TrendAviv Active instruments currently surfaced through model-driven asset pages.
Transparent model framework
How the TrendAviv model works
TrendAviv is built as a transparent market-intelligence framework. The goal is not to promise certainty, but to organize market, macro, sentiment, and volatility information into a disciplined structure that can be reviewed, challenged, and improved over time.
Risk-first interpretation, not headline forecasting.
Multi-horizon outputs with intervals and trust framing.
Evaluation layer designed to track behavior across changing regimes.
Methodology snapshot
A compact summary of the current framework so visitors can understand the product quickly before going deeper into the model logic.
Distinct market buckets used in the coverage universe.
Separate horizons matter because short-term and medium-term uncertainty behave differently.
The framework combines market, macro, sentiment, technical, and cross-asset context.
Out-of-sample and forward-style checks matter more than polished hindsight curves.
Public pages are tied to model timestamps so freshness is visible rather than implied.
What the model actually does
The model is meant to structure context, levels, and scenario logic. It is not designed as a certainty machine.
Output layer
Directional context
The model estimates whether the balance of evidence leans upward, downward, or neutral for a defined horizon.
Output layer
Expected levels
It transforms the current state into horizon-specific targets and level structure rather than a vague price opinion.
Output layer
Uncertainty ranges
Outputs are framed with interval logic so users can judge width, asymmetry, and regime stress instead of reading one point estimate in isolation.
Output layer
Risk-aware interpretation
Volatility, trust framing, and scenario discipline are presented alongside the forecast so context is always visible.
Architecture of the workflow
The system is organized as a sequence of explainable layers. That matters because a readable architecture is easier to validate and easier for users to interpret responsibly.
1
Data ingestion
Pull current asset context, historical prices, model-side parameters, and supporting public source layers into a single structured view.
2
Normalization
Bring inputs into comparable scales and consistent timestamps so different data families can be used together without hidden distortions.
3
Feature engineering
Convert raw series into interpretable features such as returns, volatility structure, regime flags, spreads, and sentiment context.
4
Signal extraction
Estimate the directional and contextual pressure that appears most relevant for the chosen horizon.
5
Forecast model
Generate expected levels and directional output through a structured modeling layer instead of hand-written trade rules.
6
Confidence and range layer
Translate output into intervals, trust framing, and scenario width so uncertainty remains visible.
7
Page rendering layer
Map the model output into a human-readable asset page with spot, target, range, signal, and evaluation cues.
Input layers used by the framework
The model does not rely on a single market series. It combines several input families because different parts of the market story matter at different times.
Input family
Market prices
Historical price behavior is used to estimate direction, drift context, recent structure, and how fast the market state is changing.
Spot history
Returns
Local structure
Trend context
Input family
Rates and macro context
Policy rates, rate differentials, and macro-sensitive context help explain why some assets behave differently even when the headline chart looks similar.
Rate spreads
Policy backdrop
Macro stress
Relative carry
Input family
News and sentiment layer
Structured news tone is used as a context feature, not as a claim that the model estimates events before they happen.
News tone
Event pressure
Headline clustering
Sentiment regime
Input family
Technical structure
Price geometry, support-resistance structure, momentum shape, and volatility behavior help define where the market is likely to react.
Support / resistance
Momentum shape
Moving structure
Volatility
Input family
Cross-asset context
Related instruments can matter because risk often moves through clusters rather than through isolated tickers.
Related assets
Spillovers
Risk transfer
Correlation context
Feature engineering logic
Raw data is not useful on its own. The model depends on feature groups that describe direction, regime, instability, and cross-market pressure in a more compact way.
Feature group
Price features
Price-based features capture trend persistence, local reversals, drift shape, and how the current move compares with the recent history of the asset.
Feature group
Volatility features
Volatility features help widen or compress forecast ranges and prevent overconfident interpretation during unstable conditions.
Feature group
Regime features
Regime features try to distinguish between calmer conditions, directional trends, and stressed environments where normal relationships can weaken.
Feature group
Macro-relative features
Relative macro features connect asset behavior to rate spreads, policy backdrop, and broader conditions that can dominate the tape.
Feature group
Sentiment features
Sentiment features translate the information environment into a usable context layer that can reinforce or challenge price-only interpretation.
Forecast horizons and why they differ
Short-term and longer-term outputs should not be read the same way. As the horizon expands, the uncertainty cone generally widens and the dominant drivers can change.
Shortest public horizon
Best read as immediate directional context and short-range level framing for the currently displayed public horizon on the asset page.
1 week
Useful for short tactical context when local drift and macro pressure are aligned but event risk still matters.
1 month
Bridges local market structure and broader macro context. The interval becomes more important than the center point alone.
1 quarter
More exposed to regime shifts, policy moves, and external macro changes. Read with wider scenario discipline.
Long horizon
Best used as structural context rather than a precise timing signal. Confidence should be judged through breadth, range width, and regime stability.
How a forecast is constructed
The page output is the result of a structured build process, not a raw chart overlay. Each layer serves a specific purpose in turning current state into a horizon-aware view.
1
Collect the latest asset state
The system starts from the latest market context for the chosen asset, including fresh parameters and related input layers.
2
Compute the feature set
Price, volatility, macro, sentiment, and structural features are assembled into a state description suitable for the model layer.
3
Estimate directional tilt
The model forms an initial directional view and translates it into expected levels for the active horizon.
4
Adjust by risk and regime
Volatility behavior and regime conditions shape how wide the interval should be and how strongly the point estimate should be trusted.
5
Render the public output
The final page exposes spot, target, signal, levels, ranges, and evaluation cues in a readable format.
How to read an asset page
Each asset page is a readable projection of the model state. The visual example below shows how to interpret the main components on a live asset page such as Index TA-35.
The reference level is the current anchor used for the model snapshot. Every target or interval should be judged relative to that anchor.
The active-horizon target is a modeled expectation for the currently displayed horizon, not a promise that the market must trade there.
The signal badge is a shorthand for directional tilt. It should be read together with interval width, not in isolation.
Support and resistance style levels provide structure around where price may react or stall within the modeled distribution.
Volatility describes expected move magnitude. The same signal looks very different under calm and stressed volatility regimes.
Trust or confidence is a framing layer that summarizes how readable the current setup appears under the model logic.
The range is one of the most important parts of the page. Narrower bands imply a more stable regime than wider ones.
Evaluation cues remind users that model output should be judged against observed behavior, not only against persuasive narrative.
Validation and evaluation logic
A model should not be judged only by how polished a forecast card looks. The framework needs evaluation layers that measure how the output behaves after publication.
Directional model accuracy estimate
55.8%
Directional model accuracy estimate tracks whether the public tilt broadly matched the realized move more often than not across observed windows.
Interval coverage
Coverage asks whether realized values fell inside the published band often enough to treat the range logic as useful.
Against a naive baseline
Model vs. simple carry-forward
A useful framework should beat trivial alternatives often enough to justify the extra complexity.
Stability across windows
Rolling-window review
Performance should be examined across different windows because a market that changes regime can distort one-off statistics.
Confidence, intervals, and uncertainty
Confidence is not the same thing as certainty. It is a contextual layer that should be read together with range width, volatility, and the stability of the current regime.
Calmer regime
When the market is relatively stable, intervals can be tighter and the public signal is usually easier to interpret.
Mixed regime
In mixed conditions the center view may still matter, but the widening band becomes more important for scenario planning.
Stressed regime
During stressed conditions the range widens sharply, confidence should be read more cautiously, and event risk can dominate.
What the model does not do
A transparent page should explain limits clearly. These are not legal afterthoughts; they are part of reading the output responsibly.
Important limit
The model does not guarantee direction or returns. It provides structured context, not certainty.
Important limit
It does not replace professional financial advice or personal portfolio suitability analysis.
Important limit
It cannot eliminate event-driven regime shifts, geopolitical shocks, or unexpected liquidity breaks.
Important limit
It does not make all horizons equally precise. Longer horizons should be treated as broader scenario context.
Important limit
It should not be read from the signal badge alone. Interval width, volatility, and validation all matter.
Data sources and update logic
Source families matter because they determine what kind of information the model can actually see. The goal is consistency and transparency, not source inflation.
| Source family | Role in model | Typical use | Refresh rhythm |
|---|---|---|---|
| Exchange and market data | Core price layer | Provides price history, asset state, and market-structure inputs for the public pages. | Aligned with market data availability and page refresh cycles. |
| Bank of Israel and macro references | Macro context layer | Supplies FX and macro-relevant context used for ILS-sensitive modeling. | As official series and relevant market context are updated. |
| Structured news / sentiment feeds | Event and tone features | Adds context around information pressure without treating headlines as guaranteed predictors. | Feature-dependent and generally tied to ingestion cycles. |
| Internal model outputs | Forecast presentation layer | Produces the page-facing target, interval, signal, and trust framing for each active asset. | Whenever the active model run publishes a new calculation window. |
How interpretation changes by market type
The same output components are used across pages, but the way users should interpret them can vary by market class.
Forex
For FX pairs, rate differentials, macro backdrop, and event sensitivity often matter more. A drift signal should be read together with policy and volatility context because FX can reprice quickly on macro headlines.
Indices
For indices, breadth, sentiment clustering, and cross-asset risk transfer matter more. Directional tilt can reflect both local structure and broader equity risk appetite.
Equities
For single-name equities, company-specific risk and wider market beta can mix in uneven ways. The model output should be treated as contextual rather than as a substitute for issuer-specific analysis.
Macro-linked assets
For macro-linked assets such as commodities or rates-sensitive instruments, the signal can be strongly influenced by policy, inflation expectations, and global risk narratives.
How to use the framework
The product is most useful when it is used for disciplined interpretation, scenario framing, and comparison across horizons rather than as a one-click trade trigger.
Use case
For directional context
Use the page to see whether the current balance of evidence leans bullish, bearish, or neutral for the active horizon.
Use case
For scenario planning
Use target and interval together to think in ranges and scenarios instead of forcing a single exact outcome.
Use case
For comparing horizons
Read the same asset across short and longer horizons to understand where the signal is tight and where uncertainty expands.
Use case
For market monitoring
Use the framework as a structured monitoring layer that helps you revisit the same asset page under changing market regimes.
Methodology roadmap
The framework is modular and designed to improve over time as the product grows and more evidence becomes available.
Expand cross-asset context so market pages can explain spillovers more explicitly.
Improve calibration and horizon-specific evaluation summaries as more public observations accumulate.
Broaden market coverage while keeping the same transparent page structure for interpretation.
Expose richer performance and validation views without turning the page into a black-box dashboard.
Add more user-facing transparency around model assumptions, refresh logic, and scenario interpretation.
Methodology FAQ
Short answers to the questions users usually ask first when they want to understand what the public forecast pages really mean.
Does this methodology does not guarantee market direction?
No. It organizes evidence into a structured model view, but it does not guarantee any market outcome.
Is this investment advice?
No. The page explains how the product structures research output. It does not provide personal investment advice.
Why do different horizons matter so much?
Because the drivers of short-term moves are not identical to the drivers of broader multi-week or multi-month behavior, and uncertainty expands with horizon length.
What does confidence really mean here?
Confidence is a framing layer that summarizes how readable the setup appears under the model logic. It should always be read together with range width and volatility.
Why can the signal change quickly from one refresh to the next?
Because market structure, volatility, sentiment, and macro inputs can all shift. A fresh model state can legitimately produce a different public interpretation.
How should I use this page together with an asset page?
Use the methodology page to learn how the output is built, then use the asset page to apply that reading framework to a specific live instrument.
Next step
Move from explanation to live market context
Once the methodology is clear, the best next step is to review the market hub or open a live asset page and read the output with proper horizon and risk context in mind.