From bd99cfaa6d1e7ef961c40c1b29cc7df78652374b Mon Sep 17 00:00:00 2001 From: Branden J Brown Date: Sat, 23 May 2026 16:18:24 -0400 Subject: [PATCH] zenno/doc/frbm: update with charts &c. --- zenno/package-lock.json | 100 ++++- zenno/package.json | 3 +- zenno/src/lib/Skill.svelte | 27 ++ zenno/src/lib/data/skill.ts | 173 +++++++++ zenno/src/lib/prob.ts | 9 + zenno/src/lib/race.ts | 34 ++ zenno/src/routes/doc/frbm/+page.svelte | 503 ++++++++++++++++--------- 7 files changed, 678 insertions(+), 171 deletions(-) create mode 100644 zenno/src/lib/Skill.svelte create mode 100644 zenno/src/lib/data/skill.ts create mode 100644 zenno/src/lib/prob.ts diff --git a/zenno/package-lock.json b/zenno/package-lock.json index f244eea..6404c6c 100644 --- a/zenno/package-lock.json +++ b/zenno/package-lock.json @@ -8,7 +8,8 @@ "name": "zenno", "version": "0.0.1", "dependencies": { - "@observablehq/plot": "^0.6.17" + "@observablehq/plot": "^0.6.17", + "mathjs": "^15.2.0" }, "devDependencies": { "@eslint/compat": "^2.1.0", @@ -40,6 +41,15 @@ "vitest-browser-svelte": "^2.1.1" } }, + "node_modules/@babel/runtime": { + "version": "7.29.2", + "resolved": "https://registry.npmjs.org/@babel/runtime/-/runtime-7.29.2.tgz", + "integrity": "sha512-JiDShH45zKHWyGe4ZNVRrCjBz8Nh9TMmZG1kh4QTK8hCBTWBi8Da+i7s1fJw7/lYpM4ccepSNfqzZ/QvABBi5g==", + "license": "MIT", + "engines": { + "node": ">=6.9.0" + } + }, "node_modules/@blazediff/core": { "version": "1.9.1", "resolved": "https://registry.npmjs.org/@blazediff/core/-/core-1.9.1.tgz", @@ -2416,6 +2426,19 @@ "node": ">= 10" } }, + "node_modules/complex.js": { + "version": "2.4.3", + "resolved": "https://registry.npmjs.org/complex.js/-/complex.js-2.4.3.tgz", + "integrity": "sha512-UrQVSUur14tNX6tiP4y8T4w4FeJAX3bi2cIv0pu/DTLFNxoq7z2Yh83Vfzztj6Px3X/lubqQ9IrPp7Bpn6p4MQ==", + "license": "MIT", + "engines": { + "node": "*" + }, + "funding": { + "type": "github", + "url": "https://github.com/sponsors/rawify" + } + }, "node_modules/convert-source-map": { "version": "2.0.0", "resolved": "https://registry.npmjs.org/convert-source-map/-/convert-source-map-2.0.0.tgz", @@ -2880,6 +2903,12 @@ } } }, + "node_modules/decimal.js": { + "version": "10.6.0", + "resolved": "https://registry.npmjs.org/decimal.js/-/decimal.js-10.6.0.tgz", + "integrity": "sha512-YpgQiITW3JXGntzdUmyUR1V812Hn8T1YVXhCu+wO3OpS4eU9l4YdD3qjyiKdV6mvV29zapkMeD390UVEf2lkUg==", + "license": "MIT" + }, "node_modules/deep-is": { "version": "0.1.4", "resolved": "https://registry.npmjs.org/deep-is/-/deep-is-0.1.4.tgz", @@ -2986,6 +3015,12 @@ "@esbuild/win32-x64": "0.27.7" } }, + "node_modules/escape-latex": { + "version": "1.2.0", + "resolved": "https://registry.npmjs.org/escape-latex/-/escape-latex-1.2.0.tgz", + "integrity": "sha512-nV5aVWW1K0wEiUIEdZ4erkGGH8mDxGyxSeqPzRNtWP7ataw+/olFObw7hujFWlVjNsaDFw5VZ5NzVSIqRgfTiw==", + "license": "MIT" + }, "node_modules/escape-string-regexp": { "version": "4.0.0", "resolved": "https://registry.npmjs.org/escape-string-regexp/-/escape-string-regexp-4.0.0.tgz", @@ -3349,6 +3384,19 @@ "dev": true, "license": "ISC" }, + "node_modules/fraction.js": { + "version": "5.3.4", + "resolved": "https://registry.npmjs.org/fraction.js/-/fraction.js-5.3.4.tgz", + "integrity": "sha512-1X1NTtiJphryn/uLQz3whtY6jK3fTqoE3ohKs0tT+Ujr1W59oopxmoEh7Lu5p6vBaPbgoM0bzveAW4Qi5RyWDQ==", + "license": "MIT", + "engines": { + "node": "*" + }, + "funding": { + "type": "github", + "url": "https://github.com/sponsors/rawify" + } + }, "node_modules/fsevents": { "version": "2.3.2", "resolved": "https://registry.npmjs.org/fsevents/-/fsevents-2.3.2.tgz", @@ -3493,6 +3541,12 @@ "integrity": "sha512-tFLRAygk9NqrRPhJSnNGh7g7oaVWDwR0wKh/GM2LgmPa50Eg4UfyaCO4I8k6EqJHl1/uh2RAD6g06n5ygEnrjQ==", "license": "ISC" }, + "node_modules/javascript-natural-sort": { + "version": "0.7.1", + "resolved": "https://registry.npmjs.org/javascript-natural-sort/-/javascript-natural-sort-0.7.1.tgz", + "integrity": "sha512-nO6jcEfZWQXDhOiBtG2KvKyEptz7RVbpGP4vTD2hLBdmNQSsCiicO2Ioinv6UI4y9ukqnBpy+XZ9H6uLNgJTlw==", + "license": "MIT" + }, "node_modules/jiti": { "version": "2.7.0", "resolved": "https://registry.npmjs.org/jiti/-/jiti-2.7.0.tgz", @@ -3869,6 +3923,29 @@ "@jridgewell/sourcemap-codec": "^1.5.5" } }, + "node_modules/mathjs": { + "version": "15.2.0", + "resolved": "https://registry.npmjs.org/mathjs/-/mathjs-15.2.0.tgz", + "integrity": "sha512-UAQzSVob9rNLdGpqcFMYmSu9dkuLYy7Lr2hBEQS5SHQdknA9VppJz3cy2KkpMzTODunad6V6cNv+5kOLsePLow==", + "license": "Apache-2.0", + "dependencies": { + "@babel/runtime": "^7.26.10", + "complex.js": "^2.2.5", + "decimal.js": "^10.4.3", + "escape-latex": "^1.2.0", + "fraction.js": "^5.2.1", + "javascript-natural-sort": "^0.7.1", + "seedrandom": "^3.0.5", + "tiny-emitter": "^2.1.0", + "typed-function": "^4.2.1" + }, + "bin": { + "mathjs": "bin/cli.js" + }, + "engines": { + "node": ">= 18" + } + }, "node_modules/minimatch": { "version": "10.2.5", "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-10.2.5.tgz", @@ -4448,6 +4525,12 @@ "integrity": "sha512-YZo3K82SD7Riyi0E1EQPojLz7kpepnSQI9IyPbHHg1XXXevb5dJI7tpyN2ADxGcQbHG7vcyRHk0cbwqcQriUtg==", "license": "MIT" }, + "node_modules/seedrandom": { + "version": "3.0.5", + "resolved": "https://registry.npmjs.org/seedrandom/-/seedrandom-3.0.5.tgz", + "integrity": "sha512-8OwmbklUNzwezjGInmZ+2clQmExQPvomqjL7LFqOYqtmuxRgQYqOD3mHaU+MvZn5FLUeVxVfQjwLZW/n/JFuqg==", + "license": "MIT" + }, "node_modules/semver": { "version": "7.8.0", "resolved": "https://registry.npmjs.org/semver/-/semver-7.8.0.tgz", @@ -4689,6 +4772,12 @@ "url": "https://opencollective.com/webpack" } }, + "node_modules/tiny-emitter": { + "version": "2.1.0", + "resolved": "https://registry.npmjs.org/tiny-emitter/-/tiny-emitter-2.1.0.tgz", + "integrity": "sha512-NB6Dk1A9xgQPMoGqC5CVXn123gWyte215ONT5Pp5a0yt4nlEoO1ZWeCwpncaekPHXO60i47ihFnZPiRPjRMq4Q==", + "license": "MIT" + }, "node_modules/tinybench": { "version": "2.9.0", "resolved": "https://registry.npmjs.org/tinybench/-/tinybench-2.9.0.tgz", @@ -4769,6 +4858,15 @@ "node": ">= 0.8.0" } }, + "node_modules/typed-function": { + "version": "4.2.2", + "resolved": "https://registry.npmjs.org/typed-function/-/typed-function-4.2.2.tgz", + "integrity": "sha512-VwaXim9Gp1bngi/q3do8hgttYn2uC3MoT/gfuMWylnj1IeZBUAyPddHZlo1K05BDoj8DYPpMdiHqH1dDYdJf2A==", + "license": "MIT", + "engines": { + "node": ">= 18" + } + }, "node_modules/typescript": { "version": "5.9.3", "resolved": "https://registry.npmjs.org/typescript/-/typescript-5.9.3.tgz", diff --git a/zenno/package.json b/zenno/package.json index 29bee09..37d50dc 100644 --- a/zenno/package.json +++ b/zenno/package.json @@ -45,6 +45,7 @@ "vitest-browser-svelte": "^2.1.1" }, "dependencies": { - "@observablehq/plot": "^0.6.17" + "@observablehq/plot": "^0.6.17", + "mathjs": "^15.2.0" } } diff --git a/zenno/src/lib/Skill.svelte b/zenno/src/lib/Skill.svelte new file mode 100644 index 0000000..aab89a6 --- /dev/null +++ b/zenno/src/lib/Skill.svelte @@ -0,0 +1,27 @@ + + +{s.name} diff --git a/zenno/src/lib/data/skill.ts b/zenno/src/lib/data/skill.ts new file mode 100644 index 0000000..97ac3b7 --- /dev/null +++ b/zenno/src/lib/data/skill.ts @@ -0,0 +1,173 @@ +import skillGlobal from '../../../../global/skill.json' +import groupGlobal from '../../../../global/skill-group.json' + +/** + * Skill data. + */ +export interface Skill { + /** + * Skill ID. + */ + skill_id: number; + /** + * Regional skill name. + */ + name: string; + /** + * Regional skil description. + */ + description: string; + /** + * Skill group ID. + */ + group: number; + /** + * Skill rarity. 3-5 are uniques for various star levels. + */ + rarity: 1 | 2 | 3 | 4 | 5; + /** + * Upgrade position within the skill's group. + * -1 is for negative (purple) skills. + */ + group_rate: 1 | 2 | 3 | -1; + /** + * Grade value, or the amount of rating gained for having the skill with + * appropriate aptitude. + */ + grade_value?: number; + /** + * Whether the skill requires a wit check. + */ + wit_check: boolean; + /** + * Conditions and results of skill activation. + */ + activations: Activation[]; + /** + * Name of the Uma which owns this skill as a unique, if applicable. + */ + unique_owner?: string; + /** + * SP cost to purchase the skill, if applicable. + */ + sp_cost?: number; + /** + * Skill icon ID. + */ + icon_id: number; +} + +/** + * Conditions and results of skill activation. + */ +export interface Activation { + /** + * Precondition which must be satisfied before the condition is checked. + */ + precondition?: string; + /** + * Activation conditions. + */ + condition: string; + /** + * Skill duration in ten thousandths of a second. + * Generally undefined for activations which only affect HP. + */ + duration?: number; + /** + * Special skill duration scaling mode. + */ + dur_scale: 1 | 2 | 3 | 4 | 5 | 7; + /** + * Skill cooldown in ten thousandths of a second. + * A value of 5000000 indicates that the cooldown is forever. + * Generally undefined for passive skills. + */ + cooldown?: number; + /** + * Results applied when the skill's conditions are met. + */ + abilities: Ability[]; +} + +/** + * Effects applied when a skill activates. + */ +export interface Ability { + /** + * Race mechanic affected by the ability. + */ + type: 1 | 2 | 3 | 4 | 5 | 6 | 8 | 9 | 10 | 13 | 21 | 27 | 28 | 31 | 35; + /** + * Special scaling type of the skill value. + */ + value_usage: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 13 | 14 | 19 | 20 | 22 | 23 | 24 | 25; + /** + * Amount that the skill modifies the race mechanic in ten thousandths of + * whatever is the appropriate unit. + */ + value: number; + /** + * Selector for horses targeted by the ability. + */ + target: 1 | 2 | 4 | 7 | 9 | 10 | 11 | 18 | 19 | 20 | 21 | 22 | 23; + /** + * Argument value for the ability target, when appropriate. + */ + target_value?: number; +} + +/** + * Skill groups. + * Skills in a skill group replace each other when purchased. + * + * As a special case, horsegen lists both unique skills and their inherited + * versions in the skill groups for both. + */ +export interface SkillGroup { + /** + * Skill group ID. + */ + skill_group: number; + /** + * Base skill in the skill group, if any. + * Either a common (white) skill or an Uma's own unique. + * + * Some skill groups, e.g. for G1 Averseness, have no base skill. + */ + skill1?: number; + /** + * First upgraded version of a skill, if any. + * A rare (gold) skill, double circle skill, or an inherited unique skill. + */ + skill2?: number; + /** + * Highest upgraded version of a skill, if any. + * Gold version of a skill with a double circle version. + */ + skill3?: number; + /** + * Negative (purple) version of a skill, if any. + */ + skill_bad?: number; +} + +export const skills = { + global: skillGlobal as Skill[], +} as const; + +export const skillGroups = { + global: groupGlobal as SkillGroup[], +} as const; + +export const ZERO_SKILL: Readonly = { + skill_id: 0, + name: "invalid skill", + description: "an invalid skill was specified", + group: 0, + rarity: 1, + group_rate: 1, + wit_check: false, + activations: [], + icon_id: 0, +} as const; diff --git a/zenno/src/lib/prob.ts b/zenno/src/lib/prob.ts new file mode 100644 index 0000000..80b6584 --- /dev/null +++ b/zenno/src/lib/prob.ts @@ -0,0 +1,9 @@ +import * as math from "mathjs"; + +export function binomPMF(p: number, n: number, k: number): number { + // Operate in log domain for precision. + const lc = math.lgamma(n+1) - math.lgamma(k+1) - math.lgamma(n-k+1); + const lpk = k * math.log(p); + const lr = (n - k) * math.log(1 - p); + return math.exp(lc + lpk + lr); +} diff --git a/zenno/src/lib/race.ts b/zenno/src/lib/race.ts index d5f0793..859176c 100644 --- a/zenno/src/lib/race.ts +++ b/zenno/src/lib/race.ts @@ -1,6 +1,8 @@ // Umamusume race mechanics adapted from KuromiAK's doc: // https://docs.google.com/document/d/15VzW9W2tXBBTibBRbZ8IVpW6HaMX8H0RP03kq6Az7Xg/edit?usp=sharing +import { binomPMF } from "./prob"; + /** * Fundamental stats of umas. */ @@ -246,6 +248,17 @@ export function spotStruggleDuration(gutsStat: number, frontAptitude: AptitudeLe return Math.sqrt(700 * gutsStat) * 0.012 * strategyProficiencyMod[frontAptitude]; } +/** + * Calculate the speed modifier for running uphill. + * Contrary to the race mechanics document, this is expressed as a negative number. + * @param powerStat Final power stat + * @param slopePer Slope percentage, generally one of 0.5, 1.0, 1.5, or 2.0 + * @returns Speed modifier for running uphill, a negative value + */ +export function uphillMod(powerStat: number, slopePer: number): number { + return slopePer * -200/powerStat; +} + /** * Calculate the forward speed boost given when moving lanewise while a skill * that grants a lane change speed boost is active. @@ -256,6 +269,18 @@ export function moveLaneModifier(powerStat: number): number { return Math.sqrt(0.0002 * powerStat); } +/** + * Calculate the probability of n of N skills activating. + * @param baseWit Base wit stat + * @param N Number of skills available, default 1 + * @param n Number of skills activating, default 1 + * @returns Probability of exactly n skills out of N passing wit checks + */ +export function skillWitCheck(baseWit: number, N?: number, n?: number): number { + const p = Math.max(0.2, 1 - 90/baseWit); + return binomPMF(p, N ?? 1, n ?? 1); +} + /** * Calculate a skill's actual duration scaled to race length. * @param baseDur Skill's listed duration in s @@ -292,3 +317,12 @@ export function speedGain(speedBonus: number, dur: number, accel: number | null, // speedBonus*(dur-accelTime) + speedBonus*accelTime/2 + speedBonus*decelTime/2 return speedBonus * (dur + 0.5 * (decelTime - accelTime)); } + +/** + * Calculate the chance to enter downhill accel mode each second while running downhill. + * @param witStat Final wit stat, including style aptitude modifier + * @returns Probability each eligible tick to enter downhill accel mode + */ +export function downhillAccelEnterChance(witStat: number): number { + return witStat * 0.0004; +} diff --git a/zenno/src/routes/doc/frbm/+page.svelte b/zenno/src/routes/doc/frbm/+page.svelte index 7705b3f..87d7a7d 100644 --- a/zenno/src/routes/doc/frbm/+page.svelte +++ b/zenno/src/routes/doc/frbm/+page.svelte @@ -1,6 +1,38 @@
@@ -12,9 +44,10 @@

This document is advanced material. The target audience intends to win Champions Meet Group A Finals and either wants to use front runners to do it or wants to understand what front runners they need to beat. This is meant for players who are already - strong at training: players who can take a target stat line and skill set and turn it into a horse. This is about what those - stat lines and skill sets should be, along with why. + strong at training: players who can take a target stat line and skill set and turn it into a horse. This document is about the + mechanics that determine what those stat lines and skill sets should be.

+ About Me

About three weeks after Global launched, my friend told me to get a job, so I sent him a screenshot of me clicking the install @@ -39,6 +72,7 @@ rationalize running triple fronts for every CM even though it's not actually very good and most of my favorite horses are late surgers.

+ Race Mechanics

Very quick gloss of race fundamentals. Races are divided into four phases: early race, mid race, late race, and last spurt @@ -49,68 +83,78 @@

The numeric value of acceleration depends on the Power stat, dueling, surface aptitude, uphills, race phase, running style. At the start of early race, horses accelerate from 3 m/s to the early race base target speed, which varies by race - distance and running style. At the start of late race, if they have enough HP remaining for their last spurt, horses - accelerate from the mid race base target speed to their spurt speed, which varies by speed stat, distance aptitude, race - distance, running style, and guts stat, in decreasing order of effect. "Last spurt" and "last spurt phase" are different and - unrelated things; the latter is only used in the condition for Homestretch Haste. + distance and running style but is generally on the order of 20 m/s. At the start of late race, if they have enough HP remaining for their last spurt, horses + accelerate from the mid race base target speed to their spurt speed, which varies by speed stat, distance aptitude, running style, race + distance, and guts stat, in decreasing order of effect. "Last spurt" and "last spurt phase" are different and + unrelated things; the latter is only used in the condition for .

Speed skills add a flat amount of target speed, generally +0.15 m/s for white skills, +0.25 m/s for double circle skills and some inherited uniques, +0.35 m/s for gold skills and most speed uniques, and +0.45 m/s for a handful of speed uniques. Accel - skills similarly add a flat amount of acceleration, either +0.1 or +0.2 m/s² for white skills and inherited uniques, or +0.3 + skills similarly add a flat amount of acceleration, typically +0.1 or +0.2 m/s² for white skills and inherited uniques, or +0.3 or +0.4 m/s² for gold skills and uniques.

-

- Generally speaking, competitive races are won by the horse who gets the most acceleration at the start of late race. The - consistency of front runner accel skills is what makes it a viable running style. -

+ Runaway

- The skill Runaway converts front runners into the Great Escape running style. However, no player has ever uttered + The skill converts front runners into the Great Escape running style. However, no player has ever uttered the words "Great Escape" when talking about Umamusume, presumably because Runaway is a much cooler name. + ("Great Escape" is a direct translation of Japanese 大逃げ oonige, whereas "Front Runner" is a more liberal localization of 逃げ nige that technically just means "escape.")

- Runaways are still front runners for most purposes, the main difference just being different base target speeds per phase. + Runaways are still front runners for all purposes. + The main difference is just different numbers for things like base speed and acceleration, stamina to HP conversion, and distance thresholds for running modes. Other mechanics that are specific to front runners also apply to runaways.

- Spot Struggle + + Win Conditions

- For each of runaways and non-runaways, there is at most one spot struggle per race. Runaways will not spot struggle with - non-runaways, nor vice-versa. When a spot struggle triggers, all front runnners of that type within range participate; I've - had a horse join while in 6th. + On Global today, competitive horses usually have stat lines that are pretty similar to each other. + Races, therefore, are more often won by skills – typically acceleration skills that activate at the start of late race. + Front runners have strong options. +

+
    +
  • + , sometimes called Rod, is the second best skill in the game. + Because only the horse in first place gets it, everything about training front runners becomes a matter of being in front at the start of late race, true to name. +
    • +
  • + On long distance tracks, takes that role instead. + The front two horses get it, which opens the opportunity for multi-front builds using / – + especially because VC tracks aren't subject to the final corner spread that makes those skills worse on sprints and miles – + but otherwise the function is the same. +
    • +
  • + On those sprints where Angling is dead, the front-specific options include (VPP, or Pasta) and (HCreek), + It takes both of them to equal Angling, so such sprints may be better served gambling on , , and possibly instead. + Front runners are especially strong on sprints for other reasons anyway. +
    • +
+

+ is the best skill in the game. + Unfortunately, for the most part, it's bad on front runners; generally not a win condition. + Activating NSM requires not being in first, which means whoever was used Angling and is pulling away from you before you accumulate the blocked time to activate it. + Again, VC tracks may be an exception if you specifically build for it. +

+ + Pace Down Mode +

+ During the first 41.67% of the race, position keep is busy arranging each running style into their respective packs. + The primary mechanism for this is pace down mode (PDM), which activates whenever a horse gets what their style defines as too close to first place.

- Spot struggle provides a target speed bonus that scales with the guts stat. If it isn't cut short, which will approximately - never happen, its duration scales with the guts stat. Unlike skills, its duration does not scale with race distance. + Watch a MANT late surger with 1000+ power and wit in a daily legend race. + As long as they don't get blocked, they should slide forward throughout the early race. + Then, around when they reach the pace chaser pack, they'll suddenly start moonwalking back to the rest of the late surgers, often near the back of the group. + That's PDM.

- Spot struggle also greatly increases HP consumption. For normal front runners, the rate is slightly less than Rushed. For - runaways, it's more than double Rushed. (This is the reason people say you can't get enough stamina for runaways on Global.) - Actually getting Rushed during spot struggle dramatically increases HP consumption, much more than just adding them together; - red-light green-light pretty much guarantees that horse won't spurt. + On lesser running styles, early race and sometimes mid race speed skills are effectively converted from distance gain into HP conservation via PDM. + The thing that really makes front runners good is that they don't have to worry about that – they aren't subject to PDM at all. + Their mid race speed skills always gain distance.

- Position Keep -

Position Keep is the process by which pace chasers don't pass front runners in the mid race.

-

- During Position Keep, the frontmost front runner of each type uses speed up mode to try to stay at least 4.5m (a length - is 2.5m) ahead of second place, and other front runners use overtake mode to try to become the frontmost. Both of these modes - take wit checks to enter and apply a target speed bonus for their duration; overtake mode is slightly better than speed up mode. -

-

- Lesser running styles instead have pace up mode and pace down mode. Whether horses enter these is based on their - distance from the frontmost horse. Pace up also requires a wit check, but PDM doesn't – if a non-front horse gets too close to - the frontmost runner in the first 41.67% of the race, they are guaranteed to switch into PDM, which converts distance into HP. -

-

- Watch a MANT late surger with 1000+ power and wit in a solo practice match. When you see her advance forward early but then - moonwalk for two seconds to the back of the pack, that's PDM. -

-

- Converting distance into HP is quite a lot worse than just having enough HP. The fact that front runners don't have to worry - about PDM is one of their major strengths. In particular, it means that early race speed skills always gain distance for front - runners, which is not the case for the inferior styles. -

- Skill Timing + + Skill Timing

Thought experiment.

Picture two cars driving on a straight freeway, both at exactly 59 mph because I am American, adjacent lanes, keeping exactly @@ -143,37 +187,163 @@ This thought experiment shows that speed skills are actually more valuable before late race than during it. Thus, front runners not having to worry about PDM is even more of an advantage.

- Skill Stacking + + Gate Skills

- In a void, the fact that skill effects stack doesn't change the total distance you get from them. However, getting multiple to - activate at the same time drastically improves a front runner's ability to overtake. + Gate skills are (GW), (TTL), and (Conc), as well as all green skills including . + These skills activate the moment the race starts.

- The practical consequence of this is that Tail Held High and, for multi-front builds, Ramp Up are extremely good skills - for front runners. THH has a long duration and will pretty much always trigger off another speed skill, pushing your horse far forward. - Ramp activates upon overtake mid-race, which can turn a lucky order change into a full pass. + GW is an absolutely mandatory skill for all front runners. + Even runaway blockers should have it, otherwise they will be passed by the normal fronts they're trying to block. + It requires three other gate skills, which should be active greens to avoid overreliance on wit checks. + For reference, the chart below shows proc chances of one of one, one of two, or two of two skills with wit checks.

- Duels +
+ +

- Aside from spot struggle, the other mechanic that tries to make guts a stat that matters is dueling. This requires two horses - on the final straight to sustain a certain distance both along and across the track, and to be close in actual speed. There is - some evidence that three horses might be able to enter a single duel, although I am not certain this is confirmed. + TTL must be combined with GW if they want any chance of being first out of early race. + Since the main source of it is the Mihono Bourbon Wit SSR from the first Halloween event, VBourbon can suffice with its white version and get to the front with her unique instead. + (Her other option is the Twin Turbo SSR that does generate a lot of stats but requires winning three 50/50s to get the gold skill.)

- Duels give a substantial bonus to target speed, a gold speed skill's worth at only 500 guts, for the entire duration of the - final straight. (Also a bit of acceleration, but this rarely matters.) + Conc is less critical. + It's worth taking on horses who have it, but it isn't worth using support card slots just to get it. + On the other hand, its white version is bad; its only real use is as a backup gate skill for GW when you don't have enough greens available. +

+ + Spot Struggle +

+ For each of runaways and non-runaways, there is at most one spot struggle per race. Runaways will not spot struggle with + non-runaways, nor vice-versa. When a spot struggle triggers, all front runnners of that type within range participate; I've + had a horse join while in 6th a couple times.

- Unfortunately, winning front runners almost never get duels. After Angling, no one else will be nearly close enough to - trigger; other fronts are left behind, and other styles are still catching up. On tracks like Tokyo 1600 where Angling is - right before the final straight, the problem instead becomes width, since horses spread out on the final corner. + Spot struggle provides a target speed bonus that scales with the guts stat. If it isn't cut short, which will approximately + never happen, its duration also scales with the guts stat. Unlike skills, its duration does not scale with race distance. +

+
+ + +
+

+ Spot struggle also greatly increases HP consumption. + For normal front runners, the rate is slightly less than Rushed. + For runaways, it's more than double Rushed. (This is the reason people say you can't get enough stamina for runaways on Global.) + Actually getting Rushed during spot struggle dramatically increases HP consumption, much more than just adding them together; red-light green-light pretty much guarantees that horse won't spurt. +

+

+ In medium+ races, the extra HP consumption is a serious consideration; front runners need more stamina and recoveries than other styles. + At 1600m and shorter, the fact that Spot Struggle doesn't scale with race distance means that it can be worth multiple gold speed skills in total distance gained. + See the mechanical speed calculator for precise analysis. +

+ + Lane Combo +

+ While under the influence of a skill that increases lane movement speed (shoe icon skills), and while actively changing lanes (i.e. moving sideways), horses gain a (forward) target speed boost that scales with power. + This was a change Global received with the Unity Cup scenario. +

+
+ +
+

+ Front runners have access to the skill , which forces a horse who uses it to move outward to a specific distance from the rail. + DD almost always ends shortly before the horse has finished accelerating to early race speed, so it does not convert the move lane speed modifier into distance.

- For CM12 Aries Cup (Satsuki Sho), I used a guts/wit NSM Suzuka whose goal was to be in third place at late race start, trigger - NSM to stick with the Angling user, and duel with high guts to finish out the race. This plan never materialized. I did get - one win off the duel itself when it enabled Suzuka to pass back a pace chaser in the closest win I had in the whole event; but - one win in eighty races is not a great record. + We get advantage from move lane speed modifier by following DD with or . + DD created an opportunity for those return skills to convert into huge forward speed. + This setup is called lane combo.

+

+ Lane combo is only viable on tracks where early race ends before or at most very early into the first corner. + Since PP and Ignited WIT are phase_random==0 skills, they can activate at the very end of late race. + If there's a corner there, and your horse is still on the outside from DD, you are now physically running a longer distance than those on the inside. + That can more than undo the gain from the lane combo itself. +

+

+ The mechanical speed calculator has an approximation of lane combo's benefit. + A more precise lane combo simulator exists, + but I am not sufficiently confident in my Japanese to try to guide readers through it. + +

+ + Slopes +

+ Different slopes can be of different angles; the SlopePer parameter is positive for uphills and negative for downhills. + SlopePer values that currently exist on tracks include 1, 1.5, and 2, positive or negative. +

+ + Uphills +

+ Running uphill carries a penalty to target speed. + This penalty scales negatively with the power stat; that is, higher power means faster uphill running. + It scales positively with slope angle. +

+
+ +
+

+ Note that surface aptitude does not affect uphill speed, nor power generally. + It only affects acceleration. +

+

+ The practical impact is that steep early- and mid-race hills filter out front runners with low power. + Even with an otherwise perfect build, an 800 power VBourbon is likely to be passed by a 1280 power ( + ) Seiun Sky. +

+ + Downhills +

+ Running downhill allows horses to enter downhill accel mode. + Contrary to its name, downhill accel mode does not affect acceleration at all; + it gives horses a target speed boost that scales with the slope angle, plus lowered HP consumption via a flat multiplier. +

+

+ Entering downhill accel mode requires passing a wit check. + The success rate scales linearly with wit. + Style aptitude does affect the chance to pass the check. + Its duration is random with a geometric distribution; it does not scale with stats. +

+
+ +
+

+ Similar to uphills disproportionately rewarding front runners with higher power, downhills tend to reward high wit. + However, the random elements of downhill accel mode mean that lower wit horses may still keep up on downhills, depending on luck. + Conversely, the HP savings on long downhills can be enough to drop a recovery skill or two on some tracks. +

+ + Section Speed +

+ Each section, each horse gets a random modifier to target speed. + The modifier's range is determined by the wit stat. + (Curiously, the calculation uses both wit as modified by style proficiency and green skills as well as base wit.) +

+

+ Section speed is generally very small; at 1200 wit with style S, it has a range of about -0.15% to 0.5% of race base speed. + At 2000m, that translates to an actual speed range of 19.97 to 20.10 m/s. +

+

+ Unlike anything affected by the speed stat, though, it applies during the early and mid race, where front runners are trying to become frontest runners. + Wit difference alone can +

+ + Phase Speed +

+ Race base speed is multiplied by the strategy–phase coefficient for each horse. + As the name suggests, SPC is different per running style and per race phase. + It's the thing that makes runaways take off in early race, and the thing that makes pace chaser promotion scary in late race (for those not using any of the correct running style). +

+

+ Front runners, and even moreso runaways, have particularly punishing SPC for late race. + This makes sense; if they weren't forced to be substantially slower than the late surgers they're thirty meters ahead of at late race start, then they would be guaranteed to win every time. +

+

+ Late race, or more precisely the last spurt, is also the only place where the speed stat and distance aptitude apply. + In terms of lengths gained, distance S actually does more for front runners than any other style due to SPC. +

+ Stats Speed @@ -184,39 +354,6 @@ Wit Skills - Win Conditions -

- On all medium races, almost every mile, many sprints, and some longs, Angling and Scheming is the second strongest single - skill currently in the game. Any time late race starts on (or, in the case of Nakayama 2500, very shortly before) a corner, Angling - is how front runners win. This means that Seiun Sky is the most important horse to front runner trainers. Everything about training - a front runner is for the sake of improving the odds of being the horse to trigger Angling – because only the horse in first gets - it. -

-

- Long races get a different primary win condition: Victory Cheer! from Kitasan Black. VC is more forgiving than Angling, activating - for both first and second place, but it's also weaker. -

-

- The sprints where Angling is dead can try to work with Victoria por plancha ☆ (Pasta) and Give Mummy a Hug ♡ - (Mummy Creek) instead. These skills are also on the weaker side, so front running becomes more of a matter of gambling with skills - like Turbo Sprint, Rushing Gale, and Unrestrained. The closest anyone has to not-gambling on such sprints is Nishino Flower's unique, - though, and front runners are especially strong in sprints for other reasons. -

-

- If Angling is the second strongest skill, then the strongest is No Stopping Me! - Unfortunately, in most races, it is generally not a front runner skill. Because only one horse gets Angling, they will pull away - from whoever is in second, not giving a chance to activate NSM. For NSM to win on a front, the horse who got Angling has to not - be Seiun Sky (since Sei's own Angling is equal to NSM in strength), and there has to be no gap in the front pack so that it can - trigger. As I write this, I am 80 races into CM12 using a Silence Suzuka who has NSM, and it has been involved in exactly one win - where she leapfrogged off (my) Sei to pass (my) VBourbon (who placed second). It was cool, but she would have been better off in - the event if she hadn't had Yukino Wit in her deck. -

-

- The above paragraph notwithstanding, NSM and even its white version, Nimble Navigator, are a decent choice for - multi-front builds on VC races. Since the top two horses both get VC, and the accel from it is weaker, having an extra push - for your second front is strong. VC maps also don't suffer from the final corner spread that makes NN/NSM hard to trigger on - sprints and miles. -

Gate Skills Lane Combo @@ -365,131 +502,159 @@ Kitasan benefits a lot from running as a Pace Chaser early on, just for the higher effective speed. Rivals are best defeated with your intended style, but in career, winning is more important than front running.

- My CM Teams + My CM Teams CM13 – Taurus Cup (Tokyo Derby)

- Maruzensky's unique is live as an order≤5 for approximately everyone. Filling the ranks with front runners should be a - strong means to delay it for later positions, especially COC. + Maruzensky's unique is live as an order≤5 for approximately everyone. + Filling the ranks with front runners should be a strong means to delay it for later positions, especially COC.

- I even considered using Maruzensky herself, since she's a front runner. That line of thought led me to some interesting - experiments in Umalator. Redshift hits 25m into the start of late race, as a 0.4 accel on Maruzen and 0.2 inherited, just like - Angling. It turns out that that delay has a substantial impact. Sei with Redshift beats Maruzen with Angling by about 0.4 - lengths. + I even considered using Maruzensky herself, since she's a front runner. + That line of thought led me to some interesting experiments in Umalator. + Redshift hits 25m into the start of late race, as a 0.4 accel on Maruzen and 0.2 inherited, just like Angling. + It turns out that that delay has a substantial impact. + Sei with Redshift beats Maruzen with Angling by about 0.4 lengths.

- The story doesn't end there, either. As it turns out, Redshift gains less than half a length for Sei. Ines Fujin's unique - (which has a strong version on Tokyo turf specifically) is worth about 0.2 lengths more! So, for Sei specifically, Ines Fujin - is the ideal inherit, not Maruzensky. + The story doesn't end there, either. + As it turns out, Redshift gains less than half a length for Sei. + Ines Fujin's unique (which has a strong version on Tokyo turf specifically) is worth about 0.2 lengths more! + So, for Sei specifically, Ines Fujin is the ideal inherit, not Maruzensky.

- Realistically, my team comp probably should be Seiun Sky, VBourbon, and Maruzensky. However, we run our oshis, and Silence - Suzuka is my favorite front runner, so she's going in. The question then becomes whether to run VBourbon or Maruzen. + Realistically, my team comp probably should be Seiun Sky, VBourbon, and Maruzensky. + However, we run our oshis, and Silence Suzuka is my favorite front runner, so she's going in. + The question then becomes whether to run VBourbon or Maruzen.

- Maruzensky has the advantage of working even as far back as 5th place. However, what does that actually beat? She's a front - runner, so she can only outrun another front runner, and only if she has a significantly higher spurt speed than whoever got - Angling. That basically means she needs to be a guts horse hoping for duels, which in turn means probably both of Professor - and Escape Artist aren't happening. That's tough. + Maruzensky has the advantage of working even as far back as 5th place. + However, what does that actually beat? + She's a front runner, so she can only outrun another front runner, and only if she has a significantly higher spurt speed than whoever got Angling. + That basically means she needs to be a guts horse hoping for duels, which in turn means probably both of Professor and Escape Artist aren't happening. That's tough.

- On the other hand, Maruzen isn't relying on a wit check for her big accel. She's also free to take VBourbon or Ines Fujin as - her non-Angling inherit, whereas VBourbon is forced into Sei and Maruzen parents. So, basically, what Maruzen would be trying - to beat is a VBourbon who hits both Angling and Redshift (>80% chance), matching 0.4 accels but winning in spurt speed. + On the other hand, Maruzen isn't relying on a wit check for her big accel. + She's also free to take VBourbon or Ines Fujin as her non-Angling inherit, whereas VBourbon is forced into Sei and Maruzen parents. + So, basically, what Maruzen would be trying to beat is a VBourbon who hits both Angling and Redshift (>80% chance), matching 0.4 accels but winning in spurt speed.

- I'm not convinced that's good for my comp. I'd rather just be that VBourbon, having approximately every good front runner - skill built in. So, final team comp: + I'm not convinced that's good for my comp. + I'd rather just be that VBourbon, having approximately every good front runner skill built in. + So, final team comp:

-
    -
  1. Seiun Sky as a gambler, where the gamble is getting into first in midrace.
    2. +
    1. - VBourbon as an ace. 1200 wit is basically mandatory thanks to the requirement of double accels. Final Push won't be a bad - take as a gamble-y backup. + Seiun Sky as a gambler, where the gamble is getting into first in midrace.
    2. - Silence Suzuka as Silence Suzuka. If you prefer winning over running your favorites, this should be Maruzensky instead. + VBourbon as an ace. + 1200 wit is basically mandatory thanks to the requirement of double accels. + Final Push won't be a bad take as a gamble-y backup. +
      3. +
    3. + Silence Suzuka as Silence Suzuka. + If you prefer winning over running your favorites, this should be Maruzensky instead.
    - CM12 – Aries Cup (Satsuki Sho) -

    - One of COC's best tracks, because U=ma2 is at worst only slightly less good than 777 as a trigger. If there is any other front - runner, triple front pushes pace COC out of range for U=ma2, making her at best as reliable as the usual. -

    -
      + CM12 – Aries Cup (Satsuki Sho) +

      + One of COC's best tracks, because U=ma2 is at worst only slightly less good than 777 as a trigger. + If there is any other front runner, triple front pushes pace COC out of range for U=ma2, making her at best as reliable as the usual. +

      +
      1. - Seiun Sky's Angling is a 0.4 accel that lasts for the entire accel period, better than COC's 0.3 that's only up for 2/3 of - it. I want her to be my ace in front, so capped wit, high power, strong spot struggles, huge mid-race skills. Didn't get a - guts build to come together after three weeks of attempts, so switched to a standard speed/power/wit build and got a high - roll on the first try. 1181/786/1185/474/1185 A/A/S. + Seiun Sky's Angling is a 0.4 accel that lasts for the entire accel period, better than COC's 0.3 that's only up for 2/3 of it. + I want her to be my ace in front, so capped wit, high power, strong spot struggles, huge mid-race skills. + Didn't get a guts build to come together after three weeks of attempts, so switched to a standard speed/power/wit build and got a high roll on the first try. + 1181/786/1185/474/1185 A/A/S.
      2. - VBourbon is a horse that exists. She can beat other people's front runners, so great as a backup. Ideally she lets Sei in - front, but it's better to let this happen naturally off the lack of TTL than to force low stats. Second attempt got charming - and fast learner for free, medium S, and manageable stats. Skill hints were a bit sparse, but not worth rolling more. + VBourbon is a horse that exists. She can beat other people's front runners, so great as a backup. + Ideally she lets Sei in front, but it's better to let this happen naturally off the lack of TTL than to force low stats. + Second attempt got charming and fast learner for free, medium S, and manageable stats. Skill hints were a bit sparse, but not worth rolling more. 1164/662/1010/599/1167 A/S/A.
      3. - Silence Suzuka is my favorite front runner, so I will run her. Her primary task is to be in third or fourth so COC can't be, - so I don't need amazing stats. To maximize her effectiveness, there are two possible plans: I could make her a debuffer, - which needs 1200 power and wit but no other stats matter, or I could experiment with something wacky like NSM into duels. - The latter sounds more fun, even if it is obviously bad. First attempt didn't get aptitudes but did get Lone Wolf to disable - it for everyone else and surprisingly decent stats, which is good enough for me; her job isn't to win anyway. + Silence Suzuka is my favorite front runner, so I will run her. + Her primary task is to be in third or fourth so COC can't be, so I don't need amazing stats. + To maximize her effectiveness, there are two possible plans: + I could make her a debuffer, which needs 1200 power and wit but no other stats matter, + or I could experiment with something wacky like NSM into duels. + The latter sounds more fun, even if it is obviously bad. + First attempt didn't get aptitudes but did get Lone Wolf to disable it for everyone else and surprisingly decent stats, which is good enough for me; + her job isn't to win anyway.
      -

      Win rates after 40: VBourbon 35%, Sei 17.5%, Suzuka 15%. Not quite executing the plan, but I'll take the wins.

      - Win rates after 80: VBourbon 30%, Sei 22.5%, Suzuka 12.5%. I believe this is my best round 2 performance ever. I lose more to - other fronts than to COC. "Most dominant racing horse for a year" continues to get trounced by the wacky triple front build. + Win rates after 40: VBourbon 35%, Sei 17.5%, Suzuka 15%. Not quite executing the plan, but I'll take the wins. +

      +

      + Win rates after 80: VBourbon 30%, Sei 22.5%, Suzuka 12.5%. I believe this is my best round 2 performance ever. + I lose more to other fronts than to COC. "Most dominant racing horse for a year" continues to get trounced by the wacky triple front build.

      CM11 – Pisces Cup (Hanshin 3200 Heavy Rain) -

      N.B. This CM was before I started writing this document, so henceforth, there is much less info.

      -

      Late race starts on the back stretch, which means the end closers are out to play.

      -
        +

        + N.B. This CM was before I started writing this document, so henceforth, there is much less info. +

        +

        + Late race starts on the back stretch, which means the end closers are out to play. +

        +
        1. - Kitasan Black is a snap take. Her unique is the only reliable accel outside of Straightaway Spurt, and it's quite a lot - better. 1200/1200/816/777/742 A/S/A. + Kitasan Black is a snap take. + Her unique is the only reliable accel outside of Straightaway Spurt, and it's quite a lot better. + 1200/1200/816/777/742 A/S/A.
        2. VBourbon's unique has a built-in recovery, which makes her the perfect choice as the survivor if stamina debuffers show up.
          3. -
        3. Silence Suzuka is coming. 1200/1145/653/608/1000 A/A/A.
          4. +
        4. + Silence Suzuka is coming. + 1200/1145/653/608/1000 A/A/A. +

        - I floundered on parenting and ended up with not enough time to make runners. Suzuka had more wit than Kitasan could handle, so - I rarely got Kitasan uniques. + I floundered on parenting and ended up with not enough time to make runners. + Suzuka had more wit than Kitasan could handle, so I rarely got Kitasan uniques. +

        +

        + Win rates after 80: VBourbon 31.25%, Kitasan 21.25%, Suzuka 2.5%. +

        +

        + Extremely unlucky finals gave me third place for the first time ever.

        -

        Win rates after 80: VBourbon 31.25%, Kitasan 21.25%, Suzuka 2.5%.

        -

        Extremely unlucky finals gave me third place for the first time ever.

        CM10 – Aquarius Cup (February Stakes)

        - Everyone is terrified of Taiki Shuttle, who has a 3-4 ult. Triple fronts would like to have a word. It's a dirt track, but - every horse can run dirt if you're brave enough. + Everyone is terrified of Taiki Shuttle, who has a 3-4 ult. + Triple fronts would like to have a word. + It's a dirt track, but every horse can run dirt if you're brave enough.

        -
          +
          1. - Smart Falcon is the obvious choice, being the only actual dirt front runner to exist. Her unique isn't terribly strong for - this track, but her gold skills are – Trending makes it extremely difficult for others to overtake her. 1200/467/920/410/930 - A/S/A. + Smart Falcon is the obvious choice, being the only actual dirt front runner to exist. + Her unique isn't terribly strong for this track, but her gold skills are – Trending makes it extremely difficult for others to overtake her. + 1200/467/920/410/930 A/S/A.
          2. - Silence Suzuka in runaway mode will make positioning much easier. I don't have to think about Unrestrained on my other - horses because they won't be able to get in position for it anyway. Other Suzukas will be rare because she has G dirt and - people don't realize distance aptitude hardly matters for runaways. 1200/674/820/470/774 B/A/A. + Silence Suzuka in runaway mode will make positioning much easier. + I don't have to think about Unrestrained on my other horses because they won't be able to get in position for it anyway. + Other Suzukas will be rare because she has G dirt and people don't realize distance aptitude hardly matters for runaways. + 1200/674/820/470/774 B/A/A.
          3. - Taiki Shuttle is a front runner now. She has B dirt and C front at base. Very easy to fix. Falco's mid-race is probably - stronger than Taiki's between her unique and Trending, so Taiki should often be in position for her ult in this build. + Taiki Shuttle is a front runner now. + She has B dirt and C front at base. Very easy to fix. + Falco's mid-race is probably stronger than Taiki's between her unique and Trending, so Taiki should often be in position for her ult in this build.

          - This is probably the strongest gameplan I've been able to use, but I failed to execute it properly. In particular, this was - the CM that taught me through experience how important mid race speed skills are for front runners. Final win rate was a bit - over 50%, including my first ever five win round 2 entry. Insane luck with Unrestrained at the same time as Angling made - Suzuka the champion of the Aquarius Cup. + This is probably the strongest gameplan I've been able to use, but I failed to execute it properly. + In particular, this was the CM that taught me through experience how important mid race speed skills are for front runners. + Final win rate was a bit over 50%, including my first ever five win round 2 entry. + Insane luck with Unrestrained at the same time as Angling made Suzuka the champion of the Aquarius Cup.

          Version History