EdgeLint
Pre-launch · Building in public

Test the idea before it touches your account.

EdgeLint is a set of AI agents that turn prop firm rules and trading ideas into sourced rulebooks, testable strategy specs and strict backtests, then check every strategy against your account limits before anything runs. You stay in control of every decision.

Early access → See the pipeline

The problem

Most strategies fail for reasons you could have checked first.

Retail futures traders try ideas from videos, forums and courses. Many of those ideas break down for predictable, testable reasons long before market conditions are even the question.

costs

Fees and slippage

A backtest that ignores commissions and realistic fills can make a losing idea look attractive.

bias

Lookahead and overfitting

Tuning on the same data you evaluate on, or using information that wasn't available yet, produces results that don't hold up.

rules

Account rule breaches

Prop firm drawdown, daily loss, session and consistency rules can end an account even when the strategy itself is fine.

How it works

Five stages. Each one is a gate.

A strategy only moves forward when the previous stage passes. Nothing reaches a live account without clearing simulation and the risk guard.

  1. Rule research

    Agents read a prop firm's or broker's published rules and build a rulebook where every line links to its source.

    sourced

  2. Idea intake

    A trading video is broken down from transcript and chart frames into a testable strategy spec.

    labeled

  3. Strict backtest

    Realistic costs, no lookahead, walk-forward plus a sealed holdout, minimum trade counts and bias checks.

    validated

  4. Risk guard

    Every strategy is checked against the account's drawdown, per-trade loss, session and consistency rules.

    compliant

  5. Execution

    Automated execution on the trading platform, only after the strategy passes simulation.

    simulated first

What's different

Evidence over confidence.

Every output says where it came from and how sure it is, so you can judge it yourself.

Verified rulebooks

Account rules are extracted with citations to the firm's own pages, not paraphrased from memory.

Evidence-labeled strategy specs

Each rule in a spec is marked as stated in the video, shown on the chart, or inferred, so guesses are never mistaken for facts.

A strict backtest standard

One written standard for costs, data splits and bias checks, applied to every strategy the same way.

Risk guard before execution

Account limits are enforced as a hard check, not a reminder.

strategy_spec.exampleillustrative
  • saidTrade only during the New York morning session.
  • shownEntry on a pullback to the prior swing level, as drawn on the chart.
  • saidStop goes below the most recent swing low.
  • guessedExit rule not stated; assumed fixed 2R target. Needs review.

Example format only. Not a recommendation and not a tested strategy.

Status

Pre-launch. Building in public.

There are no users, customers or performance results yet. Here is exactly where things stand.

  • DonePrivate prototypeProp firm due-diligence dossier, sourced rules table, data plan, backtest standard, and agent skills for rule checking, video-to-strategy breakdown and backtest review.
  • NowFirst end-to-end backtestRun the first strategy through the full backtest standard and publish the method, whatever the result.
  • NextRisk guard in simulationEnforce account rules against simulated trading before any automation.
  • LaterEarly access for a small group of tradersResearch and risk-control tooling first; execution features only after they are proven in simulation.

Who's building it

Dipsy
Solo founder, Indonesia