Loading PSX real-time flow…
Walk-forward validated · zero look-ahead leakage

We test the edge before we sell it to you.

Alpha PSX is a rule-based signal engine for the Pakistan Stock Exchange. Every parameter is optimized on a training period, then measured — once, never re-selected — on a held-out test period. What doesn't generalize gets excluded, not shipped.

Latest validation cycle · train → test sector-pooled
CEMENT
0.69× VALIDATED
BANKS
0.77× VALIDATED
OMC
−0.95× EXCLUDED

Excluded sectors aren't hidden — they fall back to conservative baseline parameters until a future cycle validates them. This is a methodology snapshot, not a return projection.

Systematic Timeframe Architecture

Different horizons need different math, and different risk. Stop and target distances are tuned per horizon — and, where the data supports it, per sector.

1–7 DAYS

Short-Term Catalyst

Momentum bursts on a 1–7 day window, read through fast oscillators and structural price-momentum divergence.

Fast RSI (7) Fast MACD Hidden Divergence Vol-Ratio (RVOL)
2–8 WEEKS

Medium-Term Trend

An ADX chop filter neutralizes signals in flat markets. Direction is confirmed by EMA slope and money flow before a trade is ever proposed.

ADX > 18 Filter EMA 20/50 Slope Chaikin Money Flow Supertrend v2
2–12 MONTHS

Macro Position

Position trades built on Ichimoku cloud structure and PnF supply/demand columns — the same tools used to read institutional accumulation.

SMA 200 Macro Ichimoku Structure PnF X/O Columns ROC (126d)

How a parameter earns its way onto the platform

No hand-picked numbers. Every stop and target multiplier is the output of a four-step gate, run per sector so the sample size is large enough to trust.

01

Train / test split

Parameters are grid-searched on one period and never touch the other until testing.

02

Pooled across a sector

Trades from every stock in a sector are combined per candidate, so thin single-stock samples can't masquerade as edge.

03

Measured once

The winner is chosen on training data alone, then evaluated on the test period exactly once — never re-selected by its test score.

04

Excluded, not hidden

A sector that doesn't generalize falls back to a conservative baseline. It stays visible below, not swept under a headline number.

Sector coverage, most recent cycle

refreshed every optimization run
CEMENT
validated
BANKS
validated
FERTILIZER
partial
POWER
partial
AUTOS
partial
E&P
baseline
TECH
baseline
OMC
baseline
PHARMA
baseline
FOOD
baseline

Validated — generalized cleanly from train to test across all three horizons. Partial — validated on at least one horizon. Baseline — no sector-specific edge found yet; uses the platform-wide default.

Under The Hood

Four mechanics that run on every signal, before you ever see it.

Reward-to-Risk Gate

Every signal's reward-to-risk ratio is computed before it's issued. Anything under 1.0 is suppressed to NEUTRAL rather than shown.

Regime-Switching Logic

A market regime classifier reads the KSE-100 as Trending Bullish, Macro Bearish, Sideways, or High Volatility, and adjusts thresholds accordingly.

Corporate-Action Aware Data

Prices are adjusted for dividends, splits, and bonus issues, with a freeze window around ex-dates so a corporate action can't be misread as a signal.

Cost-Aware Backtesting

Every simulated trade carries commission and slippage, and a stop that gaps past its price fills at the open — not at the stop.

See what's actually validated, sector by sector.

No single headline number. A live breakdown of what's been tested, what held up, and what's still on the baseline.

Alpha PSX is an independent analytics platform. Signals are generated using rules-based technical analysis for informational purposes only — not financial advice. Validation metrics describe historical backtested behavior under a specific methodology; they are not a projection or guarantee of future results. Past performance does not guarantee future results.