AlphaBoom

Success Rate Section#

Success Rate Calculation @ Uma Musume - My Blog

In summary, the original author believes that each probability should be 18%, with a fixed 10% absence. It was later corrected to a 1/11 probability of not appearing, with the rest evenly distributed. Additionally, the inherent calculation method does not simply add 20 to the success rate.

Training Value Calculation#

Reference: Google Doc

The formula is as follows:

Status Increase Value = (① Base Increase Value) × (② Growth Correction for Training Uma Musume) × (③ Friendship Bonus Correction) × (④ Motivation Effect Correction) × (⑤ Training Effect Correction) × (⑥ Number of People Correction)

However, it was verified in Grand Master training that if there are currently no support cards, the base growth rate may be penalized, as the actual value is lower than the calculated result from the formula. It is predicted that the growth rate is halved, for example, 20% becomes 10%.

In addition, the base attributes have a limit of 100, and the content of the Master Cup script bonus will be displayed above the basic values. Whether the bonuses for this type of data conform to the above formula still needs verification; at least for now, it seems that the two are calculated separately.

Master Cup Basic Training Values#

Training level does not affect pt acquisition but does affect the corresponding training values and stamina consumption.

Specific Value Data#

Intelligence and power training have been verified, while other training has only been verified at level one as correct, with other training data extrapolated based on the patterns of power training. It is said that the Red Goddess explosion corresponds to training level 6, and the value for level 6 can be derived using the same pattern (level 6 has not been tested for verification).

speed_train:
  level1:
    speed: 10
    power: 3
    skill_point: 5
    vital: -19
  level2:
    speed: 11
    power: 3
    skill_point: 5
    vital: -20
  level3:
    speed: 12
    power: 3
    skill_point: 5
    vital: -21
  level4:
    speed: 13
    power: 4
    skill_point: 5
    vital: -23
  level5:
    speed: 14
    power: 5
    skill_point: 5
    vital: -25
stamina_train:
  level1:
    stamina: 8
    guts: 6
    skill_point: 5
    vital: -20
  level2:
    stamina: 9
    guts: 6
    skill_point: 5
    vital: -21
  level3:
    stamina: 10
    guts: 6
    skill_point: 5
    vital: -22
  level4:
    stamina: 11
    guts: 7
    skill_point: 5
    vital: -24
  level5:
    stamina: 12
    guts: 8
    skill_point: 5
    vital: -26
power_train:
  level1:
    power: 9
    stamina: 4
    skill_point: 5
    vital: -20
  level2:
    power: 10
    stamina: 4
    skill_point: 5
    vital: -21
  level3:
    power: 11
    stamina: 4
    skill_point: 5
    vital: -22
  level4:
    power: 12
    stamina: 5
    skill_point: 5
    vital: -24
  level5:
    power: 13
    stamina: 6
    skill_point: 5
    vital: -26
guts_train:
  level1:
    guts: 9
    speed: 2
    power: 3
    skill_point: 5
    vital: -20
  level2:
    guts: 10
    speed: 2
    power: 3
    skill_point: 5
    vital: -21
  level3:
    guts: 11
    speed: 2
    power: 3
    skill_point: 5
    vital: -22
  level4:
    guts: 12
    speed: 3
    power: 4
    skill_point: 5
    vital: -24
  level5:
    guts: 13
    speed: 4
    power: 5
    skill_point: 5
    vital: -26
wiz_train:
  level1:
    wiz: 8
    speed: 2
    skill_point: 5
    vital: 5
  level2:
    wiz: 9
    speed: 2
    skill_point: 5
    vital: 5
  level3:
    wiz: 10
    speed: 2
    skill_point: 5
    vital: 5
  level4:
    wiz: 11
    speed: 3
    skill_point: 5
    vital: 5
  level5:
    wiz: 12
    speed: 4
    skill_point: 5
    vital: 5

Failure Rate Calculation#

Based on the obtained failure rates, there are several hypotheses:

• It may be a sigmoid function with an additional failure rate for stamina below 10.

• It may be a Bezier curve fitting.

• After playing with ChatGPT, several curves were found for fitting, and it was discovered that the Gaussian function performs very well, and the Gaussian function controls the graph more intuitively.

  Copy$$f(x)=ae^{-\frac{(x-b)^2}{2c^2}}$$

Failure Rate Statistical Data#

The recorded data is stamina, with the first column being the failure rates for speed, stamina, power, guts, and intelligence, and the second column being the corresponding training levels.

57
-

49
6 5
5 5
6 5
8 5
0 5
44
12 1
10 1
13 4
13 1
0 2
41
17 2
15 2
18 5
19 2
0 3
40
18 1
17 1
20 4
20 1
0 1
37
24 1
22 1
25 3
26 1
0 1
36
26 1
24 1
28 5
28 1
0 2
32
34 1
32 1
34 2
37 1
0  1
31
39 5
37 5
39 5
42 5
3 5
31
37 1
34 1
38 4
39 1
2 1
30
39  1
37  1
39  2
41  1
5   1
30
39 1
37 1
40 4
41 1
4 1
26
49 1
46 1
51 5
51 1
12 2
21
62 1
59 1
64 5
64 1
22 2
19
67 1
65 1
68 3
70 1
27 1
16
76 1
73 1
79 5
79 1
34 2
15
79 1
76 1
80 3
82 1
37 1
10
95 1
92 1
96 3
98 1
50 1
5
99 5
99 5
99 5
99 5
66 5
0
-
81 2

Skill Related#

By examining the database, several useful fields can be found. Here are the key fields listed briefly, with the fields in red explained in the sections below.

Skill Trigger Conditions#

Related to precondition_x and condition_x. Here are some common examples; for more complete data, you can refer to GameTora.

An example of the content in the conditions is as follows:

distance_type==4&phase_random==1&bashin_diff_behind>=1&order==1

The structure is such that the expression & indicates "and" and @ indicates "or".

Skill Types#

ability_type_x_x and ability_value_usage_x_x

The following descriptions are based on analysis through existing skills and available numerical information.

Target Types#

target_type_x_x and target_value_x_x analysis.