# THE DAILY PLAN (TDP) — System Overview
## What This Is
The Daily Plan is an automated trading intelligence system built for NQ (Nasdaq futures) day trading. It combines institutional-grade market data, AI-generated trade plans, real-time price monitoring, and a self-learning journal — all running continuously and posting directly to a structured Discord server. It is designed so that someone with zero trading experience can follow its alerts and trade profitably from day one, while gradually learning the underlying methodology, and so that an experienced trader can use it as a disciplined second set of eyes that never gets emotional, never hesitates, and never deviates from the rules.
## Core Methodology
The system is built on a top-down, multi-timeframe structural approach — not indicators alone, not breakout chasing, and not opening-range strategies (which were tested and intentionally excluded after backtesting showed they were not consistent enough on NQ). Every trade plan drills from the Weekly chart down through Daily, 4-Hour, 1-Hour, 30-Minute, and 10-Minute charts to find "brick wall" zones — price levels with a proven history of holding or reversing.
Every potential entry requires five confirmations to align: MACD, Volume, VWAP, RSI, and the Keltner Channel middle. There are three valid ways to enter a trade: a limit order placed directly at a brick wall (only valid on Weekly/Daily-level zones), a rejection/hold confirmation on a 10-minute candle close, or a breakthrough of a level followed by a retest confirmed on a 10-minute close. Every trade carries a hard 30-point stop loss and three profit targets at 60, 100, and 150 points, with built-in staggered entry logic (splitting entry across two prices when a zone can't be narrowed to one precise level) and add-to logic (adding to a winning position only after a confirmed breakthrough and retest).
## Data Infrastructure
The system pulls live and historical market data from two professional-grade sources. Rithmic provides direct futures data for NQ, ES, and CL — including real-time tick-by-tick price action and historical replay covering up to several years on daily/weekly timeframes and meaningful depth (weeks to months) on faster timeframes. IQFeed supplies market internals and correlated instruments: VIX, VXN (volatility), UVXY, DXY (dollar index), TNX (10-year yield), TICK, ADD (NYSE breadth), Put/Call ratio, and a basket of tech leaders and indices — NVDA, MSFT, AAPL, AMZN, META, QQQ, and SOX — across multiple timeframes.
As of this build, the system also maintains its own permanent historical database. Every 15 minutes, all market data currently being tracked is saved to a local database that persists across restarts — meaning the system is continuously building its own growing historical archive of every instrument it monitors, independent of how much history any single data request can pull at once. This archive grows larger and more valuable every single day the system runs.
## Plan Generation
Six trading plans are generated automatically each day, one for each market session: London (2:00 AM), Pre-Dawn (5:00 AM), Pre-Market (8:15 AM), Morning (9:30 AM), Afternoon (1:00 PM), and Tokyo (6:30 PM Central Time). Each plan is generated by an AI model (Claude) using a detailed prompt built from real-time volume profile analysis, HVN/LVN (high and low volume node) detection across four timeframes, pivot zone calculations, trend alignment scoring, economic calendar data, and a live market regime classification (explained below). Each plan only proposes trades valid within its own session window — a morning plan will not generate trades meant for the afternoon.
## Real-Time Trade Monitoring
Once a plan is posted, the system actively monitors live price action every minute. As price approaches a planned entry level, it sends a sequence of alerts: an early warning at 100 points away, a "get ready" alert at 50 points away, and a full "at the level" alert when price arrives, including the entry, stop, all three targets, and a live confidence score. Every single test of a level is recorded and alerted — there is no longer a hard rule blocking trades after multiple tests, since the system is built to take every signal it generates and let real outcome data determine over time whether such rules are even necessary.
## Intelligence Layer (Recently Added)
On top of the core trade alert, four additional intelligence checks now run automatically:
A regime classifier evaluates VIX level, overnight price range, and TICK extremes before each session plan generates, tagging the day as Trending, Rotational, Breakout, or Mixed — this context is fed directly into the AI's plan generation so trade type selection accounts for current market conditions rather than treating every day identically.
A volume acceleration detector checks whether trading volume is increasing or fading as price approaches a planned level, distinguishing levels with real buyer/seller participation behind them from thinner, less reliable ones.
A cross-instrument confirmation check evaluates whether ES futures, NVDA, and QQQ are trending in agreement with or against a trade's direction at the moment it triggers, appending a confirmation or conflict note to the alert.
An economic news gate checks a live economic calendar feed in real time — not just once at plan generation — and flags any trade alert that falls within 20 minutes of a high-impact USD economic release, since news can break mid-session after a plan has already posted.
A weekly drawdown monitor reviews the last five trading days each afternoon and posts an informational note if losses cross a meaningful threshold — this never blocks trading, it simply provides visibility so unusual stretches don't go unnoticed.
## Self-Learning Components
Every trade — win, loss, or in progress — is logged to a structured database along with the full market context at the time: RSI, MACD, volume, VWAP position, VIX level, NVDA/ES/QQQ trend, TICK and ADD readings, and which of the five confirmations were met. A Pattern Discovery Engine runs automatically every Friday afternoon, analyzing this trade history for repeatable setups with 85% or higher historical accuracy, and posts any findings to a dedicated Discord channel. Over time, as more trades accumulate, this allows the system to identify which setups, sessions, and conditions produce the most reliable results — separate from the model's general AI judgment.
## Discord Integration
The system posts to over a dozen dedicated Discord channels covering daily trading plans, real-time trade alerts, trade management updates, market internals, session summaries, support/resistance levels, weekly bias, overnight session plans, pattern discovery results, the economic calendar, high-impact news, daily edge summaries, and a wins/losses results journal that only displays resolved trades. A separate journal channel posts an AI-generated narrative each day explaining what happened in the market, why NQ moved the way it did, which instruments drove the move, and what to watch the next day — going beyond raw numbers to actual analysis.
## What Makes This Different
Unlike many automated trading alert systems, this one does not simply react to price crossing a static line. It requires structural proof a level has historically mattered, requires multiple independent indicators to align, accounts for the broader market regime before ever proposing a trade, checks correlated markets for agreement in real time, watches for news risk continuously rather than only at plan generation, and records everything it does so its own performance data — not assumptions — eventually determines which rules are worth keeping. It is built to operate without requiring the operator's constant attention, while remaining fully transparent about its reasoning at every step.
## Current Status and Roadmap
The system is fully operational and considered production-ready for live trading. Near-term planned additions include time-of-day confidence weighting (once sufficient live trade data accumulates to make it statistically meaningful) and a predictive cross-instrument pattern engine that would analyze a full year of historical data across all monitored instruments to identify the specific conditions that have historically preceded large NQ moves before they happen — this is dependent on expanding historical data depth on the non-futures instruments, which is currently growing daily through the system's new persistent database.