Appearance

Personalize your experience

Mode
Accent Color
Direction
Analysis Marketplace
QuaScript

Jump Diffusion Monitor

Admin's Choice NewFree

Realized vs bipower volatility, Lee-Mykland jumps and Hawkes jump clusters

HFT MicrostructureOwn pane WebGPU CanvasHybridDashboardSignalsAlerts
quascripthftwebgpujumpsbipowerhawkes

Favourites and Admin's Choice analyses are listed in the terminal's indicators window under Analysis Marketplace.

Language
QuaScript
Engine
Native C++, incremental on live ticks
Rendering
WebGPU + Canvas
Placement
Own pane
Settings
4
Plots
8
Lines of code
57
Source
presets/analysis/hft/jump_diffusion_monitor.qs
Jump Diffusion Monitorengine result
Loading the engine result…
quatick-charts · QuaScript engine on synthetic 1-minute bars · drag to pan · scroll to zoom · Run executes the script on the engine (⌘/Ctrl+Enter while editing)

About

Splits volatility into its continuous (diffusion) part and its jumps, the decomposition used by high-frequency risk desks. The filled area is realized volatility, the line inside it is bipower volatility (blind to jumps); the gap between them, the jump share, is shaded by size. Every return the Lee-Mykland test flags as a jump is marked, and the jumps feed two Hawkes intensities (up and down) so clusters of jumps, the signature of news shocks and liquidation cascades, show as the mirrored waves along the bottom. The baseline is the realized semivariance balance: upside variance against downside variance.

8
Plotted series
3
Signal marks
1
Drawings
2
Alert rules
4
Settings
45
Lines of code

What it draws

Own pane · counted from the engine's plot table and the script's draw calls

Series

  • Line7 visible · 1 helper
    FloorRealized volDiffusion volCluster axisUp clusterDown clusterSemivariance balance
  • Band fills (render.band calls)4
Legend
Floor Realized vol Diffusion vol Cluster axis Up cluster Down cluster

Signal marks

Up jumpDown jumpDown-jump cascade

Drawings and chart effects

  • Dashboard tables1 created

How it runs

From source to the live chart, for this script

  1. 1
    Compile

    The QuaScript compiler checks the 57-line script (45 lines of code) and lowers it for the native C++ engine. Compile errors point at the exact line and column.

  2. 2
    Place

    The header sets pane = hpane with height = 0.3 (30% of the chart), so the output gets its own pane under the price with its own scale.

  3. 3
    Compute

    The script runs once per bar, oldest to newest. 1 persist variable keeps state (objects, counters) across bars.

  4. 4
    Stay live

    On the terminal the script keeps one live engine session: a tick rolls the forming bar back to its checkpoint and re-executes only that bar, and new bars are appended, so history is never rerun while the market moves. The live result is identical to a full run.

  5. 5
    Render

    The results are handed to the chart: series on WebGPU, drawings on the Canvas 2D QuaScript layer.

  6. 6
    Tune

    4 settings in 3 groups appear in the chart's settings dialog. Changing one reruns the engine with the new value.

Rendering

Hybrid: this script draws on 2 surfaces at once

WebGPUSeries renderer

The browser's modern GPU API. The chart uploads every line, band and mark of the script as GPU buffers and draws them in one pass, so dense heatmaps and dozens of series stay smooth while you pan and zoom. Without WebGPU support the chart's GPU layer falls back to WebGL 2, then Canvas 2D.

In this script
  • • 8 plotted series
  • • 4 band fill calls
  • • 3 signal mark types

Chosen by render = webgpu in the header.

WebGLNot used

GPU drawing through WebGL 2, available in every current browser. Used for chart-native profiles and for series that ask for it with backend = "webgl". Fast for large numbers of bars and wide fills.

CanvasUsed

The browser's 2D drawing API. Text-heavy and object-style output (dashboard tables, zones, tags, segments), candle colors and backgrounds are painted by the QuaScript layer, a Canvas 2D surface over the chart. Series use Canvas when the script's header does not ask for a GPU.

In this script
  • • 1 dashboard tables (QuaScript layer)

Alerts

2 alert rules declared with alert.rule

Price jump

The Lee-Mykland test flagged this return as a jump

Jump cascade

Down jumps are clustering (Hawkes intensity above 2)

Settings

4 settings, grouped as in the chart's settings dialog

General
SettingTypeDefaultRange / optionsId
SourcePrice sourceclose—source
Decomposition
SettingTypeDefaultRange / optionsId
Window (returns)Integer5010 – 1000window_returns
Jumps
SettingTypeDefaultRange / optionsId
Jump test significanceNumber0.010.0001 – 0.2, step 0.001jump_test_significance
Cluster decayNumber0.080.005 – 1, step 0.005cluster_decay

Script

packages/quascript/presets/analysis/hft/jump_diffusion_monitor.qs

Header

analysis "Jump Diffusion Monitor" {
    pane = hpane
    height = 0.3
    render = webgpu
}
Total lines
57
Code lines
45
Comment lines
7
Helper functions
0
Loops
0
Persistent variables
1
Higher-interval feeds
0
Engine plots
8

Built with

Technical analysis (ta.*)
ta.highest
Microstructure kernels (hft.*)
hft.bipowerhft.hawkeshft.jump_testhft.semivariance
Math (math.*)
math.maxmath.min

Discussion

Sign in to join the discussion. Your posts and replies show up on your profile.