This Week's Comprehensive Market Analysis
Loading comprehensive market analysis with real-time data provider server data...
AI Equity & Stock Market Forecast
Equity — Quantitative Target Prices
Two models, one public scoreboard. Every asset gets a BUY or SELL side with the odds of finishing higher over the next 5 trading days, the limit price to bid and the stop that protects it — refreshed every trading day.
Loading...
Target: —
Publicly scored — misses included
Loading today’s answer…
Loading forecast data…
What to do today (5 trading days ahead · two models, one scoreboard)
Strategy — BUY above 50% odds of finishing higher, SELL below. Limit orders only,
stop attached, 1% of equity at risk per trade.
The order geometry alone loses net of costs
(8,770 walk-forward trades). It says where to bid, not whether to — only a real
directional edge pays for it, and the live ledger is measuring whether ours has one.
The part that is profitable is elsewhere: a volatility-targeted book,
Sharpe 1.11 over eleven years, −14% worst drawdown, no forecast —
The Portfolio →
Read this first — what the numbers mean, and what “unvalidated” means
“Unvalidated” is a status, not legal boilerplate. It means this model has not yet beaten a coin flip on a public, out-of-sample, scored record. It keeps a column here because it is being measured in the open: every prediction is booked at its target date, hits and misses. If the live hit rate stays at 50%, the honest conclusion is that the column is worth nothing — and this page will keep saying so. If it climbs meaningfully above 50% over a few hundred scored calls, it graduates. Information, not instruction.
Why no probability here exceeds 55%. Taken literally, our model's raw output sometimes implies “83% chance this falls this week”. Nothing measured on this site supports that confidence: the scored directional record is ~48% over 63 calls, the Kronos backtest was 50.1% over 1,086 predictions, and four pre-committed alpha studies all failed their bars. So every tilt is published on a squashed 45–55% scale — the ranking survives, the overclaim does not. Hover any probability to see the uncapped number.
The three numbers that are genuinely calibrated are the fill odds (how often price comes down/up to your limit), the 80% price range, and the volatility behind both. Those come from 20,000 simulated paths built on each asset's own return history and are verified by walk-forward coverage tests. A probability of direction is a forecast; a probability of travel is arithmetic on volatility — the second is much more reliable, which is why the orders are built around it.
How to trade a row. Place the limit at the entry — never a market order. If price never comes to you, you have no trade that week, and that is a normal outcome. The moment you are filled, set the stop and the take-profit. Risk at most 1% of the account between entry and stop, so every position carries identical risk whatever the asset's volatility.
Model 1 — Swiss Quant (our engine: momentum, volatility, cross-asset and macro features)
Model 2 — Kronos (open-source candlestick foundation model, AAAI 2026, run daily on this server)
Move Probabilities
Where price is likely to end by the horizon: the chance it finishes up at least X%, stays inside ±X%, or finishes down at least X% — the three always add up to 100%. Hover a chip for the touch odds: the chance price trades through that level at least once during the window. Touch odds are what fill a limit order, which is why they can exceed the end odds and why up and down can both happen in the same week.
Loading real-time market data from data provider server...