Evidence behind the tools
VEKTOR is a market-research workspace. Its tools explain observed data and hypothetical trading scenarios; they cannot guarantee future prices or profitable trades.
Exchange coverage
Public spot adapters: Binance, Bitget, KuCoin, OKX, Bybit, Gate, MEXC, Poloniex. Coverage varies by token and provider availability. An automatic fallback uses one actual source and labels it. Adding exchanges expands coverage; it does not turn correlated indicators into independent evidence or combine all books into a single tradable price. Unrelated assets can share ticker symbols.
Activity heatmaps
Hourly heatmaps group completed candles by hour and weekday in your chosen timezone. Quote volume measures historical activity; the high–low range and close-to-close return volatility measure different aspects of price movement. Lookbacks request up to 90 days; available history, candle counts and gaps determine actual coverage. An hour with historically high activity is not a promise of volatility at that hour today.
Rules-based Predictor
Up to 300 completed spot candles from a single exchange are analyzed. Directional alignment uses net votes from EMA trend, MACD histogram, RSI above 55 or below 45, and rolling VWAP, with ADX at least 20. These signals are correlated. Without sufficient alignment the result is “No clear setup.” At least 60 completed candles are required; EMA 200 needs 200. Invalid, gapped or stale data blocks fresh analysis.
A directional scenario uses the last completed close as a hypothetical entry, a stop 1.5 times ATR away, and targets 1.5 and 2.5 times that risk. These are planning levels, not executable quotes or trained AI price forecasts. No validated win rate or probability is claimed.
Understanding green, red and unscored results
The original TP1 and stop are evaluated over the next 24 completed candles on the original exchange. TP1 before the stop is green. Stop first, or no target by the deadline, is red. Missing coverage, no direction, or both levels hit in one candle are unscored. OHLC candles cannot tell which level was touched first within the same candle.
Illustrative example—not a live prediction
Suppose a long scenario has entry 100, target 106 and stop 96. If completed candles reach 106 before 96, its price-level test is green. If 96 is reached first, it is red. A candle with high 107 and low 95 is unscored because order of touches is unknown. A green result is not evidence of net profit after fees, funding or slippage. Your actual saved predictions show their original levels and observed results in Predictor.
Order books, headlines and trading costs
Demand/supply context is a sampled spot order book, with displayed source and receipt/provider time—not a promise that orders will stay in place. Spoofing, cancellations and latency can change it. Related headlines have linked sources and times; proximity to a move does not establish cause. Profit/loss tools calculate hypothetical outcomes from your inputs. Enter actual fees, leverage, funding and slippage; approximate liquidation does not reproduce every venue.
Historical testing and movement ranges
Research requests up to 1,800 completed base-volume candles from the same exchange. The frozen technical rules use a 300-bar warm-up and non-overlapping 24-bar observation windows. Hypothetical entry is the next open with adverse slippage; invalid entries already beyond target/stop are skipped. Stop gaps fill at the worse open. Same-bar target/stop touches stay ambiguous; expiry exits use the final close. Fixed illustrative costs are 0.055% per-side fees and 0.05% per-side adverse slippage. Funding, borrowing, leverage, liquidation, current books and current news are excluded. The report shows skipped, ambiguous, invalid and no-setup counts separately.
Net expectancy is the mean net R across scored historical trades, including expiry exits. R is gross entry-to-stop risk; drawdown sums fixed 1R allocations, not account-equity percent. Excluding ambiguous or missing observations can bias results. A descriptive historical test is not proof of an edge or a prediction of future returns.
Historical ranges map log-return 10th and 90th percentiles onto the reference price for 1, 4 and 24 candles. Non-overlapping observations are split chronologically: first 70% fits the bounds, last 30% measures held-out coverage. At least 30 fit and 20 held-out samples are needed. Displayed coverage refers to that historical sample, not a calibrated future success probability.
Liquidity, timeframe and timing calculations
Order-size estimates walk sampled asks for buys and bids for sells; weighted average fill is total quote value divided by covered quantity. Uncovered quantities receive no invented price. These are spot snapshots, not executable futures liquidity. Multi-timeframe comparison uses one same-exchange supported higher timeframe, with no bar closing after the original cutoff. Hourly timing requests 30 days of quote volume and high–low ranges, grouped using each historical timestamp in the chosen timezone. The latest completed hour is excluded from its same-weekday/hour median baseline, which needs three prior samples.
Cost-adjusted reward/risk divides positive net target gain by positive modeled stop loss. The break-even target-hit rate is loss divided by gain plus loss, only for a two-outcome model. It is not a winning probability. Fees, funding, slippage, expiry exits and liquidation can materially change actual outcomes.
Available now and future features
Current tools include market screening, activity heatmaps, technical scenarios, available headline context, sampled order-book depth, historical rules testing, historical movement ranges, timeframe comparisons, activity timing, cost calculators and demo practice. Paid X/social and whale feeds, trained forecasting and background alerts remain marked “Coming soon” and are not included as available paid features.