STRATEGY PARAMETER ANALYSIS
20EMA Fast Period
Controls signal sensitivity and turnover.
LOCAL AI QUANT RESEARCH DESKTOP
Make the strategy explicit.
Then test what it depends on.
QuantAniki is an AI research desktop centered on a local workflow. Work with Aniki to turn a trading idea into executable conditions, study parameters, run per-symbol backtests, and test robustness.
Built for quant learners and independent researchers. No live execution. No return promises.
01Conversational strategy builder02Conversion quality gate03Pine compatibility gate04Parameter value research05Per-symbol backtests06Six robustness families
PRODUCT, NOT A CHAT WRAPPER
Conversation on the left. Evidence on the right.
The real product uses a three-part desktop workspace: strategy and data navigation, an Aniki conversation, and collapsible research reports. Every step continues from the same strategy state.
Build in ordinary language
Aniki asks for symbols, timeframe, sizing, fees, and slippage while updating the condition card. Text and model switching are built in; voice transcription is optional after FunASR is installed and configured.
Pass an execution quality gate
The summary separates strategy conditions, parameters, and backtest configuration. A strategy can be saved only after conversion checks pass.
Continue research in one window
After saving, the adjustable workspace keeps parameter research, backtests, and robustness reports side by side with Aniki.
ONE RESEARCH LOOP
Move from idea to evidence through the product's real flow.
Each step has a defined input, output, and next action.
Describe the rules
Enter a strategy in natural language or Pine Script. FunASR voice transcription is optional after installation and configuration.
Complete the specification
Confirm symbols, timeframe, sizing, entries, exits, fees, and slippage.
Structure and validate
Produce savable conditions and verify that the conversion is complete and executable.
Research parameters
Select candidates, design a coarse scan, and use binding and baseline-trade guards to reject invalid scans.
Run local backtests
Inspect portfolio equity, trades, risk metrics, and the contribution of selected symbols.
Test robustness
Run out-of-sample, walk-forward, regime, cost, neighborhood, and random perturbation tests.
IMPLEMENTED CAPABILITIES
Capabilities already connected to the workflow.
Each item below has a corresponding implementation in the current UI, API, or test suite.
01
DIALOG → DSLConversational strategy building
Complete entries, exits, symbols, timeframe, sizing, and costs, then produce a structured strategy.
02
PARSER → IRPine safe-subset analysis
For the executable subset of Pine v5/v6, Parser, AST, IR, and semantic audits identify unsupported features, repaint risk, and broker differences.
03
COARSE → VALUEParameter value research
Choose meaningful parameters before constrained coarse scans, value analysis, and finer validation.
04
PORTFOLIO / SYMBOLBacktest and symbol diagnostics
Inspect portfolio equity, trades, drawdown, Sharpe, profit factor, and selected symbols.
05
OOS / WFO / STRESSSix robustness families
Out-of-sample, walk-forward, regimes, cost stress, parameter neighborhoods, and random perturbations.
06
REPORT → NEXTReport-aware Aniki
Use parameter, backtest, and stability context to explain results and plan the next experiment.
PINE COMPATIBILITY LAYER
Do not trust a script before proving it can enter the research loop safely.
QuantAniki supports a safely executable subset of Pine v5/v6, not full TradingView compatibility. Source passes structural, semantic, temporal, and resource checks; unsupported paths fail closed.
Source coverage
Track provenance and conversion nodes so IR can be traced back to source.
Semantic audit
Check series, history references, multi-timeframe behavior, orders, and runtime semantics.
Risk report
Collect unsupported features, repaint risk, broker differences, and MTF warnings.
Execution gate
Separate safe-to-backtest and safe-to-optimize decisions; insufficient evidence fails closed.
SOURCE→PARSER→AST→IR
01series history references audited
02order semantics normalized
03repaint risk: none in executable path
04unsupported branch blocked before runtime
BACKTEST REPORT · SYNTHETIC DEMOEMA trend strategy
12 SYMBOLSBTCUSDT+4.8%52%ETHUSDT+2.1%49%SOLUSDT−0.7%47%
RESEARCH REPORTS
One equity curve is not enough.
Inspect distributions, symbol details, parameter rationale, and stability boundaries to separate broad validity, narrow contribution, and overfitting.
- Portfolio equity and drawdown
- Positive symbols / total
- Median return and drawdown
- Trades and win rate
- Sharpe and profit factor
- Per-symbol equity curves
Aniki can summarize the supplied metrics; when a model is unavailable, the interface still produces a computed conclusion without inventing data.
LOCAL-FIRST WORKFLOW
Data, models, and research history stay organized around your local environment.
The product centers on a Windows desktop and local API. Research records stay local by default; cloud models receive the prompts and context required for the selected task.
01
DATA DOWNLOADERBinance coverage and sync
USDT-margined futures, coin-margined futures, and USDT spot; inspect data lines, timeframes, coverage, and local sync.
02
MODEL CONTROLConfigurable model services
OpenAI, Anthropic, Qwen, DeepSeek, MiniMax, Ollama, and compatible endpoints. Full AI workflows require a configured model; API keys stay in environment variables.
03
RESEARCH MEMORYVersions, backtests, and notes
Keep strategy versions, backtest history, free-form notes, and Markdown reports in one research trail.
SCOPE & SAFETY
It helps you research. It does not trade for you.
No live execution
The current product has no brokerage connection, live order routing, or custody.
No return promises
Historical metrics do not represent future performance.
Failures remain visible
Missing Pine evidence, incomplete conversion, or invalid scans are blocked or explicitly reported.
Cloud models have a data boundary
Research records stay local by default; a selected cloud model receives the strategy context needed for the task.
FAQ
What you may want to know first.
01Can I build a strategy without coding?＋
Yes. Aniki asks for the missing conditions and turns your answers into a structured strategy.
02Does QuantAniki execute Pine directly?＋
No. It supports a safe subset of Pine v5/v6 and applies semantic, provenance, and execution gates; this is not full TradingView compatibility.
03Is optimization just brute force?＋
No. The product identifies candidates and research value before a guarded coarse scan and result validation.
04Will it connect to an exchange and trade?＋
No. The current product is a local research tool with no live order connection.
WINDOWS DESKTOP · COMING SOON
Research the strategy before you decide to trust it.
The first public QuantAniki build for Windows is in preparation.
View release status