Bayfall Morrigan is a quantitative proprietary trading firm that develops structure-first systems built from structural reasoning about market microstructure. The firm was founded by Aidan Ronan and operates with a methodology grounded in structural reasoning, discipline, and rigorous forward validation.
Quantitative Proprietary Trading&Market Microstructure Research
Built on structural reasoning rather than statistical discovery.
Lender Access
Systematic Integrity in Quantitative Trading
Bayfall Morrigan designs and operates quantitative trading models engineered from market microstructure. We strictly decouple structural architecture from empirical testing. Instead of data snooping for spurious correlations, we rigorously validate behavioral hypotheses and calibrate our systems through structural logic.
Method
At Bayfall Morrigan, our models originate as structural hypotheses. We map recurring market dynamics into explicit causal chains, tracing the path from the structural condition that should create an inefficiency to the observable behavior it produces. Price impact is one of the primary observables through which we characterize these mechanisms, including the extent to which an inefficiency can actually be extracted and where its capacity becomes constrained. Our predictive signals are derived from these causal relationships. Data is therefore empirical evidence, not the authority on model structure. Statistical discoveries can expose relationships worth investigating, but they do not independently justify a change to the model. A parameter, calibration decision, or structural change must remain supported by a coherent explanation for why it should improve the model's ability to capture the inefficiency it was designed to capture. This distinction prevents statistical relationships from silently determining the architecture of the system and keeps the resulting models interpretable when market conditions change. When performance diverges from expectations, the model retains a structure that can be examined rather than a collection of fitted relationships whose origins are difficult to distinguish.
Model Integrity
Regime shifts are inevitable. What matters is whether a model remains interpretable when the environment changes. An interpretable model gives us something to pull apart: its assumptions, mechanisms, conditions, and parameters. When performance changes, we can examine the reasoning that produced the model rather than simply observing that the output changed.
Parameters should not exist simply because historical observations improved a fit. Calibration decisions should not be accepted simply because the data preferred them. Statistical discoveries can reveal relationships worth investigating, but they do not, by themselves, establish why those relationships exist. The model must remain grounded in a reason for what it contains.
Our models are built around claims about how a structural inefficiency should behave. We test those claims against predicted conditions and defined failure boundaries.
Production Architecture
We utilize historical testing strictly to assess how a completed methodology performs across prior market regimes, never to parameterize a system to a transient historical window. Final validation runs exclusively against out-of-sample, live data forward in time, forcing the system to prove its logic in real time across the ordinary variations of an active market. The result is an authentic behavioral profile under real operating conditions, entirely insulated from historical backtest optimization.
Every component of our pipeline, spanning from our cross-sectional screening frameworks to our non-parametric bootstrap simulators, runs on completely custom-built infrastructure. We operate a bespoke data architecture that logs every microstructural metric underlying a model's resolution, strictly separating setup variables from terminal outcomes to preserve absolute statistical integrity.
Prior to capital allocation, every system undergoes extensive simulation across a complete distribution of adversarial outcomes. We do not evaluate models based on average-case performance. We stress-test them under tail-risk conditions to fully characterize their behavioral boundaries and select the exact execution configuration that aligns with how the system operates in reality.
The Limits of the Approach
The mechanics can resolve exactly as the thesis describes while the realized edge approaches zero, because another participant reaches the same structural condition first. This is a discovery problem rather than a thesis failure, and it is diagnosable in live data. The entry condition confirms, but the opportunity has already resolved before the position can capture it.
A mechanism is a claim about a condition in the market that produces the edge, not about the edge itself. That condition can be removed by changes outside the model: a rule change, a shift in market structure, or a change in how the underlying condition is produced. When the condition no longer exists, the thesis no longer applies. This is a hard-reset condition, not a recalibration. External changes relevant to each mechanism are therefore monitored as structural risk.
Interpretability is only as strong as the independence of the information being interpreted. Commitments within a model, including the condition, failure boundary, and expected residue, can depend on the same upstream data, classification scheme, and reasoning. An error upstream can therefore produce apparent agreement across multiple commitments without being independently exposed by any of them. The number of commitments does not necessarily reflect the number of independent reasons supporting the model.
Every model in this methodology begins with a stated mechanism, specific enough to be wrong. This constrains the space of inefficiencies the methodology can address. High-dimensional relationships without an interpretable structural mechanism fall outside that space.
Strategic Evolution
The Laboratory
The largest quantitative funds require strategies capable of deploying institutional-scale capital. A pure structural edge in mid-liquidity equities reaches a hard capacity ceiling at a fraction of what moves the needle for a multi-billion-dollar operation. Institutional mega-funds do not bypass these microstructural mechanics because they are blind to them; they bypass these positions because their capital cannot physically fit through the door. Deploying institutional size into these specific spaces would immediately consume the very market structures the edge depends on before a meaningful position could even be built.
Mid-liquidity is where we pointed our methodology, not where structural mechanics are confined. We've always preferred short, controlled risk exposure, and that preference is what made mid-liquidity equities the fit: in highly liquid instruments, the same fast mechanisms are hyper-contested and priced out by speed, while mid-liquidity left room for correct structural reasoning to actually participate.
Short, controlled exposure was a preference, not a structural limit. We chose it because a low-capacity environment returns live-capital feedback fast, letting us validate the framework quickly, cheaply, and without competition. The capacity ceiling on our current deployment is what let the underlying science get proven in the first place. It defines where we started, not where the methodology stops.
Scaled Architecture
Structural reasoning applies wherever a market mechanism does, across all liquidity regimes. But scaling the model means giving up our preference for short, controlled exposure. Capital scale dictates execution horizon: a small position can move instantly, an institutional one can't. Deploying real size alters the very environment it enters, so the position has to be worked across a longer horizon to avoid moving the market itself.
Adapting the methodology to institutional horizons is a distinct engineering problem: an architecture that holds the thesis together over much longer timeframes, past the short-horizon wrappers that defined the laboratory phase. Because the design principle is rooted in invariant microstructural mechanics rather than patterns specific to a moment in history, the underlying logic doesn't change. What we're setting aside is a preference, not a principle.
As we scale, the structural mechanisms carry forward into size; the short-horizon exposure does not. The scaling architecture exists to preserve the same methodology inside a larger, slower execution regime. We aren't changing how we reason. We're building a different operational framework to sustain that reasoning at institutional capacity.
Systems
Each system advances independently through our validation lifecycle, from structural specification and historical regime testing to live forward validation and complete simulation characterization. Risk parameters, position sizing, and capacity limits are derived natively from each mechanism rather than arbitrarily imposed on them. Once deployed, each engine stands alone as a discrete operational system. Specific logic remains proprietary.
A high-precision momentum reentry framework. When short-term direction aligns with the prevailing intraday trend, the engine monitors the initial pullback to test that structural level. It triggers execution the exact instant the retest holds, capturing the immediate microstructural pivot where recent participants actively defend their positions. The execution window is deliberately brief and precise. This high precision inherently caps its capacity; the system targets thin, fast-moving equities where outsized volume induces adverse market impact. It is engineered to operate a contained book rather than an expansive one, constraining deployable capital, not its structural performance.
Momentum reentry
A system class focused on reentering directional movement after temporary structural retracements.
Structural continuation
Captures trend continuation immediately following a validated structural retracement.
Fully operational
Verified under live conditions and characterized via simulation.
Price structure
Short-term levels treated as decision points for the participants who established them.
An intraday system that trades equities whose observed response to order flow has become directionally asymmetric. It measures how a given name's own order flow moves price in each direction, and acts only when that asymmetry is present, stable, and the move has begun. Positions are built and unwound deliberately over minutes, and held only while the condition that justified entry remains observably true.
Intraday response-asymmetry trading
A system class that acts on how a name's price reacts to order flow, not on a view of where price should go.
Market response measurement
Quantifies the name's own reaction to aggressive flow, live from the session, rather than inferring it from price charts.
Persistent directional flow-to-price asymmetry
Names where aggressive flow in one direction moves price materially more than equal flow in the other, and where that imbalance holds rather than flickers.
Research phase
Pre-build. Measurement design and regime characterization underway.
Contact
Bayfall Morrigan is a private firm. We are not raising capital and this site is not an offering. General correspondence is welcome at the address below.