Typology of Prediction & Forecasting Projects - A Technical Guide

Introduction

At the turn of a new decade, prediction markets and forecasting platforms have emerged as one of the most innovative and intriguing applications of monetized information. By turning valuable information into tradable assets, these systems seek to crowdsource wisdom, aggregate probabilities, and - in some cases - challenge the accuracy and precision of even the most complex and sophisticated institutions. From drastic political changes to macro-economic indicators, weather events to corporate earnings, prediction and forecasting projects are steadily evolving into a new layer of financial and informational infrastructure by mastering the art of drawing parallel lines and connecting the dots between two or more seemingly unrelated events with uncanny precision and near-perfect accuracy.

Strangely, not all prediction and forecasting platforms are built the same. Behind every market lies a set of technical and mechanical design choices that determine how truth is resolved, how markets operate, and ultimately the extent of trust which participants can have in the outcomes. Some depend solely on decentralized oracles and data; others rely on human arbitration and curated feeds. Some perform fully on-chain with transparent smart contracts; others strike a strategic hybrid balance to optimize for performance, regulatory compliance and/or user experience.

These differences are not just academic - they shape real-world usability, resilience, scalability and even regulatory legitimacy. A forecasting project’s success heavily hinges on tradeoffs between speed and decentralization, accessibility and precision, simplicity and expressiveness.

This publication proposes a typology of prediction and forecasting platforms, organized around the core technical dimensions that define them.

For each category, we will:

• Identify the underlying mechanism,
• Provide real-world project examples, and
• Analyze their strengths, weaknesses and tradeoffs.

The goal is to equip builders, researchers, and users with a structured way to evaluate forecasting systems - and to highlight how emerging models such as Trepa’s accuracy-scaled payouts are redefining what prediction markets can achieve.

Key Classification Dimensions

Before we proceed to categorize forecasting and prediction projects, there is the need to establish the dimensions that meaningfully distinguish one system from another. The intersection of these axes of classification are not all arbitrary—they reflect the foundational technical and mechanical design choices that shape how scalable, reliable, trustworthy and pivotal a forecasting or prediction platform can and should be.

In analyzing prediction and forecasting projects, four technical dimensions stand out as the most useful for classification, and they are:

1. Resolution Mechanism,
2. Technical Infrastructure,
3. Market Design, and
4. Governance & Incentive Alignment.

Each provides a different lens into how projects operate, and together they offer a holistic view of the tradeoffs domiciled in prediction platforms.

2.1 Resolution Mechanism

The resolution mechanism of a platform defines and dictates how an outcome is determined, i.e., how the system actually decides what happens in the real world. This is by far the most pivotal technical layer and this is so because without trustworthy and transparent resolution, the entire forecasting market collapses and falls through.

Human-based resolution:

Heavily hinges on a trusted and vetted set of moderators, arbiters and/or community votes to determine outcomes and results.

• Pros: Easy to implement; evolves to ambiguous events and changes.
• Cons: Potential risk of biases, corruption or collusion; leads to the advent of subjectivity.
• Examples: Early Augur markets often fell behind on the votes of token-holders.

Oracle-based resolution:

Employs the use of cryptographic oracle networks and protocols, (e.g., Chainlink, Pyth, UMA etc.) to deliver real-world data on-chain and off-chain.

• Pros: Objective, scalable, and automatable.
• Cons: Limited to a finite number of outcomes (prices, scores, statistics); oracle downtime or risk of manipulation.
• Examples: Platforms like Polymarket and Zeitgeist heavily rely on oracles.

Hybrid strategies:

Synergizes human contribution with oracle feeds (e.g., oracles for quantitative data + community arbitration for ambiguous claims).

• Pros: Adaptability to handle the demands of both clear-cut and subjective markets.
• Cons: Increased system complexity.

Automated Feeds / APIs:

For extremely complex and structured data (e.g, sports scores, financial metrics, etc.), direct API calls and integrations can resolve outcomes.

• Pros: Fast, low-friction, and minimized human subjectivity.
• Cons: High risk of centralization, trust in the API source/provider is still required.

Key Takeaway:

Resolution tradeoffs are always a split-decision between objectivity and flexibility. No one single method can be used as the standard across all use cases.

2.2 Technical Infrastructure

The dimension of infrastructure covers where computation and settlement converge. This has effects on scalability, transaction costs, and user experience.

On-chain systems:

All logic, settlement, and records are stored on a decentralized blockchain network.

• Pros: Transparent, immutable, censorship-resistant technology.
• Cons: High transaction (gas) fees, much slower throughput, scalability & sustainability bottlenecks.
• Examples: Augur (Ethereum) and Zeitgeist (Polkadot).

Off-chain systems:

Market operations (e.g., matching orders, computing & calculating payouts) occur off-chain, with only settlement data that is anchored and domiciled on-chain.

• Pros: Faster and cheaper with more user-friendliness.
• Cons: Reduced transparency; trust assumptions in off-chain operators and administrators.
• Examples: Kalshi, Predict-It.

Hybrid systems:

Utilizes a layered approach — off-chain for speed, on-chain for finalization.

• Pros: Balances performance with security; flexible and upgradable architecture.
• Cons: Increased design complexity as the system grows.
• Examples: Polymarket uses off-chain matching + on-chain finalization.

Key Takeaway:

The core compromise and tradeoff in this dimension is between efficiency (off-chain) and trust minimization (on-chain).

2.3 Market Design

The dimension of market design determines how forecasting actually takes place — whether users make binary or spread bets, trade continuous shares or turn in numerical predictions.

Binary prediction markets:

“Yes-or-No” markets on specific outcomes and results.

• Pros: Simple and easy to understand.
• Cons: Low-level of details, outcomes are reduced down to black-and-white.

Scalar or continuous markets:

Users forecast numerical values and changes (prices, percentages, event magnitudes).

• Pros: Richer and holistic information; aligns closely with real-world forecasting needs.
• Cons: Complex pricing models; requires an in-depth user understanding.
• Examples: Trepa’s specialty is in scalar forecasts (e.g., GDP growth, inflation).

Parimutuel/Pool-based models:

Users and contributors fund liquidity pools; payouts depend on share of the correct and accurate side.

• Pros: Easy implementation process with scalable liquidity.
• Cons: Can experience setbacks from poor odds if liquidity is out of balance.

Orderbook models:

Markets operate and function like exchanges, with the presence of bids/asks for outcome shares.

• Pros: Much more accurate and precise price discovery, deeper liquidity pools.
• Cons: Increased complexity; high demand of market makers.

Key Takeaway:

Market design tradeoffs balance simplicity vs. expressiveness. Simpler markets attract casual/seasonal participants, while continuous models open up greater opportunities for more sophisticated forecasting.

2.4 Governance & Incentives

Prediction and forecasting projects can be made or marred by how they incentivize honest participation and align stakeholders.

Decentralized governance:

Voting by token holders or DAO frameworks decide on disputes, upgrades, and parameters.

• Pros: Community-driven, censorship-resistant.
• Cons: Voters’ apathy and risks of plutocracy.

Centralized governance:

Project/founding team makes key decisions.

• Pros: Faster and decisive actions with easier product iterations.
• Cons: Higher risk of centralization, potential increase in user distrust.

Key Takeaway:

Governance is about who controls the rules and incentives are about why users play honestly. Projects that misalign typically collapse into manipulation or low participation and then rapidly die.

Incentive mechanisms:

• Dispute bonding which requires stake to challenge outcomes.
• Reputation-based systems which reward historical accuracy.
• Token rewards disbursed for liquidity provision and/or accurate forecasting.

Quick-View Table: Key Classification Dimensions

Dimension	Variants/Models	Pros	Cons	Examples
Resolution Mechanism	Human, Oracle, API Feeds, Hybrid	Flexible, objective (with oracles), scalable	Risk of bias, oracle failure, limited scope	Augur (human), Polymarket (oracle), Zeitgeist (oracle + human)
Technical Infrastructure	On-chain, Off-chain, Hybrid	Transparent, secure (on-chain); efficient, fast (off-chain)	On-chain: costly & slow; Off-chain: trust assumptions	Augur (on-chain), Kalshi (off-chain), Polymarket (hybrid)
Market Design	Binary, Scalar/Continuous, Parimutuel, Orderbook	Simple (binary), expressive (scalar), scalable (parimutuel)	Limited granularity, complexity in pricing/liquidity	Trepa (scalar), Polymarket (binary), Predict-It (parimutuel)
Governance & Incentives	DAOs, Centralized team, Incentive models (staking, reputation, token rewards)	Community-driven, aligned incentives	Voter apathy, centralization risks, potential of manipulations	Augur DAO (decentralized), Kalshi (centralized), Zeitgeist (bonding disputes)

Summary:

These four dimensions — Resolution Mechanism, Technical Infrastructure, Market Design and Governance & Incentives — synergize together to create the backbone of any classification framework. They are deeply interconnected: for instance, a project’s choice of resolution mechanism poses constraints on its infrastructure (on-chain or off-chain), which in turn shapes market design and inevitably dictates governance needs. Together, they define the technological and mechanical DNA of prediction and forecasting platforms.

Typology of Prediction and Forecasting Projects

In this section, we employ the use of the four core technical dimensions from the previous section - Resolution Mechanism, Technical Infrastructure, Market Design and Governance & Incentives - which serve as the holistic lens for classifying forecasting projects. The following categories are some practical clusters you’d see more often than not in industry and research. For each category, I list the typical resolution choices, infrastructure patterns, market structure, and governance tendencies, followed by real-world examples and their main tradeoffs.

3.1 Short-Term Operational Forecasts

Definition:
Near-perfect real-time forecasts (minutes → days) that power operational systems and automated choices and decisions.

Typical choices:

• Resolution Mechanism:
  Autonomous data feeds, sensor streams, oracles for market prices; void of human arbitration and participation.

• Technical Infrastructure:
  Predominantly based off-chain or hybrid (real-time processing architectures, event streams, caches); reliable pipelines with negligible latency.

• Market Design:
  Not usually public markets, but more continuous numerical systems or private “prediction” APIs; if properly marketed using scalar and/or orderbook-style markets for extensive liquidity and instant repricing.

• Governance & Incentives:
  Centralized ownership of products; short and closed-circuit feedback loops; incentives are often tied to meeting the operational KPI targets.

Examples:
Load balancing, short-term forecasting demands (ride-sharing, package delivery), day-to-day financial price forecasts.

Tradeoffs/Challenges:
Low latency is heavily prioritized; risk of over-crowding to noise; lapses have instant operational costs.

3.2 Strategic Business Forecasts

Definition:
Medium-horizon forecasts and predictions (quarters → a few years) utilized for planning, investments, and proper resource allocation.

Typical choices:

• Resolution Mechanism:
  Aggregate of external data sources in synergy with human judgment and executive review.

• Technical Infrastructure:
  Hybrid: cloud-based analytics and insights, in-house BI (Business Intelligence) tools; auditable records housed on-chain on an occasional basis.

• Market Design:
  Scalar forecasts (revenue, expenditure, ARPU) or in-house continuous markets used to gather distributed judgments and decisions; outputs are typically based on scenarios, rather than continuous traded markets.

• Governance & Incentives:
  Centralized or semi-centralized governance (management & operational oversight); incentives is a culmination of bonuses, reputational metrics, and in-house prediction scoring.

Examples:
Revenue forecasts, projections of product adoption, market survey and sizing for new launches.

Tradeoffs/Challenges:
A holistic blending of quantitative models with managerial judgment; handling of structural shifts and changes; aligning forecasts with incentives (optimism bias).

3.3 Long-Horizon Policy & Infrastructural Forecasts

Definition:
Multi-year → multi-decade forecasts used for public policy, infrastructural planning, as well as climate/economic scenario modeling.

Typical choices:

• Resolution Mechanism:
  A human panel of experts, an assembly of scientific models, or long-run empirical datasets. Oracle-format automation is useful for inputs (e.g., emissions measurements) but overall human interpretation is central and pivotal.

• Technical Infrastructure:
  Heavy-duty off-chain computing power such as high performance compute (HPC), climate models, and simulations; replicable data pipelines and archives that are readily available; transparency is very important but on-chain settlement is uncommon.

• Market Design:
  Not generally available in public markets; when markets exist, they act as supplementary signals (e.g., forecast markets for policy outcomes). Forecasts are based more on scenarios, rather than transactions.

• Governance & Incentives:
  Multi-stakeholder governance systems (governments, NGOs, academic institutions, etc.); incentives are often institutionalized (funding and policy outcomes) rather than market payouts.

Examples:
Projections of climatic conditions, national infrastructure demand models, energy-transition planning process.

Tradeoffs/Challenges:
Characterized by high level of uncertainty, long validation horizons, political & electoral landscape, conflict of interests & competing agendas, effectively communicating probabilistic outcomes and projections to non-technical stakeholders and audiences.

3.4 Risks & Crises Forecasts

Definition:
Forecasts that are primarily focused on tail risks, extreme events, unforeseen circumstances, or crises (financial, epidemics, cyber incidents) that are rapidly unfolding.

Typical choices:

• Resolution Mechanism:
  Oracles for quantifiable triggers (price crashes) in combination with human adjudication for unplanned and ambiguous crises; disputes mechanisms are commonly used when stakes are high.

• Technical Infrastructure:
  Hybrid: systems for simulations and stress-testing are domiciled off-chain with real-time feeds for input; high levels of reliability and auditability is a must-have.

• Market Design:
  Can be binary (based on unfavorable events) or scalar (based on the severity and extent of events); sometimes in-house hedging and insurance instruments are utilized, as opposed to using public prediction markets.

• Governance & Incentives:
  Risk management teams are centralized; strong emphasis is placed on accountability and conservative decision-making processes. Incentives may be tied to risk-mitigation results and outcomes.

Examples:
Insurance models are built in the event of catastrophes, financial stress-test forecasting, early-warning systems for epidemic breakouts.

Tradeoffs/Challenges:
Wide range of data on extreme events, uncertainties in the efficiency of models, difficulty in convincing stakeholders to act on low-probability or high-impact warnings.

3.5 Socio-Political Forecasts

Definition:
Such forecasts are about collective human behaviour regarding circumstances such as elections, public sentiments, regional protests, and geopolitical events.

Typical choices:

• Resolution Mechanism:
  Mixed: fixed, final and verifiable data points (official election results return via oracles/APIs) where available; in other scenarios, human adjudication, special data pipelines or crowdsourced consensus are used to validate results and outcomes.

• Technical Infrastructure:
  Large-scale data ingestion process (social media, news scraping), off-chain sentiment systems; can be settled on-chain if hybrid.

• Market Design:
  Binary and categorical markets are very common; scalar forecasts could be used to distinguish turnout percentages or margins. Market structure varies across the board.

• Governance & Incentives:
  Markets are often curated and permissioned to mitigate manipulation and misinformation; anti-abuse systems (KYC protocol, curation) and reputation assessment systems are a common occurrence.

Examples:
Election and result prediction markets (Predict-It, Polymarket) conflict-risk forecasting, protest likelihood systems.

Tradeoffs/Challenges:
Data bias (social media buzz ≠ population), reflexivity (forecasts determine behaviour), high risks of misinformation and manipulation.

3.6 Scientific & Discovery Forecasts

Definition:
Forecasts like these are to guide research, scientific replication, or discovery timelines (predicting replication success and time-to-breakthrough).

Typical choices:

• Resolution Mechanism:
  Peer review, experimental validation systems, long-term empirical verification process. Markets are rare, are only available to a select few and are used when forecasting tournaments and reputation tallies are common.

• Technical Infrastructure:
  Off-chain research compute hours and power, publicly available data repositories, reproducible pipelines; provenance and audit trails are extremely valuable and important.

• Market Design:
  Non-market or tournament-style scoring system; where markets exist, payoff may reward calibration and/or accuracy, rather than trades.

• Governance & Incentives:
  Academic norms, fund-driven incentives, and reputational systems, as opposed to having pure financial rewards.

Examples:
Forecasts of reproducibility and replicability in science, possible timelines for technological and medical breakthroughs.

Tradeoffs/Challenges:
Unusually long feedback loops, disparity in funding and incentive alignment, transmuting probabilistic scientific forecasts into policy/commercial actions.

Quick Reference Table - Typology X Core Classification Dimensions

Category	Resolution Mechanism (typical)	Technical Infrastructure (typical)	Market Design (typical)	Governance & Incentives (typical)
Short-Term Operational	Automated feeds/oracles	Off-chain/Hybrid (low latency)	Continuous/scalar or in-house APIs	Centralized operations; KPI-based incentives
Strategic Business	Aggregated APIs + expert review	Hybrid cloud/BI platforms	Scalar forecasts; internal markets are occasional	Centralized executive oversight; bonuses
Long-Horizon Policy	Expert panels & ensembled models	Off-chain HPC & data archives	Scenario-based; non-market	Multi-stakeholder governance system
Risks & Crises	Oracle triggers + human adjudication	Hybrid (simulation engines + feeds)	Binary/scalar (severity)	Centralized risk panels; regulatory oversight
Socio-Political	Oracles for verifiable data + human review	Off-chain web-scale integration	Binary/categorical markets are common	Curated & permissioned + reputation systems
Scientific & Discovery	Peer review/experimental validation	Off-chain research compute	Non-market/tournament-style	Academic & funder-incentives; reputational

Typology of Prediction and Forecasting Projects (Continued with Relevant Case Studies)

While the first six categories provide a structured typology, the concepts are best illustrated through the use of real-world examples. Each category below is closely related or similar to the ones outlined previously and they are paired with some live or historical projects, showcasing how they operationalize predictions and their degree of alignment with the four classification dimensions.

3.7 Political & Event Forecasting

Case Studies:

• Polymarket (Ethereum/Polygon, 2020 - Present): Users bet on U.S. elections, geopolitical conflicts, and cultural outcomes. Resolutions are tied to trusted oracles (UMA/Chainlink).
• Predict-It (U.S., 2014 - 2023): A regulated event market focused on U.S. politics operated under a CFTC no-action letter but later had to inevitably call it quits.
• Kalshi (U.S. 2020 - Present): A federally monitored exchange offering event contracts across politics, weather, and macroeconomics.

Dimension Mapping:

• Resolution Mechanism: Oracles + regulated oversight (Kalshi/Predict-It).
• Technical Infrastructure: Hybrid - Polymarket fully on-chain, while Predict-It and Kalshi are fully centralized.
• Market Design: Continuous order books with fixed bet sizes.
• Governance & Incentives: Centralized governance system (Predict-It/Kalshi) vs. crypto-economic community incentives (Polymarket).

3.8 Assets & Financial Markets Forecasting

Case Studies:

• Numerai (Ethereum, 2017 - Present): A hedge fund sourcing forecasts from thousands of data scientists who stake $NMR tokens on their models.
• Augur (Ethereum, 2015 - Present): One of the foremost decentralized prediction protocols, allowing markets on asset prices, commodities, and crypto results.
• Kalshi (again, no doubt): Fits in here for regulated financial derivatives.

Dimension Mapping:

• Resolution Mechanism: Smart contracts + token staking (Numerai/Augur).
• Technical Infrastructure: Augur being fully on-chain while Numerai utilizes hybrid off-chain aggregation.
• Market Design: Data scientists vs. open trading participation; order books or AMMs.
• Governance & Incentives: Numerai possesses a strong token alignment (staking for accuracy), while Augur had struggled with low liquidity despite complete decentralization.

3.9 Scientific, Research and Discovery Forecasting

Case Studies:

• Metaculus (2015 - Present): A structured forecasting platform focusing on science, technology, medicine and existential risks.
• Good Judgment (2011 - Present): A forecasting community stemmed from the IARPA research program focused on aggregating “superforecaster” insights.
• CrowdCast (past): Used previously by organizations for in-house R&D forecasts.

Dimension Mapping:

• Resolution Mechanism: Expert-curated or community consensus (Metaculus, Good Judgment).
• Technical Infrastructure: Primarily traditional platforms with custom aggregation algorithms.
• Market Design: Reputation-weighted scoring, as opposed to financial stakes.
• Governance & Incentives: Prestige and reputation systems (non-financial), though Metaculus is currently exploring tokenized models.

3.10 Sports & Entertainment Forecasting

Case Studies:

• Betfair Exchange (2000 - Present): A massive centralized sports betting exchange with an extensive global reach.
• ZenSports (2017 - Present): A blockchain-based P2P sports betting marketplace.
• DraftKings Prediction (U.S., 2020 - Present): Bridging sports fantasy leagues with outcome-based wagers.

Dimension Mapping:

• Resolution Mechanism: Licensed sportsbook oracles (Betfair, DraftKings); smart contract verification (ZenSports).
• Technical Infrastructure: Centralized but regulated (Betfair); decentralized apps (ZenSports).
• Market Design: Order books, betting pools, and spread-based systems.
• Governance & Incentives: Operator-driven (Betfair/DraftKings) vs. token/community incentives (ZenSports).

3.11 Enterprise & Organizational Forecasting

Case Studies:

• Cultivate Forecasts (2013 - 2018): Provided enterprise forecasting tools for corporate strategy and risk management.
• Consensus Point (2006 - Present): One of the longest-surviving providers of internal prediction markets, used by companies like Ford Motors and Best Buy.
• Google’s Internal Prediction Markets (2005 - Present): Used to forecast product launches, hiring outcomes, and adoption trends.

Dimension Mapping:

• Resolution Mechanism: Internal subject-sensitive experts or HR/business data feeds.
• Technical Infrastructure: Proprietary enterprise software solution, typically private servers.
• Market Design: Limited-access markets, often with digital currency.
• Governance & Incentives: Oversight is by organizational leadership; incentives are reputation-based, instead of financial.

3.12 Public Good & Crises Forecasting

Case Studies:

• COVID-19 Forecasts Hubs (2020 - 2022): Aggregating epidemiological models and systems for pandemic cases and death forecasts.
• UN Global Pulse: Exploring crowdsourced and data-backed predictions for humanitarian response.
• INFER (Intelligence Forecasting Enterprise, U.S. Government, 2022 - Present): Using publicly available forecasts for policy-relevant questions.

Dimension Mapping:

• Resolution Mechanism: Scientific data in synergy with government-verified indicators.
• Technical Infrastructure: Centralized dashboards, with some publicly available and open-sourced.
• Market Design: Aggregated probabilistic scoring rather than trading financial instruments.
• Governance & Incentives: Operated by public agencies and NGOs; incentives are social good & impact, reputation and policy influence.

Comparative SWOT Analysis

Part 3

Where Trepa Fits In

Part 4

Conclusion

Closing statement and finalization