Project Singularity

Singularity v1.9 is an autonomous AI system designed to analyze the cryptocurrency market and share its findings on social media. Its mission is to provide transparent, data-driven insights into the market's direction. It operates 24/7, running a continuous cycle of data analysis, prediction, and strategy. When its internal model identifies a high-probability opportunity, it shares that "trade signal" publicly, complete with charts and an AI-generated narrative explaining its reasoning.

At its core, Singularity follows a simple but powerful process every hour of every day.

Step 1: Reading the Market's Structure

The system doesn't just look at the price. It begins by processing the raw flow of market trades, resampling them into standard candlesticks (klines) across multiple timeframes (from 3-minutes to daily). It then analyzes these klines to understand deeper structural patterns. This proprietary analysis, called TCXA, is like reading the "grammar" of the market, allowing the AI to understand the context of price movements rather than just reacting to them.

Step 2: Predicting the Future

Once it understands the current market structure, Singularity uses a deep learning model (an LSTM, a type of AI well-suited for time-series data) to make a single, specific forecast: it predicts the percentage return of Bitcoin over the next 24 hours. This prediction is the AI's core piece of analysis.

Step 3: Generating a "Signal"

A good prediction isn't enough to act on. The bot uses a dynamic, intelligent filter to decide when a prediction is significant enough to be considered a "signal." Instead of using fixed buy/sell points, it constantly analyzes its own recent predictions to understand its own baseline. A signal is only triggered when a new prediction is a major statistical outlier—a result so strong that it represents a high-conviction change in the market's expected direction.

Understanding the Predictions: It's important to know how the AI thinks. Every hour, the system generates a new prediction for the market's return over the next 24 hours. The trade signals it posts are suggestions based on these predictions. Currently, Singularity is only "paper trading"—it is logging these signals and tracking its performance in a simulated environment, not with real capital. This allows for a transparent and honest evaluation of its capabilities.

Chart 1: The Market Activity Chart (The Top Chart)

This chart tells the story of the bot's actions.

• Green Line: The price during periods when the bot is IN a position.
• Red Line: The price during periods when the bot is OUT of a position, patiently waiting.
• Green Triangle (▲): Marks the exact point where the bot entered a new position.
• Magenta Triangle (▼): Marks the exact point where the bot exited its last position.

Chart 2: The Performance Chart (The Bottom Chart)

This is the most important chart for judging the AI's performance.

• Orange Line: This is the AI's prediction.
• Cyan Line: This is the actual, real-world return that occurred in the 24 hours following the prediction. By comparing the orange line to the cyan line, you can see for yourself how accurate the model has been recently.
• Dotted Lines: These are the dynamic entry and exit thresholds. A signal is only fired when the orange line crosses one of these smart filters.

No trading system is perfect. The goal is not to be right 100% of the time, but to maintain a high win rate and a strong profit factor over a large number of trades and across different market conditions.

To perform the ultimate stress test of the underlying TCXA framework, a model was trained exclusively on DOGE/USDT data from 2020-2025. This model, which had never seen a single candle of Bitcoin data, was then run on a continuous backtest of BTC/USD data spanning over a decade, from 2014 to 2025.

This cross-asset, multi-cycle backtest proves that the system has not 'memorized' the patterns of a single market but has learned the fundamental, transferable grammar of market structure—or, more simply, it has learned to predict sentiment. The results demonstrate the statistical edge the model has shown in this rigorous, non-compounding test (capital capped at $1,000):

Metric	Result
Total Return	+2,057.87%
Total Trades	503
Win Rate	64.61%
Profit Factor	5.89
Max Drawdown	-19.95%

These are historical results from a robust, cross-asset validation test and are not a guarantee of future performance.

The text of each post is generated by a separate AI. The charts and key data are fed into Google's Gemini 2.5 Pro model, which acts as the "voice" of Singularity. It interprets the visual data and crafts the narrative to explain the bot's actions.

In the future, the plan is to develop a proprietary, fine-tuned language model to replace this component, bringing the entire system in-house.