01
Evidence over instinct
Every strategy earns its place through out-of-sample evidence and survives a deliberately hostile validation process. Ideas that don't clear the bar are discarded, honestly and on the record.
In supervised live validation
Meridian Quant.
Systematic. Evidence-led. Built to endure.
A systematic, multi-asset trading research practice — strategies engineered from independent research and validated with institution-grade rigour, then proven in supervised live operation before any capital is committed.
0.00
Out-of-sample risk-adjusted return (Sharpe), flagship strategy
Simulated · out-of-sample
0 / 36
Combinatorial validation paths with positive performance
CPCV
0%
Estimated probability of backtest overfitting
Lower is better
Zero
Real or client capital deployed at any stage of validation
Capital-protected
01 Approach
A research practice
engineered for discipline.
We treat finding durable edge as a problem of evidence and statistics — not conviction or narrative. The result is a process designed to disappoint us early, so it doesn't disappoint when it matters.
02
Diversification, by design
Returns are built from decorrelated sources across currencies, interest rates and digital assets. Independent bets — not leverage — are the engine of risk-adjusted performance.
03
Risk before return
Capital preservation is the first constraint. Drawdown controls, position limits and a staged deployment ladder govern everything downstream of research.
04
Technology, applied with restraint
Machine learning and statistical modelling are used where they measurably add edge — and rejected where they merely add complexity. The method serves the result, never the reverse.
02 Validation
What separates a backtest
from a track record.
Most strategies fail forward for one reason: they were overfit to the past. Our defining discipline is a validation framework built to detect that failure before any capital is exposed to it.
A
Purged walk-forward
Chronological, leakage-controlled testing. A strategy is only ever measured on data that came strictly after it was built.
B
Combinatorial cross-validation
A distribution, not a single history. CPCV rebuilds the strategy across 36 independent paths to ask whether the edge is robust or merely lucky.
C
Overfit gating
We quantify the chance we're fooling ourselves. The probability of backtest overfitting (PBO) is a gate every component must clear.
D
Deflated performance
Discounted for the search. Every result is haircut for the number of ideas tested, so we never mistake the best of many trials for genuine signal.
E
Institution-grade cost realism
Frictions from real markets. Spread, commission and financing are calibrated from tens of millions of live market ticks, then stress-tested for adverse conditions.
F
Research–live parity
We test exactly what we run. Every live risk control exists identically in the backtest — no flattering gap between research and reality.
CPCV — 36 / 36 paths net-positive
Illustrative · simulated
03 Process
From research to
controlled deployment.
One pipeline, deliberately gated at every stage. Most ideas are eliminated before they reach a market — by design.
01
Research
Independent, first-principles signal research across asset classes.
02
Construction
Decorrelated sources combined into a single, risk-weighted strategy.
03
Validation
The full gauntlet: walk-forward, combinatorial CV, overfit and cost stress.
04
Simulation
A faithful operational simulation with live risk controls — before market contact.
05
Deployment
Supervised, staged release governed by drawdown breakers and a kill-switch.
04 Flagship strategy
Four decorrelated sources.
One disciplined strategy.
Our flagship multi-currency strategy spans 28 FX pairs and combines four economically distinct return sources. Each reads the market in a genuinely different way, so the whole is steadier than any single component.
Mean reversion
Relative value
Carry
COT positioning
01
Mean reversion
Short-horizon price reversion — the all-weather core of the strategy.
02
Relative value
Reversion toward long-run fair value — a slow, structural counterweight.
03
Carry
Interest-rate differentials across currencies — the macro-rate dimension.
04
Positioning
Speculative crowding read from regulated futures (COT) data — a source independent of price itself.
Out-of-sample simulation · net of costs
Risk-adjusted return (Sharpe)0.76
Validation paths positive36 / 36
Overfit probability (PBO)25%
Under adverse-cost stress (p90)0.72
Indicative return, validation scale~8% / yr
~0.75 Sharpe is the realistic ceiling of free FX data; we present ~0.76 as a defensible figure, not a headline. The v2 construction's win is robustness — it roughly halved the overfit probability (PBO 46% → 25%) and added a price-independent return source, not a higher headline number. Costs are calibrated from 49,086,313 real market ticks. Indicative return is simulated and scales with capital — it is not income.
05 Current status
In supervised
live validation.
The research has graduated to operational testing against live markets — building an execution track record with no capital at risk.
Live · operational
Flagship multi-currency strategy
Runs 24/5 in supervised live validation on a brokerage demo account, placing real demo orders to build an operational track record. No client capital is involved.
Track record accruing
Validated · inactive
Digital-asset carry
A delta-neutral digital-asset carry strategy, validated on exchange testnet. It is gated on funding ≥ 6%/yr — currently below the floor, so it sits in cash.
Inactive — below funding floor
Armed
Risk controls
Kill-switch and drawdown breakers are armed across all execution, on supervised, self-recovering infrastructure. No position opens while a safety flag is set.
Controls armed
No client capital is deployed at any point.
Operational results validate execution quality and reliability — fills, latency, uptime, reconciliation — not profitability, and are never presented as realised returns.
06 Risk & governance
Capital preservation
is the first constraint.
—
Layered controls
Position limits, drawdown breakers that pause new risk, and a hard kill-switch that latches the strategy off. Safety flags are honoured before any order is considered.
—
Staged deployment
Research, validation, simulation, supervised live validation, and — only on evidence and approval — a measured live pilot. No stage is skipped.
—
Reproducible & auditable
Versioned data, model registry with controlled promotion and rollback, and a clean trail from raw data to every decision. Process integrity is not optional.
07 Trajectory
A deliberate path
to live capital.
Build the research engine. Discover with rigour. Validate ruthlessly. Prove execution safely. Deploy capital only once it is earned.
Foundation
Research infrastructure
Multi-asset data, a leakage-controlled research environment, and the validation framework.
Discovery
Strategy research
First-principles signal research and candidate selection under strict controls.
Validation
Acceptance testing
Out-of-sample, combinatorial and cost-stress validation with overfit gating.
Simulation
Operational simulation
Faithful execution simulation with the full suite of live risk controls.
Live validation
Supervised execution
Continuous, supervised execution against live markets, building an operational track record — no capital at risk.
Capital
Measured live pilot
A deliberate, small-scale live deployment — on evidence and explicit approval only — with full monitoring and rollback.
08 Frequently asked
Straight answers.
What does Meridian Quant do?
We are a systematic trading research practice. We engineer multi-asset strategies from independent research and validate them with institution-grade rigour, then prove their execution in supervised live operation before any capital is committed.
Are the figures shown realised returns?
No. Every performance figure on this page is from out-of-sample simulation, or from supervised live operation with no capital at risk. Nothing here represents realised profit. Simulated results carry inherent limitations — they are prepared with hindsight and cannot fully reflect real-world execution.
Is this a fund? Can I invest today?
No. Meridian Quant is a research & technology practice — not a registered fund, adviser or broker — and nothing here is an offer or solicitation to buy or sell any security or interest. We work with a small number of aligned partners interested in the research and technology.
Why present a modest Sharpe rather than a headline number?
Because an honest, overfit-resistant ~0.76 — net of realistic costs, with positive performance across every validation path — is worth more than an unaudited backtest showing an implausible figure. The discipline is the point, and the return scales with capital.
How do you guard against overfitting?
Combinatorial cross-validation across many independent paths, an explicit overfit-probability gate, performance deflated for the number of trials, realistic cost stress, and strict research–live parity. A strategy must survive all of it to advance.
What are you looking for in a partner?
We work with a small number of aligned partners — investors, allocators and collaborators — who value rigorous process and transparent risk over headline numbers. If that resonates, we'd welcome a conversation.
09 Get in touch
Let's talk.
Built on rigour.
We work with partners who value process and discipline over hype. Tell us what you're building, and we'll tell you whether we're a fit.
Get in touch →Or email us directly — meridianquant.capital@gmail.com
Important disclaimer
Meridian Quant is a research and technology project. The information on this site is provided for informational purposes only and is not investment advice, nor a recommendation, offer or solicitation to buy or sell any security, financial instrument or interest, or to adopt any trading strategy. Meridian Quant is not a registered investment adviser, broker-dealer or fund, and makes no claim of any regulatory registration or licensing.
All performance figures shown are hypothetical, simulated results, or results from supervised operation on a brokerage demo account or exchange testnet with no real capital at risk. Hypothetical and simulated results have inherent limitations: they are prepared with the benefit of hindsight, do not involve financial risk, and cannot fully account for real-world trading frictions, liquidity, or the emotional factors of actual trading. No representation is made that any account will or is likely to achieve profits or losses similar to those shown.
Past or simulated performance is not indicative of future results. Trading foreign exchange, derivatives and digital assets involves substantial risk of loss and is not suitable for everyone. Conduct your own due diligence and consult a qualified, licensed professional before making any financial decision.