Systematic trading · Practice overview · 2026Singapore · Daejeon · Bangkok
Apex Quant Capital · Multi-asset systematic trading

Systematic trading across stocks, indices, commodities and crypto, on-chain.

We design, build and trade quantitative strategies on on-chain markets for equities, indices, commodities and crypto, and in Korean cash equities. From the data lake to live execution, work with us on agentic trading systems, quant strategies, market-making, or independent advisory.

What we do

  • Build agentic trading systems, from data lake to live execution
  • Trade our own systematic strategies across asset classes
  • Provide liquidity and run market-neutral funding-basis books
  • Advise investment teams building a systematic capability

Where we trade

  • Stocks, indices and commodities as on-chain perpetuals
  • Crypto perpetuals across the leading on-chain venues
  • Korean equities on KOSPI and KOSDAQ via a regulated broker
  • Venues: Lighter, Hyperliquid, Aster, edgeX, Paradex, dYdX

Who we work with

  • Family offices seeking disciplined, systematic exposure
  • Funds and trading desks scaling research into production
  • Platforms that want an investable strategy for their users
  • Token teams and exchanges that need professional liquidity
40%
Annualised return after fees on our flagship live system
2.78
Sharpe ratio, with maximum drawdown held to 3.78%
50M+
One-minute bars in our research data lake
169+
Instruments in the lake: crypto, equities, indices, commodities
Markets we trade LighterHyperliquidAsteredgeXParadexdYdXKOSPIKOSDAQ
Quant Strategies

Two live books, traded on-chain, with every trade on the record.

Both strategies trade live on Hyperliquid, so every fill, funding payment and position can be checked on-chain. Open a live book for the full track record.

Strategy 01 · Discretionary long/short

Apex Tiger 100

An asymmetric risk:reward long/short CTA on leveraged perpetuals across crypto, equities and commodities. Losing ideas are cut fast; winners are given room to run.

  • Long/short
  • Leveraged perpetuals
  • Crypto · equities · commodities
Closed to new capital. Register interest for the next opening. See the live book
Strategy 02 · Dollar-neutral long/short

Apex MS Cross-Asset Book

A USD-neutral long/short CTA on the most liquid universe of stocks, commodities and crypto. Long and short books are sized to offset, so returns come from selection, not market direction.

  • Dollar-neutral
  • Long/short
  • Stocks · commodities · crypto
Live on Hyperliquid. See the live book
01 / Practices

Four practices, one engineering standard: every strategy we trade runs on a system we could build for you.

We are traders who build and engineers who trade. The same data, research and execution stack sits under our own books and under every client system we deliver.

Exhibit 1Practice map
Practice 1

Agentic trading systems

What it is
The full pipeline, built for you: data lake, research and backtesting, AI analyst, agent orchestration, production execution, vaults
For
Funds, desks and platforms that want their own system
Flagship
An AI-run on-chain vault trading live capital
Practice 1 →
Practice 2

Quant trading

What it is
Systematic strategies we research, run and deliver: trend following, relative strength and multi-asset momentum
For
Family offices and allocators seeking systematic exposure
Flagship
A long-only AI trader across crypto, equity and commodity perps
Practice 2 →
Practice 3

Market-making & funding basis

What it is
Automated liquidity provision, delta-neutral funding carry, and AI-driven equity screening for value mandates
For
Token teams, exchanges, family offices and investors seeking market-neutral yield
Flagship
Funding-basis on Korean equities: long cash, short on-chain perp
Practice 3 →
Practice 4

Advisory

What it is
Independent counsel on strategy validation, trading infrastructure, risk and AI for investment teams
For
Family offices, asset managers and trading desks
Flagship
Build-versus-buy and validation reviews for systematic programmes
Practice 4 →
02 / Track record

Our systems trade live capital today, with returns reported after fees.

Our flagship dollar-neutral long/short system trades the most liquid crypto, equity and commodity perpetuals on-chain, fully automated.

Exhibit 2aLive system, key statistics
MetricLive system
Annualised return, after fees40%
Maximum drawdown3.78%
Sharpe ratio2.78
Sortino ratio2.9
Net exposure (target)Dollar-neutral
UniverseCrypto, equity & commodity perps
Source: live on-chain account, since inception (August 2026) to September 2026. Past performance is not indicative of future results.
Exhibit 2bAnnualised return against worst drawdown, live system
Annualised return after fees of 40 percent against a maximum drawdown of 3.78 percent: about 10.6 times return per unit of drawdown. 0%10%20%30%40%50% Annualised return after fees: 40% Maximum drawdown: 3.78% Annualised return,after feesMaximum drawdown 40%3.78% 10.6× annualised return per unit of maximum drawdown · Sharpe 2.78 · Sortino 2.9
Swipe to see the full chart →
Source: live on-chain account since inception (August 2026). Return-to-drawdown = annualised return ÷ maximum drawdown.
03 / Practice 1
Agentic systems

Six stages sit between a trading idea and a machine that trades it. We deliver all six.

Most teams own one or two stages and stitch together the rest. We engineer the full chain as one system, so data, research and execution never drift apart. Take the whole pipeline, or only the stage you are missing.

Exhibit 3The agentic trading system pipeline
01
Data lake
02
Research & backtesting
03
AI analyst
04
Agent orchestration
05
Production execution
06
Vault & custody

Candles, funding and order-book data from every venue, cleaned, layered and refreshed nightly.

A deterministic engine and an agentic loop that sweeps strategies and scores every result.

Ask the data lake in plain English. Get the answer, the statistics and the SQL behind it.

Fleets of research, execution and monitoring agents, coordinated under human-owned rules.

Order service with sizing, hedging, stops and circuit breakers, live around the clock.

On-chain vaults with institutional custody, NAV reporting and investor-grade letters.

Advisoryspans every stage: strategy design, validation, market structure, risk and operating model (see Practice 4).
Stages 05–06 shaded: the stages where most in-house efforts stall, and where our production record is deepest.
Exhibit 4aInstruments under coverage, by market class
169 instruments: 100 on-chain equity, index and commodity markets and 69 crypto perpetuals 20015010050 Crypto perpetuals: 69 On-chain equity, index & commodity markets: 100 Total: 169 69100169 CryptoEquities · indices · commodities Crypto perpsEquities, indices,commoditiesTotal
Swipe to see the full chart →
Source: Apex data lake audit, September 2026. On-chain equity, index and commodity markets include builder-deployed (HIP-3) perpetuals.
Exhibit 4bLake architecture and scale
LayerContentsScale
BronzeRawVenue payloads exactly as received: candles, funding, market metadataAll venues
SilverCleanOne-minute and hourly bars, normalised symbols, data-quality ledger50.8M 1-min bars
714K hourly bars
GoldResearchDaily bars, funding history, approved metrics and a permanent trial registry1.24M funding rows
121K daily bars
AccessAgentsRead-only, scoped keys for researchers, clients and AI agents; queried in SQL or plain EnglishRefreshed nightly
Stored as columnar Parquet on object storage, queried in place with DuckDB. Funding history from 2023.
Exhibit 5Services matrix: each stage is a service you can buy on its own
StageWhat we deliverKey capabilitiesOutput you own
01Data lakeA research-grade market data lake for your venues and instrumentsMulti-venue ingestion · one-minute candles and funding · bronze/silver/gold layering · data-quality ledger · nightly refresh · scoped read-only accessYour data lake and its pipelines
02Research & backtestingA deterministic backtest engine and an agentic research loopStrategy library · parameter sweeps · statistical promotion bar (probabilistic Sharpe, drawdown, exposure, sample size) · permanent trial registryValidated strategies with a full evidence trail
03AI analystNatural-language access to your data lake and engineMCP-connected agents · plain-English questions answered with statistics and SQL · house rulebook the agent must follow · new trials written on requestAn analyst desk that works around the clock
04Agent orchestrationCoordinated agent fleets across research, execution and operationsMulti-agent workflows · research, validation and monitoring agents · human-owned guardrails · task routing and hand-offs · reporting agentsAn automated operating layer for your desk
05Production executionYour strategy, running live and unattendedOrder service (sizing, entries, exits, take-profit and stop-loss, hedging) · isolated-margin presets · circuit breakers · liquidation-wick protection · copilot-to-autopilot rollout · 24/7 monitoringA live trading system and its runbook
06Vault & custodyOn-chain vaults that take in capital safelyVault design and deployment · audited contracts · institutional multisig custody · permissioned execution · NAV and fund-letter reportingAn investable product with investor reporting
Exhibit 6Venue-agnostic by design: venues we build on, by pipeline stage
VenueMarket structureDataResearchExecutionRiskVaults
LighterZK-rollup order-book perpetuals
HyperliquidL1 order-book perpetuals, incl. builder-deployed equity and commodity markets
AsterMultichain order-book perpetuals
edgeXStarkEx order-book perpetuals
ParadexStarknet appchain order-book perpetuals
dYdXCosmos appchain order-book perpetuals
One data model, one research engine and one execution service behind every venue connector. Other venues on request, including centralized exchanges for hedging legs.
04 / Practice 2
Quant trading

We trade our own systematic strategies, and build them for clients who want the same discipline.

Every strategy is a written rule set, a single decision engine shared by research and live trading, and a control stack that keeps it inside its limits.

Exhibit 7aAnatomy of our long-only AI trader: one decision engine from backtest to live
Daily cycle: universe snapshot, signal, allocate, manage, exit, reconcile. The same decision step drives both the backtest and live execution. 01 Universe Audited, write-oncesnapshot of eligiblemarkets 02 Signal Trend slope onclosed daily bars 03 Allocate Capped equal-weightslots, ranked byopen interest 04 Manage Marks at the close;funding accruedevery day held 05 Exit Volatility-scaledtrailing stop 06 Reconcile Journal checkedagainst venue statebefore next run One pure decision step, called by both the backtest driver and the live runner Backtestdriver → ← Liverunner
  1. 01UniverseAudited, write-once snapshot of eligible markets
  2. 02SignalTrend slope on closed daily bars
  3. 03AllocateCapped equal-weight slots, ranked by open interest
  4. 04ManageMarks at the close; funding accrued every day held
  5. 05ExitVolatility-scaled trailing stop
  6. 06ReconcileJournal checked against venue state before next run
Steps 02–05 are one pure decision step, called by both the backtest and the live runner.
The backtest driver and the live runner call the same one-day decision step, so the strategy that trades is exactly the strategy that was tested.
Exhibit 7bEngineering standard behind every strategy we run
PrincipleHow it is enforced
One engine, research to liveA single pure decision function drives the backtest and the live runner. No second implementation to drift.
Frozen, verified dataUniverse snapshots and datasets are written once and checksum-verified on every load.
Tested invariants117 automated tests, including randomised portfolio invariants across staggered listings, funding and execution modes.
Unlevered by constructionCapped slots, and every fill clamped to settled cash, so the book can never lever up.
Funding-aware P&LFunding accrues daily on every held position and settles at exit.
Crash-safe executionJournaled orders, pre-send checks, a kill switch and journal-to-venue reconciliation on every run.
Exhibit 7cStrategy specification, summary
StyleLong-only trend following
UniverseEligible crypto, equity and commodity perpetuals, snapshotted and audited
SignalTrend-slope entry, confirmed on closed daily bars
PortfolioCapped equal-weight slots; contention ranked by open interest
ExitVolatility-scaled trailing stop
LeverageNone
ExecutionDaily, at the signal close or the next open
Parameters withheld. Full specification available under NDA.
05 / Practice 3
Market-making

We provide liquidity and run market-neutral carry: we earn spread and funding, not direction.

An automated market-making operation across centralized and decentralized venues, delta-neutral funding-basis books that pair on-chain perpetuals with the cash market, and AI-driven equity screening for value mandates.

60+
Token teams provided with liquidity
50+
Exchange relationships, incl. direct access to Korean exchanges
CEX + DEX
Order-book and on-chain venues, one automated operation
In writing
Contractual spread, depth and uptime commitments
Exhibit 8aFunding basis on Korean equities: how the trade earns
Long the stock in the Korean cash market, short the same stock's perpetual on-chain. Price moves cancel; the funding paid by leveraged longs is collected. LONG · CASH MARKET Korean equity Held through a regulated brokerKRW · unlevered · no liquidation SHORT · ON-CHAIN PERPETUAL Same stock, perpetual Builder-deployed equity marketUSDC · isolated margin Δ ≈ 0 price hedged COUNTERPARTIES Leveraged on-chain longs, tradingafter the Korea Exchange closes pay funding NET RESULT Funding collected, with thestock's price moves offset
Long · cash market
Korean equity

Held through a regulated broker. KRW, unlevered, no liquidation.

Δ ≈ 0 · price moves offset
Short · on-chain perpetual
Same stock, perpetual

Builder-deployed equity market. USDC, isolated margin.

Counterparties ↑ pay funding

Leveraged on-chain longs, trading after the Korea Exchange closes.

Net result

Funding collected, with the stock's price moves offset.

Korean retail pays to hold leveraged longs on-chain after the Korea Exchange closes. The book collects that funding with its price exposure hedged out.
Exhibit 8bControls on the basis book
ControlWhat it does
Hedge matchingMicro-rebalances keep the short's USD notional matched to the long, FX-adjusted
Margin defenceTops up isolated margin automatically to hold the liquidation price away
Live monitoringMargin ratio, liquidation distance, funding and hedge drift, alerted in real time
Liquidity gateRefuses to open on markets below a minimum daily volume
Safe by defaultDry-run unless armed; one command flattens both legs
Equity research

An AI investment committee for Korean equities, delivered monthly to a value-oriented family office

Exhibit 8cFrom the full Korean market to a 20-name shortlist
Funnel: liquid KOSPI and KOSDAQ universe, exclusions, magic-formula ranking to 400 candidates, AI committee scoring, top 20 published monthly. Universe: Liquid KOSPI and KOSDAQ stocks after a price and liquidity gate Universe Liquid KOSPI and KOSDAQ stocksafter a price and liquidity gate Screened: Excludes financials, utilities, holding companies, SPACs, REITs Screened Excludes financials, utilities,holding companies, SPACs, REITs Top 400: Ranked on earnings yield plus return on capital Top 400 Ranked on earnings yield plusreturn on capital Scored: AI committee of three value lenses agrees a 0–10 score Scored AI committee of three valuelenses agrees a 0–10 score Top 20: Published monthly to the client, with full reasoning Top 20 Published monthly to theclient, with full reasoning
Universe

Liquid KOSPI and KOSDAQ stocks after a price and liquidity gate

Screened

Excludes financials, utilities, holding companies, SPACs, REITs

Top 400

Ranked on earnings yield plus return on capital

Scored

AI committee of three value lenses agrees a 0–10 score

Top 20

Published monthly to the client, with full reasoning

Quality-and-value ranking combines earnings yield with return on capital. Committee scores run from 0 to 10; most months, most names are rejected.
  • Grounded in filingsEvery number in a committee report must appear in the company's filings, or the report is rejected.
  • Three philosophies, one scoreThree value-investing lenses judge each company, debate, and agree one conviction score.
  • Frozen, versioned committeeThe committee prompt is fingerprinted, so scores stay comparable month to month.
  • Calibrated before it runsEach model is checked against a ground-truth set before a full batch is approved.
  • Research, not tradingThe system ranks and explains. The investor decides, sizes and trades.
Built to be rebuilt: a written playbook ports the screener to any other equity market.
06 / Practice 4
Advisory

Independent advice from practitioners who run capital, for investment teams building a systematic capability.

We advise from the desk, not the whiteboard. Every recommendation is something we have built, broken and fixed on our own books.

Exhibit 9Advisory mandates
MandateWhat we advise onTypical client
AStrategy validationBacktest review, overfitting and multiple-testing checks, statistical promotion bars, live-versus-backtest driftFunds, family offices, allocators
BTrading infrastructureData architecture, research tooling, execution stack, build-versus-buy decisions and vendor selectionFunds, prop and treasury desks
CAI for investment teamsAgentic research desks, AI investment committees, and governance of model output in the investment processFamily offices, asset managers
DRisk & operationsControl stacks, staged rollouts, incident playbooks, custody and execution separationFunds and trading desks
EMarket accessOn-chain perpetual markets for traditional investors; Korean market structure and cross-market carryFamily offices, TradFi desks
07 / Engagement

Where you start depends on what you already have: strategy, infrastructure, or neither.

We meet clients in any quadrant. Funds usually bring the strategy; family offices and platforms usually bring the capital and the mandate.

Exhibit 10Engagement matrix
Two by two: strategy (need or have) against infrastructure (need or have), mapping to Full build, Productionise, Research partner and Scale and operate. NEED IT HAVE IT TRADING INFRASTRUCTURE → NEED IT HAVE IT STRATEGY → Productionise Scale & operate Full build Research partner Execution, risk, vault andreporting built aroundyour strategy. Agent orchestration, researchacceleration and an operatingmandate on your stack. Our strategies and researchloop, feeding your existingdesk and infrastructure. Strategy, data, engine andexecution: a complete system,live and investable. TYPICAL: FUNDS TYPICAL: FAMILY OFFICES TYPICAL: TRADING DESKS TYPICAL: PLATFORMS, BROKERS
Have strategy · need infrastructure
Productionise

Execution, risk, vault and reporting built around your strategy.

Have both
Scale & operate

Agent orchestration, research acceleration and an operating mandate.

Need both
Full build

Strategy, data, engine and execution: a complete, investable system.

Have infrastructure · need strategy
Research partner

Our strategies and research loop feeding your desk.

You bring

The mandate, the capital, the decisions.

We bring

Data, strategies, engine, code, custody and upkeep.

Engagement modelBest for
Full buildA complete system delivered and handed over
Build & operateWe build it and run it for you, day to day
Strategy licenceOur systematic strategies on your infrastructure
Single stagePlug the one gap in your existing stack
Advisory retainerSenior guidance through every decision
08 / Selected work

Four systems, four different problems, one team that built and runs every one of them.

Client names withheld. References available on request.

Exhibit 11a · Practice 1 · Asset-management platform

An AI-run trading vault, from empty repository to live capital

Situation
A platform wanted an on-chain vault its users could deposit into, backed by a systematic strategy rather than discretionary calls.
Approach
We designed a dollar-neutral relative-strength long/short book across the most liquid crypto, equity and commodity perps, then built the execution, risk controls, custody and investor reporting around it.
Result
40% annualised after fees, 3.78% maximum drawdown, Sharpe 2.78, live and fully automated since August 2026.
Exhibit 11b · Practice 2 · Systematic trend

A long-only AI trader, scaled from three majors to a multi-asset universe

Situation
A trend strategy proven on the three largest crypto assets needed to scale across crypto, equity and commodity perpetuals without adding leverage or operational risk.
Approach
We generalised the rule set to an audited universe, added a capped-slot portfolio layer, and built a single decision engine shared by backtest and live, behind a journaled, crash-safe runner.
Result
One codebase from research to live across the full universe, backed by 117 automated tests.
Exhibit 11c · Practice 3 · Funding basis

Funding capture across a Korean equity and its on-chain perpetual

Situation
On-chain perpetuals on a Korean semiconductor leader were paying persistent funding that the underlying cash market did not.
Approach
We paired a long cash position through a regulated Korean broker with a short on-chain perpetual, with FX-aware hedge matching, automatic margin defence and live monitoring.
Result
Live since September 2026, collecting funding with the price risk hedged out.
Exhibit 11d · Practice 3 · Family office

A monthly AI investment committee for a value-oriented family office

Situation
A family office with a value mandate wanted a disciplined monthly shortlist of Korean equities without building an analyst team.
Approach
Quality-and-value ranking across KOSPI and KOSDAQ, then an AI committee of three value-investing lenses scoring the top 400 names from their filings.
Result
A monthly top-20 report in which every figure traces back to the filings.
09 / Team

Traders and engineers, headquartered in Singapore with desks in Daejeon and Bangkok.

One team runs our own books and builds every client system, across the three time zones that span Asia's trading day.

Exhibit 12aWhere our team comes from
Banking & insurance
  • Deutsche Bank
  • J.P. Morgan
  • Allianz
  • DBS Bank
  • OCBC Bank
Trading & digital assets
  • Huobi
  • Deriv
  • MakerDAO
Technology & consulting
  • Singtel
  • Dyson
  • Thomson Reuters
  • Infosys
  • Tata Consultancy Services
Firms where members of our team have previously worked, across global banking, trading and technology. Trademarks belong to their respective owners.
Exhibit 12bOur locations
Bangkok · Thailand

Bangkok

Desk covering Southeast Asian markets and clients.

Indochina Time (UTC+7)

Singapore · Head office

Singapore

Head office: strategy, research, engineering and client coverage.

Singapore Time (UTC+8)

Daejeon · South Korea

Daejeon

Desk covering Korean markets, exchanges and clients.

Korea Standard Time (UTC+9)

10 / Risk

Production risk is engineered in layers, so no single failure reaches client capital.

Every system we trade or deliver carries the same control stack. Each layer is designed to catch what the layer above it misses.

Exhibit 13Defence-in-depth control stack
Layer 1Pre-trade
Position sizingVolatility-aware sizing within capital limits
Position capsHard per-market and gross exposure limits
Funds checksOrders blocked when margin headroom is thin
Layer 2In-trade
Isolated marginEach position ring-fenced by preset
Protection ordersReduce-only stops that defend against liquidation wicks
Hedge controlAutomated hedge ratios on paired legs
Layer 3Portfolio
Circuit breakersAutomatic de-risking on drawdown triggers
Neutrality monitorNet exposure held to target, continuously
Leverage guardAccount-level leverage ceilings enforced
Layer 4Operations
Staged rolloutCopilot mode before autopilot, on every system
24/7 monitoringAlerts and agent-driven incident triage
ReportingDaily marks, NAV and fund letters
Layer 5Custody
Institutional multisigClient capital held under multi-party control
Permissioned executionTrading authority separated from custody
Audited contractsVault contracts independently audited
Next step

Tell us the strategy you want to run.

A 30-minute call with the team that trades. We map your mandate onto our practices and show you what it takes to put it live.

Email the team →
Family offices and funds: ask about strategy licences and advisory mandates.