Live — paper trading since Feb 2026

AI market intelligence,
one cent per day

27 capabilities across regime detection, LLM alpha, portfolio optimization, alternative data, and multi-timeframe signals. Built by a solo developer. Free on Telegram.

27
Capabilities
20K+
Lines of code
85+
ML features
$0.01
Cost / day
apexquant — multi-timeframe scan
Not a chatbot. A quant system.
Every layer does what it's best at: LLMs for qualitative selection, math for sizing, rules for risk, ML for regime detection.

4-regime detection

XGBoost classifies the market as Trending Bull, Mean-Reverting, Stress, or Crisis. Every downstream decision adapts — position sizing, stop distances, sector limits, strategy weights.

Core

LLM alpha generation

Each symbol gets an AI research analyst with full context: technicals, fundamentals, news sentiment, short interest, options flow, earnings revisions, multi-timeframe alignment, and portfolio state.

Core

Portfolio optimization

Four methods: max Sharpe, min variance, risk parity, and Kelly criterion. The LLM picks what to hold. Covariance math decides how much. No single name dominates portfolio risk.

Enhanced

Multi-timeframe signals

Daily bars say what to buy. 1-hour and 15-minute bars say when. Cross-timeframe alignment, VWAP positioning, session patterns, and institutional accumulation/distribution detection.

New

Alternative data

Short squeeze scoring, put/call ratio, IV skew, unusual options activity, analyst revision tracking — signals orthogonal to price that capture information not yet in the chart.

New

Walk-forward validation

Every ML model tested on data it has never seen. Multi-fold time-series CV with purge gaps. Overfit detection flags models before they touch capital.

New

Kelly criterion sizing

Bankroll-optimal bet sizing from edge estimate and variance. Half-Kelly scaling for estimation error. Higher alpha + lower volatility = bigger position. Math, not vibes.

New

Adaptive trailing stops

ATR-based stops that tighten in bull markets and loosen in stress. Ratchet up as profits grow. Profit lock at 5%+. Never moves down. Regime-aware multipliers.

New

Ensemble models

XGBoost + LightGBM + neural net stacked per regime. Ridge meta-learner combines them with walk-forward validated weights. Different algorithms capture different patterns.

New

Sector exposure control

72 symbols mapped to GICS sectors and correlation buckets. Limits tighten during stress. Pre-trade compliance gates. Pre-market contagion detection across the portfolio.

Core

PCA factor model

Decomposes returns into market beta, sector factors, and true alpha. Answers: how much of your return is from stock picking vs just being long the market?

New

Transaction cost analysis

Almgren-Chriss market impact model. Estimates spread + impact + commission before every trade. Filters unprofitable rebalances. Self-calibrates from realized fills.

New
Seven layers, fifteen minutes.
Every trading day at 3:30 PM CT. Each layer feeds the next.
1Data ingestionOHLCV + FRED macro + SEC filings + alt data
2Regime detectionXGBoost classifies Bull / MR / Stress / Crisis
3Multi-timeframe analysis15m + 1h + daily alignment + session patterns
4Alpha generationLLM per symbol with full portfolio context
5Portfolio optimizationRisk parity / Kelly sizing + sector constraints
6Risk + complianceTCA + adaptive stops + SLPS tiers + heat check
7ExecutionPaper fills at real prices → TimescaleDB
Built lean. Runs daily.
Daily cost
$0.01
Pipeline time
~15 min
Symbols / day
20
AI agents
4 / symbol
ML features
85+
Codebase
20K+ lines
Timeframes
3 (15m/1h/1d)
Regime models
4 states
Ensemble
XGB+LGB+MLP
Validation
Walk-forward
Optimizer
4 methods
Data history
5 years

Free daily signals on Telegram

S&P 500 top movers, macro regime, market pulse, multi-timeframe alignment — every trading day. No spam, no paywalls.

📈 Join @apexquant_signals