double player A's spread!!!