Search any NSE stock to run the Kronos forecast and get an entry or exit signal.
Research forecast only. The configured horizon comes from Wallet setup; a trade proposal must still pass the Audit, Council, and deterministic risk gate.
Last 60 sessions of history and the forecast path ahead.
Your watchlist, run through the entire desk and ranked into what to actually do today.
Runs all fifteen analysis agents — technical, fundamental, setup quality, news sentiment, market internals, smart-money footprints, options flow, macro, risk and more — and combines them into one weighted verdict tuned to your selected horizon.
The autonomous paper desk validates its active strategy configuration internally, manages stops and entries on its schedule, and sends morning/evening PDF reports. Quotes clearly label any delayed fallback.
Loading wallet status…
Read-only dashboard: the autonomous worker is the sole writer for paper fills, marks, stops, and frozen Rotation rebalances. Reports arrive as Telegram PDFs at approximately 08:35 and 18:30 IST on trading weekdays.
Hard limits: 1.5% wallet risk per trade, 4 positions, 35% single-name concentration, 150 monthly orders (hard runaway guard), a 15% drawdown halt for the Council sleeve, and a permanent desk shutdown after more than two consecutive losing completed calendar months or ₹0 equity.
The dashboard shows the active profile. Update it only from the owner Telegram chat; autopilot validates a new configuration without asking you to run a command.
At the scheduled entry cycle, the desk scans a bounded NSE universe, fully audits at most three candidates, and records at most one risk-gated Council paper entry per day. The browser cannot create or alter a fill.
A NO-TRADE, data-quality warning, or blocked risk decision records a receipt but creates no paper fill. The Council remains an experiment until 30 resolved A/B pairs select a winner.
| Sleeve | Symbol | Qty | Entry | Last | Market value | Stop | Target |
|---|---|---|---|---|---|---|---|
| No open paper positions. | |||||||
The decision log collects both paths. A winner is not selected until at least 30 resolved pairs exist.
Sweeps NSE stocks (real index constituents, fetched live from NSE) for high-quality VCP / breakout setups and ranks them A+ → Avoid. Caches for an hour.
Choose either a one-stock timing diagnostic or the portfolio Rotation sleeve, which explicitly sells dropped holdings and redeploys capital into replacements. Point-in-time data only — live-only inputs stay excluded.
A sale in this one-stock diagnostic becomes cash because no alternate historical candidate is loaded. It is useful for timing analysis, not portfolio allocation.
Each agent is graded on the horizon it predicts (next day to next month). Weights update automatically as outcomes resolve.
| Agent | Weight | Directional accuracy |
|---|
Every analysis is logged; actual prices at T+1, T+7 and T+30 grade the calls in the background.
| Date | Symbol | Rating | Start | T+1 | Status |
|---|
Each agent's accuracy split by market regime and by horizon. A cell needs 30+ graded calls before the desk trusts it over the general starting assumption — until then it shows progress, not a live weight.
| Signal | BUY means the model expects a rise beyond your entry threshold. SELL means it expects a fall. HOLD means the expected move is too small to act on. |
| Target price | Where the model thinks the price could be in about 30 days. A reasonable place to take profit. |
| Stop-loss | Your safety exit. If the price drops here, sell — no debating, no hoping. Small losses are the cost of doing business; big losses are fatal. |
| Volatility | How wildly the price swings. A calm stock might move 15% a year; a wild one 50%. More swing = more risk in both directions. |
| Kronos | The AI model behind the forecast — a neural network trained on millions of price charts, running privately on this computer. It reads candlestick patterns the way a chartist would, only faster. |
| Rating scale | Strong Buy → Buy → Neutral → Watchlist → Reduce → Sell. Watchlist means interesting but not yet; Reduce means trim your position. |
| Confidence | How much the fifteen agents agree. 50% = coin flip, 65%+ = strong consensus. High confidence can still be wrong. |
| Suggested size | A configuration-validated half-Kelly candidate. The hard risk gate calculates the final paper quantity and may shrink it to zero. |
| Weight (w) | Each agent's voting power. Starts at 1.0×; rises when an agent keeps being right, falls when it keeps being wrong. The market regime also boosts or mutes agents. |
| Evidence / Risks | Every agent must show its work — the exact numbers behind its vote, and what could go wrong. If the evidence looks thin, trust the verdict less. |
| Market Regime | Reads overall market weather (bull, bear, sideways, crisis) and decides which other agents deserve a louder voice right now. |
| Fundamental | Checks the company's financial health — is the business growing, profitable, and not drowning in debt. |
| Technical | Reads the price chart: trends, momentum, and trading volume, to judge short-term direction. |
| Quantitative | Runs statistics on the stock's returns — volatility, worst-day risk, and whether the odds mathematically favor holding it. |
| Macro | The big picture: Nifty's direction, interest rates, and the rupee — the tide that lifts or sinks all boats. |
| Valuation | Asks whether the stock is cheap or expensive compared to what the business actually earns and owns. |
| Sentiment | Reads recent news headlines (with the Kimi AI model when available) to judge whether coverage is positive or negative for the price. |
| Market Internals | Sentiment measured from money, not words: India VIX (fear gauge), FII/DII institutional buying and selling, and how many stocks are rising vs falling. |
| Setup Scanner | Grades the chart the way momentum professionals do — Stage 2 trend, leadership vs Nifty, tight quiet base, volume dry-up — into A+, A, B or Avoid. |
| Options Flow | Follows what options traders are betting — where big money expects the price to be pinned or pushed. |
| Smart Money | Hunts institutional footprints: disclosed bulk/block deals, delivery percentage, heavy volume absorbed without price movement, and money-flow diverging from price. Big players hide their orders, not their wake. |
| Risk | The safety inspector: flags dangerous volatility spikes, heavy debt, or liquidity problems. Bullish from this agent means safe. |
| Portfolio Fit | Checks whether this stock would actually diversify your portfolio or just duplicate what you already own. |
| Forensic | Looks for accounting-quality and governance red flags that can make an apparently attractive business unsafe to own. |
| Kronos AI | The foundation model's own 30-day price forecast, generated from raw candlestick data on this machine. |
| Equity curve | Growth of ₹100 over time. The strategy line should sit above the buy-and-hold line, with shallower dips. |
| CAGR | The result expressed as a steady yearly growth rate, so different periods can be compared fairly. |
| Sharpe / Sortino | Return per unit of risk. Above 1 good, above 2 excellent, negative bad. Sortino counts only downward swings as risk. |
| Max drawdown | The worst fall from a peak before recovery. The number that tests your stomach, not your math. |
| Profit factor | Total money won ÷ total money lost. Above 1.0 = net profitable. |
| Look-ahead bias | The classic backtesting cheat: accidentally using information from the future. This desk avoids it by excluding news, social, and options agents from replays — so the backtest is honest, and a floor rather than a ceiling. |
| Directional accuracy | Of the times an agent said up or down, how often was it right. 50% is a coin flip; even 58% sustained is a real edge. |
| T+1 / T+7 / T+30 | Price one day, one week, and one month after the analysis — each agent is graded on the horizon it actually predicts. |
| CORRECT / MISSED | Whether the desk's overall rating pointed the same way the price actually went the next day. |
| Two sleeves | The frozen Rotation sleeve and experimental Council sleeve each begin with ₹5,000, so their results can be compared fairly. |
| Risk gate | Hard Python rules can shrink or veto a Council proposal, but can never create a trade. It also blocks new Council entries after the wallet drawdown limit. |
| Config hash | The paper wallet starts only when its exact strategy configuration has a matching multi-stock validation artifact, preventing untested rule changes. |
Send /webkey to the configured owner Telegram bot, then enter the key here.
The key stays only in this browser tab. Typed chat works on the current site; speech input will require HTTPS.
