Here are some lookup table to check Henry Wan is useful or not (if no token ignore/cancel effect).
Strategy: I always reveal the same number of tokens. When I select the number, I choose for it to maximize the mean of success. Since I cannot choose to go ahead or stop when I fail, this strategy looks reasonable for me. But this does not consider the robustness of success.
Expectation table: the table shows the required number of nonsymbol(non,,,,) choas tokens to achieve the given success.
0 
1 
2 
3 
4 
5 
6 
7 
8 
9 


0.50  1  2  3  4  5  6  7  7  8  9 
0.75  1  3  4  6  7  9  11  12  14  15 
1.00  1  4  6  8  11  13  15  17  20  22 
1.25  2  5  7  10  13  16  19  22  25  28 
1.50  2  5  9  13  16  20  24  27  31  34 
1.75  2  7  11  15  19  24  28  33  37  41 
2.00  2  8  13  18  22  27  32  37  42  47 
2.50  3  10  16  22  29  35  41  48  54   
3.00  3  12  19  27  35  42  50       
4.00  4  16  26  37  47  57         
5.00  5  20  33  46  59           
Row(0.5~5): expected success, Column(0~9): the number of tokens. (: required number exceeds the total tokens in game (44+20)).
Usage: Find your chaos bag (close one), and then move left (by sealing tokens) and/or down (by adding bless/curse) until you reach your goal. For example, standard NotZ (5 symbols / 11 nonsymbols) may exist between 0.75 ~ 1.00. For avg 1(), it is necessary to add 2 blesses/curses tokens or seal 1 token.
Revealed number selection: In my strategy, I select the revealed number based on chaos tokens for maximizing average. Here is my selection number (which is maximizing average).
The trial number is expressed as simple formula: (reveal) = (# of total + 1) / (# of symbols + 1) (round down). If no remainder, you may choose 1 less value.
For example, standard NotZ (5 symbols / 16 toal) case, I'll reveal 2 (17/6=2.xx) tokens for maximizing (avg: 0.92). If I add 1 tokens (5 symbols / 17 total), I'll revel 3 (18/6=3) or 2 tokens (avg: 0.97); success rate is 48% for 2 revealed, 32% for 3 revealed.