AI Strategy Tools · Two Workflows, One Loop
SYSTEM-GENERATED COMPLETE PROMPTS

Move beyond generic AI ideas.Get settings you can map back to the system.

Sun-BD assembles real UI options, the saved strategy revision, and the latest trade-level backtest evidence into two complete TXT prompts.
Move from three candidate configurations to precise before → after analysis, re-testing, and simulated-trading observation.

Meet both AI workflowsSee the full operating flow
3Strategy horizons
3×13Candidate setting lines
90DEvidence-window limit
50Latest full trade cycles
THE SYSTEM DOES THE HEAVY LIFTING

Stop hand-assembling hundreds of fields.
Generate a complete AI task from the system.

This is not a one-line prompt. It packages timeframe roles, available methods, formal decision rules, the persisted strategy, and trade evidence into a constrained TXT task.

System interface illustration showing AI strategy horizon buttons and the backtest-improvement AI prompt dialog
Interface illustration redrawn from the implemented features and operating boundariesThe user attaches the complete TXT to an external AI
TWO CRITICAL AI WORKFLOWS

Two important capabilities in one strategy-growth loop

The first answers “How do I configure a viable candidate?” The second answers “What should I test next after seeing these results?”

01AI STRATEGY RECOMMENDATION

From a blank workspace
to three testable candidates

Turn the goal of a higher backtested win rate or stronger net-return profile into exact settings that can be tested—not a vague aspiration.

After you choose a horizon, Sun-BD captures the live strategy-workspace controls and formal decision content at that moment. The generated TXT instructs an external AI to return exactly three strategies, each with all 13 setting lines.

5–30 minutes1–2 hours1–2 days
  • Explicit main, trigger, and confirmation timeframe roles
  • One implemented trend, turning, and volatility method per set
  • Trend state, market phase, and trigger count each answered ON/OFF
  • Stop/target mode plus all four scaling actions
AI-RECOMMENDATION.txtCOMPLETE PROMPT
Strategy 1: <recognized market strategy>
1. Trend Direction: label (exact option ID)
2. Turning Strategy: label (exact option ID)
3. Volatility Risk: label (exact option ID)
4. Trend State: OFF or ON + timeframe
5. Market Phase: OFF or ON + timeframe + EnglishName
6. Trigger Count: OFF or ON + source + N/interval/window
...
13. Loss Reduction: OFF or ON + exact option
02AI BACKTEST ANALYSIS

Give every backtest
a disciplined next step

Do not hand AI a win-rate screenshot. Give it the settings that were truly tested and the complete causal trade evidence.

Sun-BD accepts only the latest individual backtest for the current strategy revision. Once revision, fingerprint, and detail-schema checks pass, it builds an improvement TXT containing the persisted snapshot, formal rules, signal rows, and fill evidence.

Current revisionLatest runTrade evidence
  • Causal findings must cite trade IDs and numeric evidence
  • Up to three improvement candidates, centered on one principal variable
  • Every proposal states before → after and the exact option ID
  • All 13 setting lines must be answered; weak evidence means no change
BACKTEST-IMPROVEMENT.txtEVIDENCE-BOUND
Outcome → trade evidence → current setting → causal hypothesis
Function | before (old ID) → after (new ID)
All other settings: unchanged
Re-test: Win Rate · Net P&L · PF · MaxDD
AI UNDERSTANDS THE THREE NEW GATES

No need to assemble complex parameters by hand.
All three first-entry gates enter the complete prompt.

AI Strategy Recommendation requires an explicit ON or OFF for every candidate. AI Backtest Analysis can then assess changes against the current revision and latest trade evidence.

04 · TREND STATE

Explicit timeframe and direction

Only HighFrequency, Hourly, or Daily is allowed, preventing a candidate from inventing a timeframe that the UI does not expose.

05 · MARKET PHASE

Canonical EnglishName

One phase is selected for the chosen timeframe, and the prompt states that market phase itself is not a Buy or Sell direction.

06 · NTH TRIGGER

N, interval, and window together

The source is one of two allowed modes; the full theme must agree, and only a false → true re-trigger can increment the count.

The user still controls write-back: Sun-BD generates the prompt, an external AI returns candidates, and the user reviews the actual UI before configuring, saving, testing, or enabling simulation.
FULL TXT · NO SILENT TRANSFER

Inspect it, save it, and choose where it goes

The complete prompt is visible in the application before it is exported as UTF-8 TXT. Sun-BD can open the official ChatGPT, Gemini, or Claude website, but it does not read, copy, paste, or upload the prompt.

01Generate and inspect the full prompt inside Sun-BD.
02Use “Save as TXT” to export it without chunking or summarizing.
03Open the official website of the external AI you choose.
04Attach the TXT yourself and keep control of the transfer.
Why not rely on the clipboard? The prompt can be large, and a complete TXT reduces the risk of transferring only part of it. External AI capacity, output quality, and data policy remain the provider’s responsibility—review them before sending.
SEVEN-STEP OPERATING FLOW

Recommend, test, analyze, re-test, simulate.
Every stage connects.

AI is not the finish line. It makes settings and evidence easier to validate, while every write, save, backtest, and simulation action stays behind a human review gate.

Seven-step flow: select a horizon, generate an AI recommendation TXT, receive candidates, configure and individually backtest, generate an AI analysis TXT, re-test, and observe in simulation
Complete workflow for both AI prompts and simulated tradingDashed paths show the repeatable review and re-test loop
EVIDENCE, NOT GENERIC ADVICE

AI receives more than a summary.
It receives identity, revision, and trade-level evidence.

The prompt rejects evidence from another slot, an older run, or outside market numbers. It also constrains new values to options that currently exist in the C# UI.

≤ 90DLatest-run evidence window
≤ 50Latest complete trade cycles
45 SECSignal and pessimistic-fill rows kept separate
13/13Every setting line answered
IDENTITY

Strategy identity lock

Member, slot, latest run, revision, and fingerprint must match.

CAUSALITY

Causal decomposition

Findings must trace back to trade IDs, cohort samples, raw P&L, and tested settings.

ACCEPTANCE

Re-test acceptance

Compare win rate, net P&L, PF, MaxDD, trades, average P&L, cost ratio, and stability.

WHY THIS CHANGES THE WORK

Let the system prepare the three hardest parts of AI use

01 · CONTEXT

No manual UI transcription

The prompt is built from implemented controls, reducing omitted fields, misspelled labels, and unsupported methods.

02 · FORMAT

Demand actionable output

Exactly three candidates and 13 setting lines—or a fixed diagnosis, change set, complete settings, and re-test plan—keep the answer out of generic-theory territory.

03 · ITERATE

Every change has a baseline

Start from the current revision’s latest test, control the main variable, and re-test whether higher win rate or stronger net P&L actually materializes.

AFTER BACKTEST · SIMULATED TRADING

Keep observing a refined strategy in simulated trading

After configuration, saving, and backtesting, enable simulation to observe first entry, additions, reductions, stop/target exits, closing, and transaction records without sending a live broker order. It is the bridge from historical replay to ongoing-market observation.

Learn about simulated trading
NO LIVE No live broker order path
LEDGER Independent simulated position state
LIFECYCLE Entry, scaling, stops, targets, and closing
RECORD Simulated evidence in the transaction UI
HONEST, USER-CONTROLLED AI

A high objective deserves an equally high evidence standard

Sun-BD makes strategy design and refinement easier. It does not turn AI, backtests, or simulation into a profit guarantee.

No automatic write-backAI responses are candidates. The user reviews before configuring, saving, backtesting, or enabling simulation.
No guaranteed win rate or profitHigher win rate, stronger net P&L, or lower drawdown can be research objectives; test evidence decides whether they were achieved.
No claim that simulation is liveSimulation sends no live broker order and cannot reproduce every matching, slippage, or connectivity condition.

Start with one complete prompt.
Turn strategy refinement into a repeatable validation process.

AI Strategy Recommendation solves the starting point. AI Backtest Analysis drives the next step. Individual backtests and simulation provide the evidence.

Download for WindowsExplore Custom Strategies
SYSTEM NAVIGATION

Continue exploring

01Product Overview02Features03Custom Strategies04AI Strategy Tools05Technical Values06Cloud Hosting07Download