Skip to content
Oxarus
QUANT · ENGINEERING

from
data
to
live.

Building systems that trade our own capital taught us what belongs in the infrastructure: pipelines that verify data, tests closed to look-ahead bias and risk controls that halt a live system when they have to.

Institutional projectRisk controls
  • Tests guarded against look-ahead bias
  • Verified against two independent data sources
  • Automatic circuit breakers in production
Representative flow diagram.
DataTestLive
TESTING

where
backtests
go
wrong.

Most systems that look good on historical data are feeding on a flaw in the test itself. Pick one to see the flawed test, the corrected pipeline and the automated checks side by side.

Audit suite5 / 5 passed

If any audit is red, the change does not merge into the main branch.

oxarus-research / audit/test_lookahead.pyRepresentative output

FLAWED TEST

fail

The day's close leaks into an intraday decision.

CORRECTED PIPELINE

pass

The read window ends before the current bar; anything after it is out of reach.

SYMPTOM

A value that's only known at the end of the day gets used during the day.

HOW WE CATCH IT

Every calculation draws only on the previous bar and earlier. We don't leave this to code review: by design, the data access layer can't see the future.

AUTOMATED CHECKS

pytest -k lookahead
  • feature_access: reads bars[..t-1] onlyrunning
  • forward_refs: none foundqueued
  • replay_parity: live stream == backtest streamqueued
DATA

two
sources,
minute
by
minute.

We compare the broker feed against an independent reference source, minute by minute. 1,721 shared trading days, 2.4 million shared minutes.

feed_parity · broker ⇄ referencevalidated

Same minutes, two sources

BrokerReferenceEURUSD · M1Representative path
14:01A 1.08484B 1.08508Δ +2.40 pip

Distribution of minute-close differences

n = 2.4M · square-root scale
  • Median0.00pip
  • 90th percentile0.00pip
  • 99th percentile0.00pip
  • Largest difference0.0pip

The shape of the distribution is representative; the marked percentiles and the largest difference are measured.

  • Shared trading daysmeasured0
  • Shared minutesmeasured0.0million
  • Daily range difference, medianmeasured0.0pip

Most of the time, the difference is a fifth of a pip. That's why we can verify our results without depending on any single broker's data.

RISK

five
checks
before
every
order.

Every order the live system sends passes these checks. Trip one, then send a test order to see which check stops it.

  • Daily loss limitArmed· daily_loss 19 / 100
  • Spread guardArmed· spread 1.1 / 4
  • News windowArmed· news_window 94 / 10
  • Connection lossArmed· heartbeat 0.7 / 20
  • Exposure capArmed· exposure_usd 45 / 100

Test order

Send an order through the gate and see which check passes and which one stops it.

$ order EURUSD buy 0.20

  1. loss
  2. spread
  3. news
  4. link
  5. exposure
risk-gateway · live-0114:02:07 UTCRepresentative screen

SYSTEM STATUS

Order flow openaccepted 1,284blocked 0
spreadok

1.1x baseline

heartbeatok

0.7s since reply

exposure_usdok

45% of cap

daily_lossok

19% of limit

news_windowok

94min to event

latency_p50ok

36ms

order_flow · msgs / 800 msflowing
events.logtail -f
  • 14:02:03INFOorder EURUSD buy 0.20 accepted · pre-trade checks 5/5
  • 14:02:04INFOheartbeat ok · rtt 38 ms
  • 14:02:05INFOrisk snapshot published · 3 accounts
  • 14:02:06INFOorder GBPUSD sell 0.10 accepted · pre-trade checks 5/5
INSTITUTIONAL

we
build
it
for
you,
too.

We build the infrastructure we run our own research and live systems on around your organization's data and processes.

  • SVC-01Data pipelines

    Infrastructure that collects, cleans and stores price and news data from multiple sources.

    DELIVERABLES

    • Source connectors and scheduled pulls
    • Dual-source comparison and quality report
    • Versioned, queryable data store

    Tools used

    • Python
    • Polars
    • Parquet
    • PostgreSQL
    Representative
  • SVC-02Testing infrastructure

    An environment where your team can test strategies free of look-ahead bias.

    DELIVERABLES

    • Look-ahead-safe data access layer
    • Walk-forward and parameter robustness reports
    • Reproducible, containerized test environment

    Tools used

    • Python
    • NumPy
    • pytest
    • Docker
    Representative
  • SVC-03MetaTrader 5 tooling

    Robots, panels, bridges and reporting tools built for institutional use.

    DELIVERABLES

    • Robot, panel and indicator development
    • Bridges between MetaTrader 5 and external systems
    • Source code and setup documentation

    Tools used

    • MQL5
    • C++
    • MetaTrader 5
    • REST
    Representative
  • SVC-04Risk and monitoring dashboards

    Dashboards that show account exposure, limits and system health on a single screen.

    DELIVERABLES

    • Per-account exposure and limit view
    • Circuit breaker states and event log
    • Instant alerts on threshold breaches

    Tools used

    • TypeScript
    • Next.js
    • WebSocket
    • PostgreSQL
    Representative

ENGINEERING

let's
build
your
infrastructure.

Tell us what you need, whether it's a data pipeline, a test environment or a risk dashboard, and we'll define the scope and approach together.

info@oxarus.com