Progression model

This is the complete technical description of the progression model, moved here from the player pages. It’s written for people who want the formulas: the Game Mechanics pages explain the same systems for players.

This page explains the core systems that govern character growth, class progression, and the mastery system. The MMORPG is built around an emergent multiclass model, where the character build takes shape through the activities performed, the attributes developed, and the classes unlocked over time.

Overview

Progression is built around a recurring loop that connects actions, stat growth, class development, and mastery advancement.

1. The player chooses or triggers an event.
2. The system generates a raw growth budget for that step.
3. Raw growth is reduced by diminishing returns and converted into actual stat increases.
4. Total experience increases and updates the PG Level.
5. New classes may become eligible and later unlocked.
6. The system updates the current main class.
7. The player may spend AP to improve attributes directly.
8. Unlocked and slotted classes gain Mastery Points, increasing their tier.

This creates a progression model where horizontal growth, vertical specialization, and class transitions influence each other without collapsing into the same exact progression layer.

Quick Glossary

Step

A single unit of simulation or progression. Each step represents one gameplay decision or one action-resolution cycle.

Event

An activity such as combat, stealth, arcane study, crafting, or social interaction. Each event has attribute weights, an event_strength value, and optionally a multiplier called mult.

UEA Raw

The raw growth budget generated during a step before any cost is applied.

Applied Gain

The effective stat increase granted after diminishing returns are applied.

Experience

The global progression value accumulated by the character. It is used to determine PG Level, progression scaling, and part of the dynamic growth-cost scaling.

Eligible Class

A class whose minimum stat requirements are fully satisfied.

Unlocked Class

A class that is actually available to the character. Eligibility alone is not always enough, because unlock a class from the current one require to overcome a variable friction threshold.

Main Class

The emergent primary class of the character. It is not selected manually, but determined by the closest unlocked class to the current stat profile.

Unlock Progress

A per-class progress value between 0 and 1 used by the progressive unlocking model.

PG Level

The global level of the character based on total XP. It is independent from static class levels.

AP

Applied Points that can be spent directly on stats, bypassing the normal diminishing returns of UEA growth.

Mastery

A vertical progression layer for unlocked classes. Mastery grows through Mastery Points and is converted into tiers.

Step Sequence

A typical step follows a functional pipeline that links event choice, growth generation, progression updates, class evolution, and mastery advancement.

1. Event selection: the player or an automated policy selects an event.
2. UEA raw generation: a raw attribute budget is generated using guided distribution, event weights, event strength, multiplier, and UEA configuration.
3. Diminishing returns application: raw growth is converted into effective gain, updating stats and total experience.
4. Global progression update: experience updates PG Level, XP to next level, and current progress.
5. Class unlock evaluation: eligible classes are checked and either unlocked immediately or advanced through unlock progress.
6. Main class update: the system may switch the main class only toward classes that are actually allowed.
7. AP allocation: available AP may be spent manually or through policy logic.
8. Mastery update: slotted classes gain MP based on event relevance and intensity.
9. Tracing and export: if enabled, the system records traces for debug, analysis, or external visualization.

Raw vs Applied: Diminishing Returns

Each step generates a raw growth value for one or more attributes. This raw value is not applied directly. Instead, it passes through a cost function representing diminishing returns.

Core Idea

Raw

How much training pressure or growth effort is generated.

Applied

How much actual growth is really obtained.

If a stat is already high, or if the character has accumulated a large amount of experience, additional growth becomes less efficient.

Cost Formula

For an attribute with current value x, the cost depends on the current stat value and a scaling function tied to total experience.

• a = 2.0
• p = 1.25

S(a) is a scaling function that grows with total experience and distinguishes between base and non-base attributes.

Experience also increases according to the total effective gain produced during the step.

Gameplay Effect

This system prevents players from stacking the same attributes forever with maximum efficiency. The more specialized a build becomes, the more expensive each additional point becomes.

Practical Consequences

• Stat curves naturally slow down over time.
• Overly narrow builds suffer stronger diminishing returns.
• Progression efficiency depends on balancing generated raw growth and applied gain.

UEA System: How Raw Growth Is Generated

The UEA system generates the raw delta of a step through multiple micro-extractions.

1. Number of Extractions

2. Size of Each Increment

• uea_scale controls overall growth speed.
• uea_shape controls the shape of the distribution. Higher values produce many small increments and fewer large spikes.

3. Split Between Base and Other Attributes

The generated budget is split between base attributes and other attributes using base_budget_ratio and 1 - base_budget_ratio. An additional guided bias may be applied through base_pick_bias.

4. Explore vs Guided Distribution

Explore

With probability explore_rate, the system chooses in a more random way. This mode may favor non-base attributes through explore_non_base_multiplier.

Guided

Otherwise, the system calculates a score for each attribute and applies a softmax distribution. The guided score may include current stat value, bias toward the main_class, novelty mechanisms, and softmax temperature.

Guided Tuning

• class_bias strengthens build coherence.
• novelty_boost and novelty_floor prevent neglected attributes from almost disappearing.
• temperature controls concentration. Low temperature means highly focused growth, while high temperature means broader growth.

Locked Attributes

• Talent
• Luck

These remain excluded from both UEA and AP growth.

Balancing Levers

• uea_scale
• k_min / k_max
• explore_rate
• temperature
• class_bias

Together, these parameters define progression speed, variety, and coherence.

Event Profiles and Event Strength

Events define how gameplay activities influence growth.

Each event includes an attribute-weight vector and an event_strength value between 0 and 1.

The final probability of selecting each attribute is a combination of the UEA guided distribution and the normalized event-weight distribution.

Formula

Where es = event_strength.

Interpretation

• event_strength = 0: the event has almost no impact, and growth mostly follows the general build trend.
• event_strength = 1: the event fully dominates the step, and growth reflects that specific activity directly.

Multiplier (mult)

The multiplier represents event intensity or repetition within the step. Depending on the active formula, it can affect UEA growth, Mastery gain, or both systems together.

In policy-driven simulations, mult is often sampled from a configured interval.

Event Tags

Events may also include tags useful for analysis, filtering, availability rules, or gameplay context.

Classes: Eligible, Unlocked, Main Class

The multiclass system separates three fundamental concepts: eligible, unlocked, and main class.

Eligible

A class is eligible when all minimum stat requirements are satisfied.

This is a strict requirement test.

Unlocked

A class is unlocked when it becomes actually available to the character.

• Instant unlock: eligible classes unlock immediately.
• Progressive unlock: eligible classes fill an unlock bar until the threshold is reached.

Main Class

The main class is not selected manually. It emerges automatically as the valid class closest to the current stat profile.

This usually happens by computing the distance between character stats and class vectors, typically through Euclidean distance, and selecting the nearest class.

Gating Rule

Main class changes are limited. The system may switch only toward classes that are already unlocked or classes unlocked during the current step.

Progressive Unlock with Friction

When friction is enabled, becoming eligible is not enough to unlock a class immediately.

When progress reaches 1.0, the class becomes unlocked.

Transition Cost Multiplier

cost_mult represents how difficult it is to move from the current main class to the target class.

1. Graph Friction

Graph_Edges[main][target]

If the edge does not exist, a fallback friction value is used.

2. Level Difference

• If dlvl > 0, moving upward is harder.
• If dlvl < 0, moving downward may be easier.

3. Clamp

The final multiplier is clamped within configured limits.

Gameplay Meaning

This system makes some transitions naturally slower than others. For example, moving from Rogue to Assassin may be much smoother than moving from Mage to Paladin.

Debug Output

• The full unlock_progress map.
• A reduced view of the classes closest to being unlocked.

PG Level

PG Level is the global XP-based level of the character. It does not match class level and does not depend on either main class or mastery.

Formula

• xp_level_A
• xp_level_P

Main Metrics

• Current PG Level.
• XP remaining to the next level.
• Progress percentage toward the next level.

Why PG Level Matters

PG Level influences several other systems, especially the number of Mastery slots and AP gain scaling. Even though it is global and class-independent, it strongly affects progression choices.

AP Allocation

AP are direct progression points used to increase stats without passing through UEA diminishing returns.

Basic Rule

• 1 AP = +1.0 applied directly to a chosen stat.

AP Gain

Per-Stat Cap

Tracking

• AP spent per stat.
• AP still available.

Policy Usage

Different policies may spend AP differently. Efficiency-oriented policies prioritize strategic unlocks and stat gaps, while balanced policies spread points more broadly.

Class Mastery System

Mastery is a vertical progression layer for unlocked classes. It allows the character to deepen expertise over time.

• Slots
• Mastery Points (MP)
• Tiers

Mastery Slots

Only classes placed in active slots can gain MP.

The system validates slot content by removing locked classes, removing duplicates, and trimming excess entries.

Practical Consequence

If a class is not present in a slot, it does not gain MP and cannot increase its mastery tier.

Policy Behavior

Different policies may manage slots differently. Some always prioritize the current target and the highest-value classes, while others maintain broader variety across unlocked classes.

MP Gain per Event

For each slotted class, the system calculates how relevant the current event is to that class.

Tuning

• mp_scale controls the overall speed of mastery progression.
• mp_mult_exponent determines how much high event intensity matters.

As a result, mastery grows faster when the event strongly matches the class identity, the class is actively slotted, and event intensity is high.

Tier Thresholds

The effective mastery tier is the highest tier whose cumulative threshold has been reached, up to the configured maximum.

Interpretation

• Increasing mp_tier_exponent makes higher tiers much harder to reach.
• Increasing mp_tier_base raises the entry cost of all tiers.

This allows the mastery system to remain meaningful even in long-term progression.

Extraordinary Unlocks

The system is prepared for special unlocks linked to advanced conditions such as minimum mastery tiers, combined tier totals across multiple classes, PG Level thresholds, or other progression milestones.

Design Intent

The goal of this system is to support a progression model that feels emergent, multiclass-oriented, strategic, scalable, and expressive.

• Emergent, because classes derive from what the character actually trains.
• Multiclass-oriented, because multiple classes can coexist and grow.
• Strategic, because unlocks, AP, and mastery introduce meaningful trade-offs.
• Scalable, because diminishing returns and XP-based progression prevent uncontrolled growth.
• Expressive, because events and policies shape very different builds.

Rather than imposing a fixed class path, the game allows character identity to emerge through repeated actions, stat evolution, and controlled class transitions.

How every attribute grows

Every trainable attribute takes part in the same three growth paths, so the per-attribute pages no longer repeat them:

• UEA growth: an attribute receives raw growth during a step when the current event profile, the build’s tendency and the guided distribution favour it.
• Applied growth: raw growth passes through diminishing returns, so the higher an attribute already is, the less each step adds.
• AP allocation: Applied Points raise an attribute directly, to complete class requirements, reinforce a build or patch a gap.

Attribute system design notes

These notes come from the original design of the attribute system (the legacy wiki’s overview page). The game’s character sheet groups attributes differently today (see Attributes); the original grouping is kept here for reference.

Attributes are one of the central pillars of progression in TenkaiClassOnline. They define what your character is capable of doing, shape class eligibility, influence event outcomes, and determine how your build evolves over time.

Overview

The attribute system is designed to represent different dimensions of the character. Instead of limiting progression to a few combat-only stats, TCO uses a broader model that includes physical ability, magical affinity, inner power, professions, and social/world interaction.

• Primary Attributes form the foundation of the character profile.
• Physical Disciplines describe martial and bodily ability.
• Magical Disciplines represent arcane talent and magical knowledge.
• Inner Power reflects spiritual, mental, and energetic strength.
• Crafting & Professions cover creation, enhancement, and specialized trades.
• Social & World Interaction govern influence, reputation, and non-combat presence.

Together, these groups create a flexible framework that supports a wide variety of builds, from straightforward archetypes to complex hybrid multiclass paths.

Why Attributes Matter

Attributes are not decorative values. They are tightly integrated with the core systems of the game.

• They determine class eligibility by checking whether class requirements are met.
• They influence how effective events are for the character’s progression.
• They interact with the UEA system, which distributes raw growth across attributes.
• They are subject to diminishing returns, making specialization and diversification meaningful choices.
• They influence the emergence of the main class and the pace at which new classes unlock.

Main Categories of Attributes

Primary Attributes

Primary attributes are the base values of the character. They represent broad innate or foundational capabilities and are often the most important requirements for class access.

• Strength
• Dexterity
• Constitution
• Intelligence
• Wisdom
• Charisma
• Talent
• Luck

Physical Disciplines

Physical disciplines define how the character applies bodily skill and martial ability in practical gameplay situations.

• Combat
• Athletics
• Stealth
• Survival

Magical Disciplines

Magical disciplines cover the study, channeling, and understanding of magical forces. They are essential for arcane and spell-based class paths.

• Magic
• Elemental Affinity
• Arcane Knowledge
• Ritualism

Inner Power

Inner power attributes represent mental discipline, spiritual stability, and internal energy. These are especially relevant for focused, mystical, or disciplined character concepts.

• Ki
• Spirit
• Willpower
• Focus

Crafting & Professions

These attributes support non-combat progression, item creation, enhancement systems, and production-oriented playstyles.

• Crafting
• Alchemy
• Enchanting

Social & World Interaction

These attributes represent influence over the world and other characters, including diplomacy, trade, leadership, and deception.

• Reputation
• Leadership
• Trading
• Diplomacy
• Deception

Attributes and Progression Systems

Attributes are directly connected to the progression logic described in the Game Mechanics section.

UEA Growth

During each progression step, the UEA system generates raw growth and distributes it across attributes based on guided probabilities, event weights, and tuning parameters.

Diminishing Returns

Growth becomes progressively less efficient as an attribute rises. This prevents infinite optimization of a single stat and encourages strategic planning.

Class Unlocking

Classes become eligible when the relevant attributes meet their required thresholds. Because of this, every distribution choice can accelerate or delay access to specific class paths.

AP Allocation

Applied Points can be used to directly improve attributes, allowing the player or policy system to patch weaknesses, reinforce strengths, or complete class requirements more efficiently.

Design Philosophy

The attribute system is meant to support meaningful character identity. Instead of being a simple list of RPG stats, it acts as a living structure that connects progression, class development, and gameplay specialization.

• It rewards specialization, but limits it through diminishing returns.
• It supports hybrid builds by combining multiple domains of growth.
• It allows class identity to emerge through action instead of rigid selection.
• It gives non-combat progression a real structural role in the game.

In TCO, attributes are not just numbers. They are the language through which the character’s history, focus, and future class identity are expressed.