Skip to content

Look closer at every market move.

Analyze quote behavior, order-book observations and execution quality in q/k. Use l for the numerical and column work behind pricing research, liquidity studies and the feedback loop between a quoting decision and its outcome.

Discuss your workload Explore the runtime

01 / The workload

Understand the sequence behind the spread.

  • Quotes & trades
  • Pricing inputs
  • Execution observations

An average can hide the market you traded.

Quote updates, fills and changes in available liquidity happen together. A useful analysis preserves instrument, venue and event context while letting you examine many sessions. The challenge is moving between a detailed sequence and a comparable statistical view.

Market microstructure

Study recorded quote and order-book observations across instruments and venues. Derive spread, midpoint and depth measures from the columns supplied by your market data pipeline.

  • Inspect spread and depth distributions
  • Compare normal sessions with stressed intervals

Pricing and inventory research

Express candidate pricing inputs and inventory adjustments as vector calculations. Compare the resulting quotes with observed market conditions and your own simulation outputs.

  • Evaluate pricing variants by condition
  • Explore inventory-sensitive quoting behavior

Fill quality and markouts

Analyze prepared fill records with matched benchmark prices. Calculate execution differences and study subsequent price movement at the horizons your methodology defines.

  • Break down outcomes by venue and size
  • Compare markouts across quoting variants

Event and latency investigation

Filter recorded event timestamps and sequence identifiers around a change in behavior. Group measured timing differences to distinguish broad shifts from isolated outliers.

  • Compare event timing across sessions
  • Inspect the sequence around an unusual fill

02 / From data to analysis

Connect a quoting decision to its outcome.

Bring an already normalized session into a focused investigation, retaining the time and identifier fields needed to explain each result.

  1. 01Source data

    Bring the event context

    Use your recorded quotes, book snapshots and prepared fill-to-benchmark records, with explicit venue and sequence fields.

    • Quote observations
    • Book snapshots
    • Matched fill records
  2. l>Compute with l

    Calculate the measures

    Derive price differences and liquidity measures, select an event window, and group results by the conditions under study.

    • Spread & midpoint
    • Depth & imbalance
    • Fill-price differences
  3. 03Analytical result

    Review the distribution

    Compare quoting variants and venues, inspect tail observations and retain the rows behind the summary.

    • Venue comparisons
    • Markout distributions
    • Event-window extracts
Compare quote spreads
/ Compare quote spreads

l>bid:998 1000 1001 999

l>ask:1000 1002 1003 1002

l>spread:ask-bid

l>spread
2 2 2 3

l>avg spread
2.25

l>where spread>2
,3

l>

03 / Why l

The expression stays simple.
The layout does the work.

You write q/k. The runtime considers the physical representation of the data, choosing encoding-aware operations where supported and materializing values when needed.

Vector calculations for market observations
Express price differences, masks and reductions over columns. The runtime can select wide execution paths without introducing vector intrinsics into the research expression.
A component of your analytical stack
Use native IPC clients to connect application code and Arrow files to exchange analytical datasets. Feed normalization and benchmark matching stay explicit in your workflow.
Read the research

04 / Your first evaluation

Start with a result
you can verify.

Take one recorded session with quotes, book observations and a set of fills whose benchmark prices have already been matched.

Available as a research preview for local evaluation.

Download l Try the playground

Questions to answer with your sample

  1. Do the derived spreads and execution measures match your reference?
  2. Are timestamp units, event ordering and venue identifiers preserved?
  3. What happens to query time and memory as event density increases?

05 / Go deeper

Research and integration guides.

All documentation
Runtime research

SIMD: making every cycle count.

The data layout and execution discipline behind wide kernels.

Read the paper Integration guide

Rust SDK

Connect native services to l through the Rust IPC client.

Read the guide Integration guide

Arrow interop

Read, write and stream Arrow IPC files with the native interop library.

Read the guide
All solutions
01 / Capital markets

Hedge Funds

Signal research, backtesting, execution and portfolio analytics.

03 / Capital markets

Sell-side Banks

Desk analytics, scenario calculations and execution review.

Your data. Your next question.

Which part of your market data deserves a closer look?

Let’s talk