Period Trader/Features

MT5

Binance

Bybit

Execution ready

Features

AI creation. Decision models. Research you can inspect.

Create with the assistant, test with the research engine, review with Decision AI, and supervise eligible live strategies with your own controls.

Get Started Run the backtester preview.

AI + research

An assistant to build. Decision models to review. Research to test.

Period Trader brings AI into the work you already do: turn an idea into a test, understand the results, and review decisions without replacing your research or your controls.

Create and work with AI

Chat with the platform or start with a brief in Create. The assistant reads available platform data and proposes changes for your approval.

Strategies and chart indicators: Describe your rules or paste source. Create tunable strategies for backtests and research, or supported formula-based indicators for the chart. Unsupported indicator features must be disclosed.
Backtests and research sweeps: Ask for a test or parameter search, review the proposed settings, then approve it. AI can help interpret the results; the research engine still runs the tests.
Reports and risk reviews: Ask about closed trades, fees, open positions, market context or bot risk. Use text or voice input; the answer depends on the data the app can read.

Decision AI across the workflow

Specialized decision models return structured assessments, separate from the chat assistant. Each feature has its own controls and a recorded activity history.

Entry and deployment checks: Review bot signals and backtests for potential problems. Shadow records the assessment without changing execution; Veto can refuse a flagged entry or deployment when you enable it.
Market conditions and bot health: Classify market conditions, flag bot-health concerns, and review possible reasons for a closed trade. These are model assessments, not established causes or forecasts.
Manual second opinions and alerts: Request an opinion on a chart ticket, receive exit-advisor notifications, or label and filter alerts. The advisor does not place orders or close positions itself.

One shared AI credit balance

Chat, Create, voice input and Decision AI consume the same AI credits. Check your balance and usage in the app. Access to the tools is not a promise of unlimited AI usage or a fixed credit allowance.

Your approval. Your execution settings.

Assistant changes wait for approval. Auto-deploy is a separate setting: enabling it can let passing eligible research runs go live. Decision AI is not a substitute for validation, position sizing or the account risk guard.

Local research. Connected AI.

Backtests and research run on your PC. AI processing uses external services and can send your messages, supplied code, voice input, and relevant market/account context. Do not put passwords or API secrets in prompts.

Custom strategies can be backtested and researched, but cannot currently be deployed live. Supported built-in strategies have the live execution path. AI can be wrong; no model assessment or backtest guarantees a future result.

Try it

Run the backtester preview.

Pick a strategy and a market, then watch a research run, the results and a deployment — a simulation built from the real app.

⚡Period Trader

Execution readyResearch0 running · 0 paused15M · cached locally

Research bots

Point it at a market and a timeframe. It works through the parameters, keeps what holds up and discards the rest.

01Parameter combinations are generated and backtested.
02Every result is scored for overfitting — most are discarded.
03Survivors are walked forward across three segments.
04What holds up can be deployed behind the risk guard.

Example only · Simulation using example data. Not a backtest result, and not a performance claim.

The problem

A good idea needs evidence.
Keep the research visible.

Any tool can show you a strategy with a perfect win rate. That usually means it memorised the past.

Period Trader is built to find the ones that do not survive — and to say so plainly, with the reason attached.

Sized on your real account balance, not a number you typed in
Priced with your broker’s actual spread and commission
Rejected outright when a result fails validation

⚡Period Traderbacktest · reportexample view

equity curveall cached data

PASSwalk-forward eligible

score 71.4

trades

win rate

54.7%

max DD

8.6%

warnings

1 — stress segment sample is thin

⚡Period Traderresearch · exampleexample view

Consolidation Breakout1,676 / 4,987

ATR multiple1.0

Consolidation count17

Consolidation length40

Stop multiple2.2

TP multiple 11.5

Trail multiple1.2

combo rejectedno trades generated

XAUUSD · 4H

Piper Weighted DCA + Fibonacci

equity

1,000

2.4%

open

guard armed

Research stays at the center

AI helps frame the question.
Research tests the idea.

Start from a built-in strategy or a custom idea. Run parameter sweeps, compare results and inspect costs, drawdown and holdout checks. An AI answer is not a substitute for running the test.

Built-in + custom

Strategies

MT5 + crypto

Markets

Walk-forward

Validation

Your PC

Research runtime

Validation

Inspect the evidence before execution

Not a switch you turn on. Every result passes through it before it can reach a live account.

Reality Score

An explicit PASS or FAIL against the configured validation rules, with reasons. Passing a historical check does not predict live success.

Walk-forward split

Development, validation and stress. Perform only in the first segment and you are told it is overfitted.

Live-adjusted return

Every result is re-scored on real spread, commission and slippage. The gap to the flattering number is the point.

⚡Period Traderwalk-forwardexample view

development78

validation64

stress52

overfit detected: no

Why it matters

A strategy that only works on the data it was found in is not a strategy. It is a description of the past.

One connected workspace

Create, test, review and supervise.

AI chat and Create

Build strategies and indicators, request research, and review trades with an assistant that uses available platform data.

Decision AI

Structured entry/deployment checks, market labels, bot-health reviews, exit advice, manual opinions and alert sorting.

Research and validation

Parameter sweeps, backtest reports, Reality Score, walk-forward and live-cost checks remain the foundation.

Chart workspace

Analyze candles, use built-in or supported AI-created indicators, and prepare manual order previews.

Custom strategies

Create or import sandboxed Python for backtests and research. Custom strategies cannot currently go live.

Live controls

Eligible built-in strategies use the deployment path, with position sizing, broker-side stops and account risk controls.

Trade history and reports

Review closed trades, fees and deployment ownership. Ask AI to help explain records rather than invent results.

Structured Earn

Review supported exchange products, strike outcomes and allocation previews. Conversion risk still applies.

How it works

From an idea to an evidence-led decision.

Describe or choose

Use Create to explain an idea, or choose an existing strategy. Review proposed code, parameters and actions before approving.

Research and validate

Run a backtest or research sweep. Inspect costs, drawdown, holdout results and the validation verdict.

Review with Decision AI

Use structured assessments alongside the evidence. Shadow records; enabled Veto may refuse a flagged eligible entry or deployment.

Control and supervise

Choose whether eligible built-in strategies go live. Keep sizing, risk guards and auto-deploy settings explicit; use AI reviews as assistance.

Privacy and control

Local software. External AI processing.

Research and backtests run on your PC. Connected services still receive the information needed to do their work. AI is not an offline feature.

•Keep broker passwords and exchange API secrets out of prompts.

•AI requests can include messages, code and relevant account/market context.

•Voice input is sent for transcription when you use it.

•Review the Privacy Policy and app controls before sharing sensitive material.

Separate sharing control

Collective learning

Optional collective learning shares aggregate research statistics with a persistent pseudonymous client identifier. It is not guaranteed anonymous.

It is off by default. AI use and automatic error reporting are separate features with their own data flows; disabling collective learning does not disable them.

collective learning: off by default

The workflow

Assistance and evidence, in the same place.

Manual workflow

Period Trader

Create a strategy

Manual workflowWrite and check code by hand

✓Period TraderAI proposals; sandbox validation before saving

Explore parameters

Manual workflowRepeat tests individually

✓Period TraderResearch sweeps and comparable results

Review a result

Manual workflowRead metrics in isolation

✓Period TraderValidation verdicts, costs and walk-forward checks

Review a decision

Manual workflowSeparate notes and tools

✓Period TraderDecision AI assessments and an activity log

Choose risk

Manual workflowSet sizing and limits yourself

✓Period TraderYour selected sizing and account controls still apply

Processing

Manual workflowDepends on the tools you choose

✓Period TraderLocal research; external AI processing

Custom strategy use

Manual workflowDepends on the execution engine

✓Period TraderBacktest and research; not currently live