TradingAI 12.0 ~Nexus — Devlog

What happened since the last devlog: three models became a whole system and it’s running live.

What changed

The 8.0 system could predict, but it couldn’t trade. So it grew into a two-stage system:

Stage 1: 39 candle models. Instead of one timeframe, now 13 (from 1 minute to 1 day), each with Close, High/Low, and Volume variants. Every model looks at the last 200 candles and rolls its predictions up to 100 steps forward, the familiar ChatGPT principle, just on 13 time levels at once. Trained on 5 years of Binance data (~2.6M candles).

Stage 2: 6 signal models. The predictions from all timeframes feed into six classifiers (1m to 1d). Each learns a single question:

Will my take-profit or my stop-loss be hit first?

Answer: BUY / SELL / HOLD plus a confidence value that tells you how sure the AI is of itself.

**The server is running.**live Binance data, AI prediction in real time, TradingView-style chart, paper trading with fees and stats. The legacy of 8.0, just with 45 models behind it instead of three.

Still everything local, offline, no cloud, as it has always been.

Why so many SL/TP tests and timeframe signals?

Because the AI predicts direction it doesn’t tell us how far the move will go. And that’s where the real money is decided.

Why 100 SL/TP combinations. A tight target (+0.3%) is hit often but earns little; a wide one (+7%) earns a lot but rarely hits. The same signal can be profitable with one setting and ruinous with another. Winning often doesn’t mean making money ,the win rate only matters together with the risk/reward. So instead of guessing “maybe 1% is fine,” we simulate every combination on real price paths, subtract the fees, and let the numbers decide. No gut feeling, just math.

Why a separate signal per timeframe. Each timeframe is a different game. The 1m signal trades hundreds of times a day — lots of trades, small moves, high fee impact. The 1d signal trades rarely ,few trades, big moves, more risk per single decision. A setup that works on 5m can be a disaster on 4h. Every timeframe gets its own model and its own SL/TP test, so we find out independently which horizons actually have an edge.

What’s next

The first complete test run was too good to be true: +1.79% per trade at a 99.6% win rate. The analysis showed why: the models were trained on the entire history including the test period. The AI knew the answers. A classic leak.

The plan:

1. Retrain all models with a hard training cutoff, so the test period stays unseen for every model
2. Simulate all SL/TP combinations on that honest time window, fees included
3. Only a setup with real net profits gets switched to live

The AI has to prove it.

Built by tradersquant — TradingAI 12.0 ~Nexus, August 2026

BTC Prediction AI – Devlog

I’ve had the idea of building my own AI for BTC predictions for a while. Not that whole “AI will make you rich” stuff, but just a tool that gives me a second opinion on the chart. After a few months of tinkering, the first version is done – and I’ve released it on GitHub for anyone to test.

This isn’t a tutorial or financial advice. It’s just a devlog explaining what I built and why.

What does it do?

It loads Bitcoin candles from Binance (1-minute chart, ~250,000 candles) and displays them in a chart that feels like TradingView. Three small AI models look at the last 200 candles and try to predict the next single candle:

• Close model: What will the price be at the end of the minute?
• High/Low model: How much will the price swing back and forth during the minute?
• Volume model: How many people are buying/selling in this minute?

All three work together: first Close, then High/Low as offsets to it, then Volume. The prediction is fed back as a new candle and the next step begins – using the same principle as ChatGPT, creating 100 predicted candles in one go.

Since it’s a replay, you can scroll through history, turn on the AI, and see if it was right back then. There’s also Paper Trading: Buy, Sell, Close – with a win/loss counter.

Features

• Replay: Walk through 250,000 BTC candles at adjustable speed
• AI Prediction: 100 candles into the future, shown as white/blue candles
• Paper Trading: Long/Short with play money, including fee input and stats
• Indicators: Volume Delta + SuperTrend Oscillator (toggle on/off)
• Live Chart: Real Binance data + AI in real time (/live_chart)

Tech

• Data: BTCUSDT 1m from Binance, ~250k candles, 10 features (Close, High, Low, Volume, MA10, MA20, BB, EMA, OBV)
• Training: Close (18 epochs), Volume (36), High/Low (180) – trained separately on CPU
• Server: Python HTTP server (no Flask etc.), port 8765
• Frontend: lightweight-charts (TradingView-like), Vanilla JS, ~300 lines

Try it yourself git clone https://github.com/tradersquant/BTC-PredictionAI.git cd BTC-PredictionAI pip install -r requirements.txt python trading_server.py

Then open http://localhost:8765 in your browser.

What’s next

The current version is a first draft. Three simple nets with 10 features are far from the end goal:

• More inputs: Order book data, funding rate, liquidation levels, macro data
• Larger networks: Deeper architectures, maybe transformers or attention
• Ensemble: Multiple models running in parallel, validating each other
• Better features: Automated pipeline that tests what actually has predictive power

But for now, the foundation works. Everything runs locally, offline, no API costs, no cloud dependency.

Built by tradersquant – BTC Prediction AI, July 2026

Currently, the AI uses 14 input features. I am considering increasing this to around 80 inputs. While this could potentially improve prediction accuracy, it also introduces a significant risk of excessive noise, which may ultimately degrade the model’s performance instead of enhancing it.