Changelog
What's new in the platform, newest first.
Workspace redesign and more reliable signals
- ChangedA new geometric UAP logo pairs a signal-inspired mark with a custom wordmark, with a matching browser icon and reusable dark and light versions.
- ChangedA redesigned graphite-and-lime workspace brings persistent navigation, a clearer market overview and larger text across the feed, track record and scoring guide.
- ChangedTrade cards now put entry, target, stop and AI context first. Expand the score breakdown for the five metrics and scalp reference, with live checks and position monitoring still close at hand.
- ChangedMarket tabs and compact filter controls make it easier to narrow the feed, with a responsive layout for phones and tablets.
- FixedScoring and outcome checks now use completed candles, preventing unfinished price bars from influencing signals or resolving tracked calls too early.
- FixedFinal candle values replace earlier stored snapshots when fetched again, so incomplete price history can be corrected. Previously resolved calls are not recalculated.
- FixedThe four-hour trend filter now uses complete candles aligned to UTC boundaries, keeping its time windows consistent between scans. Backtests use the same aggregation rules.
- FixedScanner, AI, alert and monitor settings now reach the worker in Docker Compose deployments, including the switch for stocks and forex.
- ChangedSetup guidance and example settings now reflect the current scanner defaults, optional AI, and differences between backtests and the live paper track record.
Scalp take-profit on calls
- ChangedThe live monitor now streams real-time prices (~1s, direct from Binance) instead of updating every 20 seconds — entered trades tick live, so you can watch a scalp reach its target as it happens.
- AddedEach call now shows a fixed 1% take-profit level (≈ +10% at 10× leverage) next to the engine's own target — an aid for manual leveraged trading. It doesn't change what the system trades or how outcomes are tracked.
Higher-timeframe alignment
- AddedSetups must now agree with the 4-hour trend — calls fighting the bigger picture are dropped before they ever surface. Backtested and walk-forward-validated: fewer but markedly better trades (expectancy up ~80%, profitable in 5 of 6 periods).
Tighter loss control
- ChangedLosing calls are now cut at the stop (1× risk) instead of running to 1.5× risk before being closed — every failed trade loses about a third less, and the live results now match the validated backtest.
Structure-based trade levels
- ChangedStops now sit just beyond the recent swing high/low — where the thesis is genuinely wrong — instead of a flat ATR distance, with the target at 1.5:1. Backtested and walk-forward-validated (profitable in 5 of 6 periods vs 3 of 6 for the old bracket), with a markedly higher, steadier win rate.
Position monitor
- AddedPosition monitor — an early-warning Telegram alert the moment an open high-conviction thesis starts heading for its stop (before it's hit), plus a heads-up as it nears target. Checks live prices every couple of minutes.
Track record focus
- ChangedTrack record headline is now scoped to high-conviction theses (score 60+), so it reflects the quality calls rather than being diluted by marginal setups.
Broader scan, focused feed
- ChangedScans now cover the top 150 coins by volume (up from 40), for broader coverage and faster data accumulation.
- ChangedFeed now shows high-conviction setups (score 60+) by default, with a “60+ / All” toggle to reveal the lower-scored ones on demand.
Backtesting, validation & polish
- AddedBacktesting harness that replays the live scoring engine over real historical candles — measuring win rate, expectancy, and whether a higher score actually means a better trade.
- AddedWalk-forward validation: performance is broken out period-by-period, so a result has to hold up consistently rather than on the average of one lucky stretch.
- AddedPoint-in-time universe in the backtest — each coin is ranked by its liquidity at the time of the trade, removing the survivorship bias of judging history by today's winners.
- AddedMarket-regime filter (ADX + trend structure): signals are gated to context — no counter-trend entries during strong trends, no trend-following in choppy ranges.
- AddedTrack record now breaks outcomes down by score bucket — the live, forward test of whether higher-scoring setups actually win more.
- ChangedTrack record now reports performance as percentages only — no longer distorted by a flat paper-stake figure, so it reflects results cleanly.
- ChangedTelegram alerts now fire only on higher-scoring setups (score 60+), cutting the noise to the calls actually worth a look.
- ChangedResponsive layout — the dashboard now adapts cleanly to phones and tablets, with the feed filters tucked behind a single “Filters” control on small screens.
- AddedThis changelog.
Live monitor sidebar
- ChangedThe live trade monitor is now a docked right-hand sidebar rather than a floating overlay, so it never covers the feed.
Alerts, monitoring & tighter signals
- AddedTelegram alerts with a rendered price chart and a full card (score, entry/target/stop, R:R, signals, news verdict, thesis) for every new setup.
- AddedLive monitor — mark a trade as entered and watch it tick live with running P&L, progress-to-target, and target-hit / stop-out notifications.
- AddedPer-trade “check” button to re-test whether a setup is still takeable between scans.
- AddedBinance futures positioning signals (funding, open interest, long/short ratio, taker flow) and a Fear & Greed regime tilt on the score.
- ChangedScans now run every 30 minutes; the feed defaults to newest-first.
- ChangedThe track record now stakes paper capital only on higher-conviction crypto theses.
- FixedEntry state correctly flips to “invalidated” once price has run past the stop.
- FixedThe news/risk gate no longer shows false “search unavailable” cautions.
Platform launch
- AddedUniversal Opportunity Platform is live: a multi-market scanner with a deterministic five-metric scoring engine and cost-tiered AI narration.
- AddedMarket-intelligence dashboard, a self-scoring track record (each thesis tracked to target or invalidation), and a scoring-methodology page.