Initialising central foundry chamber…
Chamber 01 · Central Foundry · Facility record AR-FAC-001
Trading Systems Are Not Found. They Are Forged.
The AlphaRail Foundry Laboratory is a strategy-engineering and market-research environment designed to reveal how edge, regime, execution, risk, trade flow, capital exposure and governance operate as one system.
Interactive preview · no market data is read in this phase
Planned application scope
The
Montex
Alpha Rail System
presents
01bFacility PlateAR-FAC-001·PLATE
The bench where the system is assembled.
Every metric surfaced across this facility — expectancy, drawdown envelope, Monte Carlo survival, regime profile, parameter response — originates from one bench: the Montex Alpha Rail System strategy laboratory, of which The AlphaRail Foundry Laboratory is the application extension.
02Laboratory DoctrineAR-FAC-001·02
Most platforms test rules.
The Foundry tests systems.
A rule is a fragment. A system is what the account actually experiences: the rule, plus every decision around it that determines how much is risked, how many positions are open, what they cost, and what happens when the sequence goes badly.
03Framework DecompositionAR-FAC-001·03
Seventeen decisions,
most of them inherited by accident.
Ask a trader to describe their system and you will hear about the entry. The entry is the part they consciously chose, so it is the part they can describe.
The parts that determine the equity curve are usually elsewhere: stop geometry, partial policy, how many positions may be open at once, how correlated those positions are permitted to be, what the whole thing costs, and whether anything reduces size when the account is in trouble.
The Foundry separates all seventeen so each can be interrogated on its own.
Framework anatomy · AR-FRM specification
04Parameter Calibration BayAR-SIM-047
No parameter moves alone.
Three controls from the Parameter Laboratory. Move any of them and everything downstream responds — because in a real system it does.
Live controls
Interactive demonstration — not a historical backtest
Median of nine seeded histories at the current configuration.
Reading
—
05Branch Architecture ChamberAR-BRN-001…004
One strategy.
Four populations that behave nothing alike.
Most traders run several strategies without knowing it. The same entry, managed four different ways, produces four different distributions — and blending them without measuring them separately hides which one is failing.
Branch network
Canvas scene is decorative. Everything it depicts is stated in the text panels beside it.
Branch architecture in full →06Trade-Flow and Exposure TunnelAR-SIM-052
Sequencing is a capital decision.
The same trades, taken one at a time or four at a time, are not the same system. Concurrency changes compounding speed, drawdown depth, correlation burden and how much opportunity is rejected for lack of capacity.
Flow tunnel
—
07Risk and Drawdown ObservatoryAR-EXP-052
Drawdown is terrain,
not a number in a table.
Depth is one measurement. Duration, time under water and recovery efficiency are three more, and they are the ones that decide whether a system is actually held through the decline.
Drawdown terrain
Governance comparison
The same outcome sequence, deployed with and without a drawdown-gated risk schedule. Governance compresses the decline and gives up terminal growth in the favourable case. The trade is explicit; it should be made deliberately.
Decline from the running equity peak across the modelled history. The horizontal band marks the point at which deployment authority would be withdrawn entirely.
08Regime Classification CorridorAR-EXP-071
Hold the strategy constant.
Change the market.
One unchanged framework, evaluated across eight classified market states. Expectancy ranges from clearly positive to clearly negative. Regime permission is not an optimisation — it is a precondition.
Regime corridor
09Execution Efficiency StationAR-EXP-066
A profitable strategy
can still leak expectancy.
Between the decision and the fill there is a gap, and between the best available outcome and the realised one there is another. Both are measurable.
In the modelled configurations, giveback is larger than spread, slippage and commission combined — which means most cost-reduction effort is aimed at the smallest term.
Each point is one simulated trade. Horizontal axis is maximum adverse excursion in R; vertical is maximum favourable excursion. The quadrant boundaries mark the entry-quality and opportunity thresholds used throughout the Foundry.
Decomposition from theoretical result to realised outcome across friction and management terms.
10Monte Carlo Survival CoreAR-MC-021
One equity curve is
a sample of one.
The same trades in a different order produce a materially different experience. Trade ordering alone moves modelled maximum drawdown by more than a factor of two.
Demonstration simulation based on selected assumptions. Not a forecast.
Reading the field
The median path is not a forecast. It is the middle of a distribution that also contains the paths in which the same system, with the same edge, does not survive long enough to express it.
Raising risk per trade raises completion probability up to a threshold and then reduces it, while the probability of hitting the lock boundary rises the entire way.
Open the Monte Carlo Core →11MARS Governance ChamberAR-VAL-031
Capital authority sits
above signal quality.
Montex combines Monte Carlo — the ruler — with expectancy — the grader. Monte Carlo tells you what should happen. Expectancy tells you what is happening. MARS decides what you are permitted to do about the difference.
The layers below are not a dashboard. They are a hierarchy: a good setup cannot override a drawdown gate, and a favourable daily reading cannot expand past an authorised cap.
Optional layer
MARS is presented as an optional governance layer over any framework, not as a requirement. Its cost is explicit: it compresses risk in adverse states and gives up growth in favourable ones.
12Wizard Framework ArchiveAR-PUB-142
Decision frameworks,
not personalities.
A planned library of 80–100 educational abstractions of publicly discussed trading doctrines. Each is an interpretation of what was said in public, structured so it can be compared with the others — not a reconstruction of a private method, which would be dishonest to claim.
13Data Intake LaboratoryAR-SIM-063
A trade history is not
a dataset until it is validated.
Broker exports disagree with journals, journals disagree with themselves, fees arrive in three different columns and timestamps arrive in two timezones. None of that is unusual, and all of it changes the conclusion.
The intake workflow is shown here as a labelled preview. Nothing is uploaded, processed or transmitted in this phase — a demonstration file and a static mapping flow stand in for the real pipeline.
Intake workflow
14Validation BoardAR-VAL-031
Eight conditions
before a result counts.
A result that has not been through all eight is recorded as Simulated, never as Validated. Failed results are kept, because a negative result that is discarded will be rediscovered later at greater expense.
15Future Application BlueprintAR-FAC-002
Build. Test. Diagnose.
Govern. Refine.
The planned application as a facility map. Every module carries an honest functionality level: built in this phase, represented as a preview, or described as future production capability.
16Founding Researcher AccessAR-REG-001
Enter the development register.
The Foundry is being built in the open. Founding researchers see the modules as they are finished and shape what gets built next.
What is not promised
No launch date, no guaranteed invitation, no guaranteed founder discount and no performance claim of any kind. Registration records interest and nothing more.