A methodology companion to the Research Overview, focused on process architecture, model construction logic, risk controls, and execution assumptions.
Use this page for the process-level explanation behind QSCW Balanced: universe design, sector architecture, factor families, sell discipline, the equity-model volatility exit rule, model construction, and execution assumptions.
A process-level explanation of the research model architecture for readers who want more detail than the teaser provides, without requiring disclosure of proprietary parameters.
QSCW Balanced Research Letter is a rules-based U.S. small-cap research publication built around a systematic model with a defensive orientation. The purpose of this page is to describe how the process is built and governed: what enters the universe, how the opportunity set is narrowed, what the ranking engine is trying to detect, how names are removed, how the volatility exit rule behaves, and what assumptions sit behind the simulation.
If the question is whether the research publication deserves a first look, the teaser should come first. If the question is how the standardized research publication works at methodology level, this is the right page.
The model does not begin with the whole U.S. market. It begins with a structured elimination process designed to retain only names that are genuinely investable, liquid, and structurally transparent.
The pre-screening addresses execution risk, governance quality, and structural stability before any stock enters the ranking process. The purpose is not simply to narrow the list, but to ensure that the names reaching the ranking stage are already compatible with disciplined model construction.
| Dimension | What it controls |
|---|---|
| Size range | Focuses on the investable part of the small-cap spectrum and avoids the fragility of deeper micro-caps. |
| Minimum price | Excludes penny-stock behavior and the spread and manipulation risk that comes with it. |
| Security type | Restricts the universe to U.S.-listed common stocks; non-standard structures are excluded. |
| Liquidity | Requires sufficient daily trading activity and acceptable spread behavior for realistic execution. |
| Turnover controls | Filters out dormant or unstable names where execution quality would be unreliable. |
| Reporting quality | Requires timely and fresh financial reporting rather than stale or irregular filings. |
| Event risk | Excludes pending mergers or similar situations that distort normal selection logic. |
| Crowdedness & instability | Uses additional controls to reduce exposure to structurally unstable or crowded names. |
Specific numerical thresholds are proprietary. What matters publicly is the design logic: execution realism is built into the research model before ranking begins.
The research model uses a deliberately narrow sector map. This is a design decision, not a side effect.
Eligible names are drawn only from a subset of defensive RBICS sectors. The objective is to improve model resilience and reduce exposure to more cyclical, lower-visibility segments of the small-cap market.
The model construction process also includes concentration guardrails so that no single sector cluster dominates the book even if ranking conditions favor concentration at a given rebalance.
The core engine combines more than 70 underlying financial and market variables into a single composite ranking across six conceptual factor families.
The engine is meant to identify situations where multiple independent research inputs align simultaneously. It is explicitly multi-dimensional rather than a single-style model. Factor weights and variable definitions are proprietary and therefore not disclosed here.
The composite score drives both inclusion and retention. A stock's ranking matters not only when it first qualifies, but also when the model decides whether it still qualifies in the research model.
Names are not retained by narrative or discretion. The model rotates toward stronger-ranked opportunities through a rules-based retention standard.
A name is removed when it no longer satisfies the model's minimum retention standard at the scheduled rebalance date. The decision is mechanical and ranking-driven. Original thesis, recent price behavior, or discretionary judgment do not override the rule.
The quality discipline is not confined to a single factor bucket. It is embedded across the pre-screen, the ranking logic, and the retention standard. That means a name with market strength but weak quality characteristics typically does not survive the process for long.
A separate daily-monitored mechanism operates independently of the biweekly review cycle as a binary capital-preservation response to extreme stress conditions in the equity model.
When market volatility exceeds a predefined stress threshold, the model equity allocation moves to cash within the research framework. The mechanism is binary and non-discretionary; there are no gradual intermediate steps.
Re-entry does not happen immediately when the stress condition clears. It waits for the next scheduled biweekly review window, which introduces a deliberate delay in favor of execution discipline.
The research output is concentrated, published biweekly, and shown on an equal-weight model basis. Under normal conditions, the state of the ranking engine at each scheduled review date governs the published model output.
Model changes are not driven by tactical day-to-day intervention. The research model uses a biweekly review cycle to balance responsiveness against execution assumptions. Model weights are shown on an equal-weight basis at each scheduled publication review.
Construction also respects the sector guardrails described earlier, so the resulting model output remains diversified within the defined defensive perimeter.
The simulation is built with non-zero execution friction. The aim is a practitioner-relevant result, not a frictionless theoretical optimum.
| Parameter | Assumption |
|---|---|
| Trade execution | Next-open pricing rather than intraday fill assumptions. |
| Slippage | Variable slippage to reflect differences in price and liquidity sensitivity across names. |
| Commission | Fixed flat-fee commission applied throughout the simulation. |
| Rebalance cadence | Biweekly rather than continuous repositioning. |
| Position weights | Equal weight, reset at each review cycle (static weight). |
| Simulation period | March 2001 – March 2026, spanning multiple market environments. |
This page stops at the level of methodology and assumptions. It does not attempt to reproduce the Research Overview performance presentation or the landing overview layer. Any no-cost 3-issue sample sequence, if offered, consists of the same standardized research publication for all recipients and does not include personalized advice, account-specific sizing, implementation guidance, mandate-specific guidance, or portfolio-specific recommendations.