Quant Lab — CryptoTracker

QUANT SIGNAL CHART BTCUSDT

signal = ML alpha (boosted trees) + momentum + mean-reversion — no classic TA

QUANT SIGNAL

select a symbol…

QUANT BOT

MODE

MIN CONFIDENCE %

MARGIN / TRADE $

LEVERAGE ×

MAX POSITIONS

INTERVAL MIN

exits reuse the shared engine: SL/TP + auto break-even protect quant trades too

SMA 5/20 CROSSOVER MONITOR — winner of the 3-mo XAUUSD study · signals only, nothing auto-executes

loading crossover state…

RECENT FLIPS leg P&L, newest first

–

QUANT POSITIONS (source = quant)

no quant positions yet

ENTRY ACCURACY LAB 3-mo triple-barrier framework · % of entries that hit TP before SL · train→test validated · feeds the autotrade gate

POOLED VALIDATION 100+ out-of-sample entries across symbols vs a random-entry control — the only sample big enough to trust

PROTECTIONS

loading guards…

ALL CONFIGS — sorted by OUT-OF-SAMPLE expectancy (the only column that matters)

SMART TRADE ENGINE census-pruned triggers + order-flow confirmation (delta/divergence/value area) + session filter + per-instrument news policy + zero-float levels

RISK / TRADE % of equity

WIN RATE vs PROFIT — the trade-off, measured

PER INSTRUMENT (geometry picked on TRAIN only)

ACTIVE FILTERS

TRADE SIMULATION + FUNDAMENTAL CONTEXT every signal from every config, labelled with its triple-barrier outcome AND whether it fired into a news shock

ENTRIES / INSTRUMENT

NEWS vs QUIET — did entries during information shocks do better?

BY SCHEDULED MACRO WINDOW (UTC)

LIVE CALENDAR

loading…

PER INSTRUMENT

BY TRIGGER — which entry logic actually pays

REAL-TIME OPTION PRICING ENGINE Black-Scholes greeks + CRR American binomial + IV solver

GREEKS ACROSS STRIKES

AMERICAN vs EUROPEAN + STABILITY

FACTOR EXPOSURES drag, then backtest

weekly-rebalanced dollar-neutral long/short across the 100+ perp universe

BLEND EQUITY CURVE

TOP LONGS

TOP SHORTS

SINGLE-FACTOR PERFORMANCE (each factor alone, same engine)

HIGH-FREQUENCY MARKET SIMULATOR 1ms ticks · L2 book dynamics · latency-constrained market making

LATENCY 50ms QUOTE SPREAD 6bps SPEED

STRATEGY TELEMETRY

TRADE TAPE

the market maker quotes both sides at your spread but sees the book LATENCY ms late and its quotes take LATENCY ms to land — raise latency and watch adverse selection eat the spread. Everything here is synthetic and client-side.

MACHINE-LEARNING ALPHA MODEL gradient-boosted trees + ridge · walk-forward out-of-sample

OOS EQUITY: STRATEGY vs BUY&HOLD

FEATURE IMPORTANCE (boosting gain)

CROSS-EXCHANGE ARBITRAGE SCANNER 6 venues · executable bid/ask · fees + latency + slippage simulated

FEE bps/side SLIPPAGE bps LATENCY ms SIZE $ auto 10s

OPPORTUNITIES (net of costs)

RAW QUOTE GRID (bid/ask)

CONSTRAINED PORTFOLIO OPTIMIZER max Sharpe · 100+ assets · projected gradient

MAX WEIGHT % SECTOR CAP % TURNOVER PENALTY

WEIGHTS (top holdings)

EFFICIENT FRONTIER

SECTOR EXPOSURE

PORTFOLIO RISK ENGINE 99% VaR · expected shortfall · historical crash stress tests

NOTIONAL $

DAILY RETURN DISTRIBUTION (250d) + VaR

COMPONENT VaR (top contributors)

STRESS TESTS — HISTORICAL CRASH SCENARIOS

STATISTICAL ARBITRAGE BACKTESTER 1,700+ pair strategies · cointegration screen · noise ceiling

SHARPE DISTRIBUTION ACROSS ALL PAIRS red line = expected max Sharpe from pure noise

PAIR SPREAD Z-SCORE

TOP BY SHARPE (all pairs — mostly noise!)

TOP COINTEGRATED (ADF t < −3 — the real candidates)