Regime-aware signals, ATR-based risk management, and strategies that ship only if they survive honest costs.
GuluStock does one thing: Bursa Malaysia equities, with the warehouse, regime model and cost assumptions built specifically for how KLSE actually trades.
No marginal ships, no curve-fit fantasies.
A 4-axis market state (trend, strength, volatility, breadth) read from the FBM KLCI gates every entry to the right environment.
Strategies are validated out-of-sample and across 15 combinatorial purged splits. They ship only if they are profitable across the distribution, not one lucky window.
Every trade gets an ATR-based stop and target, chandelier trailing, and position sizing that risks a fixed fraction of equity. Capital preservation first.
Backtests subtract real frictions, Moomoo commissions plus spread and slippage, so the edge you see survives the real world.
Top-conviction signals open paper trades that a server cron watches against live quotes every few seconds during Bursa hours.
The most recent months are a locked validation slice strategies are never tuned against, a guard against multiple-testing self-deception.
One purpose-built workspace, from the morning scan to the closed trade.
Scan the liquid universe, read the FBM KLCI regime, reconcile a watchlist, and open risk-managed paper trades, all in one screen.
A signal is only half the job. Each paper trade carries its stop, target and trail, and a server cron marks it against live quotes until it closes.
No fluff dashboards. Each view answers one question a trader actually asks: what to do, why, and whether the edge is real.
Open the desk and the whole KLSE is already sorted, charted and tape-read for you.
Candlesticks layered with EMAs, an anchored VWAP, a side volume profile and a KDJ pane: the full read on any counter, one click from the table.
The whole universe split into Basing, Advancing, Topping and Declining, so you fish where the fish are.
A five-level depth ladder and live last price, streamed straight onto the cell you hover.
From a flagged idea to a monitored position, with the maths to back every step.
A shelf of shipped strategies, each carrying its CPCV ship-or-kill verdict and out-of-sample stats. No marginal maybes.
An equity curve beside a full exit-reason breakdown (target, stop, time-stop) so the edge is legible, not lucky.
Adds, drops and revivals are tracked over time, so nothing you flagged quietly slips off the radar.
Open and closed paper trades with entries, exits and net P&L after real Moomoo costs, watched every few seconds by a server cron.
Most "AI signals" are one lucky backtest in a costume. GuluStock makes every candidate survive a gauntlet before it ever fires a paper trade.
Trained on the past, judged only on unseen future bars: a 60/40 in-sample and out-of-sample split across years of history.
Combinatorial purged cross-validation re-runs the strategy across 15 split arrangements. It ships only if it is profitable across the distribution, with mean PF at least 1.40 and worst-decile at least 1.10.
The most recent six months are sealed off and never tuned against, a guard against the multiple-testing self-deception that kills retail backtests.
Moomoo commissions, spread and slippage, subtracted from every trade, so the edge you see survives the real world.
The resting verdict is deliberate: SHIP or KILL. Anything in between trades at reduced size or keeps iterating, and every grade is re-earned whenever the engine improves.
No black box. These are the exact inputs, assumptions and cadences behind every signal.
KLSE equities, roughly 90 liquid counters after the turnover and quality filters.
klsescreener intraday bars and Bursa indices, refreshed by scheduled server jobs.
The FBM KLCI, read as a 4-axis state: trend, strength, volatility and breadth.
Moomoo commissions plus spread and slippage in ringgit, charged on both sides of every trade.
A fixed fraction of equity per trade, halved for strategies that grade REVIEW rather than SHIP.
Open positions checked every 5 seconds during Bursa hours, against live quotes.
The operational layer that turns a backtest into something a desk, or a team, can actually run day after day.
One login for the whole desk, with your watchlist, trades and analysis history carried across every session.
Admin-managed accounts with role-based access and per-client page visibility, so a client sees only the Analysis and Trades they need.
New accounts are reviewed before activation, and an admin can switch public registration on or off from Settings.
Submit a scan and a live status pill tracks it from queued to running to done with an elapsed timer. Heavy work runs server-side, so your tab stays free.
Clickable API-health pills, per-source freshness logs and cron-failure alerts keep the warehouse honest, so stale or broken data surfaces immediately.
Click-to-sort columns across every table, a full light/dark theme that follows your system, and a layout that works from a trading monitor to a phone.
Continuously refreshed OHLCV from klsescreener, into a clean local store.
Design a strategy, walk it forward over years of history, and grade it under CPCV and a locked holdout. Ship or kill, no maybes.
Each day (or tick) the engine reads the regime, classifies the universe, and fires only the validated, regime-matched signals.
Top signals open paper trades with ATR stops/targets that a server cron watches against live quotes until they exit.
What the platform is, what it isn't, and the jargon decoded.
