Approach
Principles, not personalities.
Evrydiki is a systematic trading firm. We design rule-based strategies for liquid futures, size risk first, and execute through automation. Our work is shaped by two intellectual traditions — one about markets, one about process.
01 — What we believe
Markets are wild.
Returns are not normally distributed. Severe moves arrive more often than theory predicts and erase years of careful work in minutes. We design for that reality from the first line of code, not as an afterthought.
Process beats reaction.
Discretionary decisions break down at exactly the wrong moments. Explicit rules — documented, testable, executed automatically — produce consistency that human judgment cannot match under stress.
Risk is the design starting point.
Position sizing, drawdown limits, and exposure caps are not constraints applied after a strategy is built. They are the specification the strategy is built to satisfy.
02 — What we operate
Quantitative models for liquid futures. Daily algorithmic signals. Execution through broker API. Continuous monitoring, governance, and research feedback.
Markets
A curated universe of liquid, exchange-traded futures — US micro equity index futures (MES, MNQ, MYM, M2K), metals (gold, silver, palladium), and selected energy, FX, and rate contracts where liquidity and cost structure are suitable. Chosen for twenty-four-hour availability, deep order books, and transparent exchange settlement.
Execution
All trades route through the Interactive Brokers API. Order placement, position management, and reconciliation are fully automated. Every action is logged for audit. Human oversight supervises the system; it does not override it.
03 — Risk framework
Risk control is not a layer on top of strategy. It is the strategy. Five components govern every position the system takes.
i
Fractional Kelly position sizing
Position size is derived from estimated edge, realised volatility, capital base, and drawdown tolerance. Kelly is fractionalised — never full — to survive estimation error.
ii
Drawdown limits
Hard limits at portfolio, instrument, and daily levels. When thresholds are reached, the system reduces exposure or enters pause mode automatically. No manual override.
iii
Volatility-adjusted exposure
Position sizes scale inversely with realised volatility. Higher volatility produces smaller positions; calmer regimes produce normal ranges. Risk stability comes before P&L stability.
iv
Concentration caps
Per-instrument and per-asset-group ceilings on contract count and notional exposure. No single instrument or correlated cluster can dominate the book.
v
Stress testing & out-of-sample validation
Long-horizon backtests, out-of-sample windows, fat-tail scenario analysis, and parameter robustness checks. Only models stable across regimes are promoted to live trading.
04 — Data & infrastructure
A research backbone built before a single live trade.
Models are only as honest as the data they were tested on. We maintain an integrated multi-source research database covering price history, factor data, macro series, derivatives, and positioning data — with database-level integrity testing applied across every load.
Sources
- Exchange and vendor price feeds (Yahoo Finance, vendor APIs)
- Ken French Data Library (factor research)
- OECD macro and structural indicators
- CFTC Commitments of Traders (positioning)
- BIS property, credit, and debt-service ratio series
- ECB CISS, MIR, and BSI macro and stress indicators
- Options-chain and crypto reference data
Engine
- Custom C++ research and execution engine
- Broker bridge tested and live on Interactive Brokers
- Mandelbrot / L-stable distribution fitting at unlimited precision (MPFR)
- Daily reporting in HTML, PDF, and DOCX formats
- Versioned model registry with model-level attribution
05 — Operational scale
Capability described, not performance promised.
23
Algorithmic models live
387
Instruments tracked
2.4M
Historical bars
7
Independent data sources
Live performance reporting is available to approved investors via the investor portal.
06 — What we don't do
We do not predict markets. We define rules and let them act on whatever conditions arrive.
We do not override the system in real time. If a model is wrong, we change the model — through research — not the execution.
We do not sell signals to retail audiences. Our work is communicated through research and engaged through selection.
07 — Two influences
The Misbehavior of Markets
Benoît Mandelbrot
We accept that markets are fractal, irregular, and produce fat-tailed events more often than Gaussian models suggest. This shapes how we evaluate volatility, design stress tests, and size exposure for extreme scenarios. We embrace these features rather than ignore them.
Principles
Ray Dalio
We believe in clarity of rules and transparency in how decisions are made. For Evrydiki this means explicit trading rules, measurable risk parameters, documented processes, and continuous feedback. Process is what allows the system to keep working when judgment would fail.