Skip to panels
DashboardF1
MarketsF2
EquityF3
WatchlistF4
PortfolioF5
NewsF6
EconomicsF7
CryptoF8
ScreenerF9
AI ChatF10
Quant
Backtest
Education
Profile
APEXTERMINAL
API · Loading…Delayed data--:--:--WIB

Quant

Free — not metered

Statistics over a return series: the shape of the distribution, the risk ratios, how they move through time, the drawdowns and what a basket looks like together. Every metric carries its definition — a statistic is not a decision, and a ratio computed on forty observations is noise with a decimal point.

Must be a symbol we hold bars for.

Daily rate as a fraction.

Return distribution

Moments, extremes and whether the series is anywhere near normal.

IDXUSSEACryptoFree
Source: quant

Jarque-Bera —, p = —. Below 0.05 the normal assumption is rejected — which matters because a normal-distribution VaR under-states exactly the days you care about.

Quantiles

Daily return at each percentile of the period.

IDXUSSEACryptoFree
Source: quant

Daily returns at each percentile of the period. The 5th percentile is the historical VaR 95% with its sign flipped.

What this screen deliberately is not

Eleven of their twenty-three modules were ruled out with a reason. These are the ones a reader might come looking for.

IDXUSSEACrypto
  • Option GreeksThere is no liquid retail equity options market on IDX. A Greeks calculator for an instrument our users cannot trade is museum-building.Cannot support
  • ML model trainingTraining deep models per user is unbounded server compute with no credit model that makes it viable, and the output is not something a retail investor can evaluate.Cannot support
  • Reinforcement-learning agentsSame compute problem, plus an actively harmful message. An agent that “learned to trade” on a hundred episodes of one ticker is overfitting with a progress bar.Cannot support
  • AutoML, ensembles, hyperparameter tuningA direct overfitting amplifier. Tuning makes it easier to find something that worked, not likelier that it will keep working.Cannot support
  • Rolling retraining schedulesLegitimate as internal infrastructure and partly built there. As a user-operated console it is the same unbounded-compute problem.Cannot support
  • Order-book microstructureIDX does not distribute retail L2 depth on terms we can meet. The IDX-native substitute is broker summary and foreign flow, which we already hold.Cannot support
  • Unmetered local computationTheirs runs on the user’s own machine, so it is free. Ours runs on a server, so anything expensive becomes a metered job with a queue — an architectural difference, not a gap.Cannot support

Pieces that live on other screens

IDXUSSEACrypto

Portfolio-shaped analysis belongs where your actual holdings are, so risk against your real positions, the optimiser and the IDX stress scenarios are Portfolio’s. The backtest engine and its honesty layer are Backtest’s. The central-limit-theorem demonstrator, the sampling-error calculator and the data-validation scorer are teaching instruments and belong in Education. This screen keeps what is about a return series.

v—Session00:00:00MarketUS0 panelsFeeds—Latency—Ready