ABOUT

ORANGEORACLE

A Public AI Crypto Trading Experiment

About

OrangeOracle is an ongoing, public experiment in AI-driven crypto trading.

Background

I'm not a futures trading expert, never have been. But I've spent a fair amount of time testing strategies:

  • Technical indicator combos: MACD, RSI, Bollinger Bands
  • ML-based prediction models
  • Grid trading
  • Arbitrage
  • Copy trading

Each time, I thought I had something. Each time, I didn't.

What I Realized

After going through all of that, I hit an uncomfortable truth: AI has real limitations, yet people keep throwing it at one of the hardest problems in finance.

Unless you're already a professional in the field, this is genuinely difficult to pull off. That's not unique to finance. It applies across industries. Some fields just have lower barriers, which makes AI look more capable there.

Back to First Principles

A few days ago, I decided to stop tinkering and start from scratch.

Question one: Can you actually make money trading Bitcoin futures?

Yes.

There are traders who do it consistently, long-term. That's not speculation. It's documented.

Question two: So why do most people fail?

Because the ones who win consistently, the top 1%, aren't relying on luck. They have:

  • A complete, coherent trading system
  • Strict risk management discipline
  • Precise position sizing
  • The ability to execute without deviation

Question three: Do any of these traders share their methods publicly?

After digging through a lot of material: yes.

On Twitter, YouTube, Discord, and in various paid communities, there are traders who openly share their:

  • Trading frameworks
  • Live trade records
  • The reasoning behind each trade
  • How they manage emotions

These aren't people hiding their edge. They genuinely love trading and want to share what they know.

Question four: Can AI systematically learn from all of this?

Yes.

The Idea

If these top traders are willing to share, what if AI could systematically absorb everything they've put out?

Not to invent new strategies from scratch, but to:

  • Collect all publicly available content from consistently profitable traders
  • Break down their logic, risk rules, and position sizing frameworks
  • Help AI understand how they make decisions across different market conditions
  • Run those battle-tested strategies 24/7 through code

In essence: make AI an apprentice to the best traders in the world.

Strategy Alone Isn't Enough

Any basic understanding of finance makes this clear: what determines outcomes isn't just the strategy, it's position sizing.

Same strategy, different outcomes:

  • Small capital, high leverage: liquidation risk
  • Larger capital, low leverage: potential for steady compounding

So position management has to be part of the solution. And at its core, it's a math problem:

  • Expected value must be positive
  • Risk/reward ratio must be sound
  • Position size must adjust dynamically based on capital and risk tolerance

The Full Framework

I've broken this into three layers:

Layer 1, Strategy

Let AI learn from traders who've already been validated by real markets, rather than guessing from scratch.

Layer 2, Position Management

Use mathematical models to calculate appropriate position sizes based on capital, risk tolerance, and trade conditions.

Layer 3, Long-Term Edge

If the strategy's accuracy is high enough and the overall expected value is positive, there's a viable path to profitability over time.

What I'm Building

OrangeOracle is the experiment I'm running to test this idea.

It's not built in isolation, and it's not another “AI trading bot.”

What I'm actually doing:

  • Systematically collecting public material from top traders
  • Reverse-engineering their trading logic
  • Continuously validating against real market data
  • Using math to improve position sizing and win-rate structure

I want to see, with my own eyes, whether this holds up in practice.

The Core Thesis

OrangeOracle isn't about letting AI place random trades. It's about testing something more fundamental: can AI learn to think like elite traders, identify better entry points, then apply mathematical position management to improve long-term win rates?

If that works, then someone like me, an ordinary person, might actually have a shot at closing the gap with that top 1%.

That's the full story behind OrangeOracle. There are plenty of implementation details and engineering challenges I haven't covered here. If you're curious, visit orangeoracle.org to follow the experiment, every trade, every decision, in real time.