Beta training

Track the edge, not just the idea.

The benchmark stack is live in beta. Preset models, webhook learning, replay imports, and style tracking are running now, while verified public result boards remain intentionally TBA until the data is clean enough to publish.

View TradingView Tools See News Desk

Benchmark status

Rank what actually works.

Public build

Tracked models

Preset buckets reporting now

Trained models

Models with learned history

Stored trades

390

Live alerts + imports

Current mode

Beta

Measured in public

Entity context

IG Group Holdings plc opened this benchmark view.

Back to entity

This view keeps model scoring, verification, and benchmark storytelling front and center while connecting the most relevant entity coverage when it adds context.

Reading the board

What the public numbers are meant to say.

The benchmark is not trying to produce one loud headline. It is built to show which workflow is being measured, what context shaped the result, and when the evidence is strong enough to publish publicly.

Identity before totals

Each board belongs to one preset or strategy so the numbers still describe a real workflow instead of a blended bucket.

Context is part of the edge

Market, session, trigger style, and directional filter stay attached to the sample so the benchmark can rank where a setup actually fits.

Publication discipline

The public board moves only when the outcomes are labeled, replayable, and clean enough to represent the product.

Model presentation

Separate models, visible logic.

The benchmark is moving away from one blended performance number and toward clearer models with their own sample, behavior, and preferred conditions.

Core Workflow Bundle

trained

Trades

165

Win

27%

Net

-257.52

TVC:GOLD / 5 / Balanced

Updated 09 Apr 2026

Sweep + Structure Bundle

trained

Trades

045

Win

29%

Net

-79.67

TVC:GOLD / 5 / Balanced

Updated 09 Apr 2026

Full BiasForge Suite

trained

Trades

005

Win

40%

Net

-6.03

TVC:GOLD / 5 / Balanced

Updated 09 Apr 2026

Execution Map Bundle

trained

Trades

004

Win

75%

Net

+253.19

OANDA:XAUUSD / 240 / Balanced

Updated 07 Apr 2026

Setup architecture

What feeds the benchmark.

The public board is only the surface. Underneath it, the benchmark is learning from direction, session context, trigger quality, and trade management as separate pieces of evidence.

Bias model

Higher-timeframe direction

Multi-Timeframe Bias and AMD keep the benchmark grounded in direction before any trigger is allowed to matter.

Multi-Timeframe Bias / AMD Phase Table

Session model

Reference range and timing

Session High/Low structure and the London-to-New-York handoff define when price is stretching and which sweep still counts.

Session HL / London Sweep references

Trigger model

Liquidity and structure confirmation

Sweeps, BOS or CHOCH, and imbalance confirmation turn the chart from a watchlist into a valid entry candidate.

Liquidity Sweeps / BOS + CHOCH / FVG

Management model

Risk and scaling logic

ATR-buffered stops, one-trade rules, and TP ladders keep outcome quality measurable instead of flattening every trade into one exit.

ATR stop / TP1 TP2 TP3 / one-trade rule

Model lifecycle

How a setup becomes a public board.

The benchmark is useful only if the pipeline stays disciplined. It should be obvious where the data comes from, how the model keeps its identity, and why a public card is trusted.

Capture

Alerts, imports, and replayed history feed one benchmark pipeline before any result is allowed to represent the product publicly.

Separate

Preset, session, trigger shape, and management logic stay attached so one model does not borrow the identity of another.

Verify

Only clean, labeled samples graduate into the public boards. Deeper diagnostics and QA stay inside the private research layer.

Verification snapshot

What is already active in the collection stack.

Measured in public

Tracked models

Receiving benchmark data now

Trained models

Have learned history attached

Stored trades

390

Closed records available for replay

Access layer

Building

Member-facing research surface

Public scorecards

Four preset boards, kept separate.

4 live / 4 tracked

Sweep + Structure Bundle

Beta live

Balanced / GOLD

Net

-79.67

Trades

045

Win

29%

Starter board for the liquidity and structure lane before the full suite is required.

Updated 09 Apr 2026

Execution Map Bundle

Collecting

Balanced / XAUUSD

240m

Net

+253.19

Trades

004

Win

75%

Value, imbalance, and POI comparisons stay grouped here once the sample is broad enough.

Updated 07 Apr 2026

Core Workflow Bundle

Beta live

Balanced / GOLD

Net

-257.52

Trades

165

Win

27%

Core stack board for the traders who want one repeatable lane from context into execution.

Updated 09 Apr 2026

Full BiasForge Suite

Collecting

Balanced / GOLD

Net

-6.03

Trades

005

Win

40%

Full-suite board for comparing the broadest workflow once enough labeled trades have been verified.

Updated 09 Apr 2026

Strategy spotlight

London Sweep + New York Expansion.

A dedicated London sweep model is now being tracked around session reference highs and lows, trend alignment, and staged exits. It is a strong candidate for the first named strategy board once live collection is broad enough.

Asia

Build range

Use the Asia session as the first reference range before London and New York start stretching price.

London

Extend range

Track London highs and lows, then merge them into the full reference range for the New York session.

Bias

EMA gate

Use the 21, 55, and 200 EMA stack plus a directional reclaim candle so the model only acts with trend support.

New York

Sweep + entry

Wait for New York to sweep the reference high or low and reclaim back through it before the trade is allowed.

Manage

Scale out

Stops use an ATR buffer and the trade scales through TP1, TP2, and TP3 so management quality stays visible.

Collection rules

Keep the London sweep strategy in its own model instead of blending it into the bundle presets.

Use the Asia and London sessions to define the range that New York is allowed to sweep.

Require trend alignment and a reclaim candle so sweeps are not treated as automatic entries.

Label only closed trades so the sample stays auditable and easy to replay.

Publish public performance only after sample size and QA thresholds are met.

Keep this strategy separate so its results stay attributable, easy to audit, and ready for a clean public board later on.

The alert payload already carries strategy name, side, session, stop, and reference levels, which makes it a strong fit for later benchmark replay and cleaner result labeling.

Example trade path

XAU / NY

Reference range

Asia and London build the high-low reference before New York becomes the trigger session.

Directional filter

The trade only qualifies if the EMA stack and reclaim candle still agree with the direction after the sweep.

Risk ladder

Stops use an ATR buffer, then the model scales through TP1, TP2, and TP3 to record managed outcomes instead of one flat exit.

Verification rules

verification

Dedicated model per strategy

This London sweep model should stay separate from the bundle presets so its sample remains attributable to one idea.

One trade per New York session

The current logic is intentionally selective: one trade per New York session keeps the benchmark cleaner during early collection.

Publication follows verification

The public board only goes live after sample size, labeling, and QA checks are strong enough to trust.

Publication split

public + private

Public benchmark

Show the named model, sample progress, scorecards, and polished trade examples once the evidence is ready for publication.

Private research layer

Keep deeper diagnostics, replay notes, QA checks, and internal review detail behind the member workspace.

Release path

What already supports the benchmark.

The benchmark is part of a broader product. These lanes show what is already live around it, what is still building, and what comes next.

Website and checkout

Live

The customer-facing layer is now focused on direct copy, cleaner navigation, working checkout, and clearer public pages.

Benchmark and training

Live

Benchmark is moving from raw strategy output into tracked styles, preset models, historical import, and repeatable learning.

News and event flow

Building

The next release lane is a sharper news stack built from exports, official APIs, and event tagging instead of filler headlines.

Access and dashboard

Building

Get Access, login, and the private dashboard entry are now public-facing instead of buried in future plans.