Our approach is primarily systematic, built on quantitative frameworks and reproducible process. But the framework is a tool, not a master. Experience, pattern recognition, and judgment shape every decision. Transparency isn't a marketing angle. It's how we operate.
The level of unemployment matters less than the direction. We track flows through labor markets, credit markets, and the financial system. Change, not state, is what moves prices.
We don't predict. We read the cycle and position accordingly. The framework tells us which regime we're in. Regime tells us how to size risk. Sizing determines outcomes.
Diversification is for capital preservation. Alpha comes from concentration. We run concentrated, high-conviction portfolios. 30 positions means 30 opinions you're not sure about.
Good process, bad outcome is acceptable. Bad process, good outcome is dangerous. We judge decisions at the time they're made, with the information available. Results are feedback, not validation.
Some research shops cultivate an air of mystery. We show our work. We explain what we're watching, why it matters, and what would change our mind. Every trade has a thesis and an invalidation trigger.
Being right and consensus is low value. Being right and non-consensus is where returns come from. We look for gaps between market pricing and our framework's signal.
Our pipeline pulls fresh data daily from primary government and institutional sources. No third-party aggregators. No approximations. Code-first means reproducibility first.
Raw data flows through standardized transformations to create each pillar's composite index. The MRI synthesizes all 12 pillars into a regime reading. The numbers provide the foundation. Experience and pattern recognition shape the interpretation.
The MRI classifies the macro regime, from supportive through crisis. But regime classification is the starting point, not the conclusion. We overlay qualitative judgment, cross-pillar divergences, and real-time context before making allocation decisions.
We scan for cross-pillar divergences. Divergences are where the edge lives. When credit says one thing and labor says another, someone is wrong. These mismatches across domains are often the highest-conviction signals.
The framework generates signals. We translate those into actionable theses, articulating why the opportunity exists, what would invalidate it, and what the asymmetry looks like. Every position has a written thesis before entry. The data informs. Judgment decides.
Conviction determines size. Higher conviction means larger positions. Lower conviction means smaller or no position. Size is adjusted based on the regime environment. We concentrate capital where the framework and our judgment agree.
Positions are implemented and monitored against both thesis invalidation and price stops. We use dual stops: if the thesis breaks (data changes), we exit regardless of price. If price breaks (stop hit), we exit regardless of thesis.
True global macro is directional, not directionally constrained. We have views. In a world where prices go up most of the time, our bias is towards owning. The framework accommodates wherever the thesis takes us: long or short.
Our primary approach is thesis-driven: three-engine convergence identifies the opportunity, conviction scoring determines the size, and dual stops (thesis invalidation + price invalidation) manage the risk. When the macro picture is clear, we lean in.
Markets don't always wait for the macro to catch up. When our framework signals caution but strong trends are running, we have the flexibility to participate through price-driven positioning with tighter risk management. This keeps us dynamic across different market environments.
When neither the macro thesis nor the technical picture supports risk, we hold cash. Meaningful cash allocations are not a failure of imagination. They're a reflection of discipline. The best trade is sometimes no trade.
Every position is monitored against two independent triggers. If the thesis breaks, we exit regardless of price. If the price structure breaks, we exit regardless of thesis. Either condition is sufficient. This prevents the most dangerous mistake in macro: holding a broken position because "the thesis is still intact."
The MRI is only one layer. Our full risk architecture operates on three distinct levels, each answering a different question. Together, they provide both the strategic view and the tactical response.
Estimates the probability of recession onset within a forward-looking window based on the current state of leading indicators. This isn't a forecast of GDP. It's a probability distribution derived from historical patterns: when these conditions existed before, how often did recession follow?
The model synthesizes labor market flows, credit conditions, yield curve dynamics, and real economy indicators. Lead times vary by indicator, and the model accounts for this by weighting inputs according to their historical reliability at different horizons.
Continuously monitors key indicators against historically significant thresholds. When an indicator crosses into territory that preceded past recessions, a warning triggers. This isn't about prediction. It's about pattern recognition.
Warnings are additive: one warning is noise, multiple warnings demand attention. When warnings across different domains fire simultaneously, the probability of coincidence drops dramatically.
The Macro Risk Index synthesizes all twelve pillars into a regime indicator. It answers a different question than the recession model: not "is recession coming?" but "what is the current risk environment for asset allocation?"
MRI operates in real-time, classifying regimes with specific implications for position sizing, sector allocation, and defensive posture. The ensemble approach ensures no single pillar dominates.
We track rates of change, not levels. The quits rate matters more than the unemployment rate. Credit growth matters more than credit outstanding. Flows lead, stocks lag.
No indicator is reliable in isolation. The yield curve has false positives. Unemployment has false negatives. Combining uncorrelated signals reduces error.
We classify environments, not predict outcomes. Being positioned for a high-risk regime is more valuable than predicting recession timing precisely.
Different indicators lead by different amounts. The model accounts for this, weighting inputs by their historical lead time and reliability at each horizon.
| Attribute | Lighthouse Approach | Consensus Approach |
|---|---|---|
| Signal Source | 12 Pillars → Proprietary Composites | Headlines, lagging data |
| Framework | Three-Engine System | Single-dimensional analysis |
| Position Sizing | Conviction-weighted | Equal-weighted |
| Concentration | Concentrated, high-conviction | 30+ marginal positions |
| Cash Treatment | Active position (meaningful allocation valid) | Residual drag |
| Risk Framework | Dual stops (thesis + price) | Single trailing stop |
| Timeframe | 3-6 month tactical core | Arbitrary calendar |
The 12 Pillars are the diagnostic framework. Asset classes are how we express views. We cover the full spectrum from liquid ETFs to individual securities, depending on client needs and conviction level.
Broad indices, sectors, and single names. Factor tilts and thematic exposure.
Duration positioning across the curve. TIPS, Treasuries, and credit.
Investment grade, high yield, leveraged loans. Spread compression and widening plays.
Gold, energy, industrial metals. Inflation hedges and growth signals.
Dollar positioning, carry trades, EM FX exposure.
Major tokens and select alts. Liquidity-driven framework tied to Fed plumbing and net liquidity dynamics.
These are signals we monitor, not rules we obey blindly. In normal ranges, they're inputs to weigh alongside fundamentals and cross-pillar readings. At extremes, they demand respect. The discipline is knowing the difference.
Where price sits relative to its primary trend. Trading against the trend raises the bar. But mature moves in either direction can offer opportunity. Context determines whether it's a warning or a setup.
The shape of the trend. Deteriorating structure signals weakening momentum, but duration matters. Fresh deterioration is more concerning than stale. Weigh against breadth and momentum.
Is the asset earning its risk? Persistent underperformance is a flag. But relative weakness during sector rotation is different than relative weakness during broad risk-off.
Price momentum standardized across timeframes. At extremes, momentum is telling you something important. In normal ranges, it's one input among many.
How stretched is price? Extreme extension means poor entry risk/reward regardless of direction. But extension in a strong trend is different than extension in a weak one.
Volatility is information, not just risk. Elevated vol compresses position sizing. Term structure tells you whether the market is hedging the present or the future.
The Diagnostic Dozen educational series walks through each pillar in detail. The framework is not a black box. You can understand exactly how we reach conclusions.
Each proprietary index is built from publicly available data sources. We share the components and the logic. The specific construction and weighting is our edge.
When we take a position, we articulate:
When we're wrong, we say so. Post-mortems are part of the process. The goal is calibration over time, not pretending to be infallible.