The VecViz framework
Two long-term inputs, one framework, three outputs. The sections that follow take you through each block in turn.
Vector Set channel
Major tops and bottoms typically occur on high volume and carry more signal than other price points. The regression line is the channel center, with standard-error bounds.
- bounds, ±1 standard error
- least-squares center
- price, lifted above the center by a bullish event
1 VecLevel (VL) = 1 standard error = 1 “channel width”
VecViz considers up to 2,028 Vector Set channels per ticker, aggregating them on an expected support and resistance weighted basis to a GMM distribution centered above the current price.
What a VecEvent is
A VecEvent is a dated storyline about the stock, with a direction and a trend that are judged separately. Each event is mapped to every Vector Set whose anchoring tops and bottoms occurred prior to or concurrently with the event.
| VecEvent | Direction | Trend |
|---|---|---|
| Regulatory relief for the sector | Bullish | Intensifying |
| Central bank easing lowers funding costs | Bullish | Steady |
| Input-cost inflation squeezes margins | Bearish | Waning |
Bullish / Bearish / Neutral · Intensifying / Steady / Waning
Each event pushes price above or below the center of the channels it maps to.
VecEvents move the target
A VecEvent’s bias, trend and timing determine its channel contribution. The model combines these contributions to update the VNA target.
Event emergence: during or after channel formation
| Bias | Bias trend | During | Post |
|---|---|---|---|
| Bullish | Intensifying | Up | Up |
| Steady | Flat | Up | |
| Waning | Down | Up | |
| Bearish | Intensifying | Down | Down |
| Steady | Flat | Down | |
| Waning | Up | Down |
Neutral bias: Flat for either timing and any trend
“Flat” treats ±0.0001 VL as effectively zero in this guide.
VNA target price
Vector-Narrative Alignment compares narrative with channel position to generate a target. Across ticker coverage, its implied return is calibrated on average to expected base-case returns over the next 6–12 months.
Base case = average expected body return. Individual ticker returns vary.
Two LLMs: build and review
One LLM generates the baseline VNA estimate. If an expert human cannot review it, VecViz recommends your agent check and supplement the VecEvents via web search and its MCP.
1 Baseline
An LLM catalogs and tags the events. The VNA model prices that baseline.
2 Review
A second LLM checks the news, revises tags and adds missing events. The model re-runs.
Every change is re-priced by the model
Illustrative review: neutralize an ended cycle, revise a trend, add a risk.
Price probability estimates
Four tail bounds plus two body averages, at a chosen horizon. The body spans the prices between 95D and 95U.
| Point | Definition | How to read it |
|---|---|---|
| 99D | 99% downside bound | 1% modeled probability below this price. |
| 95D | 95% downside bound | 5% modeled probability below this price. |
| EDB | Expected Down Body | Mean (base case) downside inside 95D, weighted by probability. |
| EUB | Expected Up Body | Mean (base case) upside inside 95U, weighted by probability. |
| 95U | 95% upside bound | 5% modeled probability above this price. |
| 99U | 99% upside bound | 1% modeled probability above this price. |
EDB/EUB are base-case means inside the 95% bounds. Bounds apply at the horizon end, not to interim extremes. Actual outcomes can breach them.
Sigma comparison
Sigma supplies a symmetric, normal-distribution benchmark from past returns. Compare it with the Vector Model at the same horizon and percentile to see how the bounds differ.
Illustrative 21D, 99% bounds
The Vector Model incorporates support, resistance and asymmetry by a machine learning model that predicts price movement on the basis of chart geometry in terms of support and resistance traversed.
V-Score
A timing signal that compares today’s setup with historical analogs using 13 chart geometry and price probability features. The headline score adds six horizon scores, each from −2 to +2.
Six horizons, one aggregate score
−12 bearish · +12 bullish
Near zero can mean neutral, mixed, or little discernible signal.
Illustrative examples. Research outputs, not investment advice.
Dashboards and API at vecviz.com ·