The MosaicIQ Platform
Technology that augments judgment
MosaicIQ is proprietary, patent-pending, AI-native research software, built and maintained in-house. It organizes our coverage universe around a custom sector taxonomy designed for the Firm’s research framework, and it preserves every analytical output with a timestamp, so any decision can be reconciled against the information that was available at the moment it was made.
End-to-end
One platform, three core workflows
1. Research & selection
Every covered company carries a living thesis: bull, base, and bear scenarios with a set of testable catalysts and risks. Valuation methods adapt to each name, a dual conviction read keeps the universe disciplined, and a thematic layer maps each hypothesis to its likely beneficiaries and to the names it constrains.
2. Entry, sizing & construction
Daily signals flag entry readiness, timed by a technical overlay that never overrides fundamentals. Exposure analytics catch hidden concentration, factor, and theme risk. Sizing follows conviction-weighted recommendations and a decision scorecard for each add, trim, or exit.
3. Performance measurement
Because every thesis, score, and decision is timestamped, the platform grades itself. Theses are measured against outcomes, automated backtests recalibrate scoring weights toward what has actually predicted returns, and a behavioral-bias layer audits the Firm’s own decisions.
What makes it different
Discipline, compounded
A thesis that changes as the facts do
Most investment theses are documents. Ours are data. Each catalyst and risk carries a status, an evidence weight, and a full history of how it got there, so when the facts move the thesis moves with them, and the earlier version is still on record to learn from.
From evidence to estimate
When new research shifts a tracked catalyst or risk, the platform produces a calibrated, audit-trailed adjustment to the estimates. Qualitative judgment and quantitative forecasts stay tied together, with the reasoning preserved.
It keeps score against itself
Price targets, scenario probabilities, and implied upside are checked against what happened. The misses feed back into the scoring weights. The model is built to learn from its own mistakes.
And it watches the analyst, too
The Firm’s own conviction, estimate, and position trajectories are tested against the CFA Institute bias framework, which surfaces anchoring, overconfidence, and herd behavior in our own decisions.
On the roadmap
Toward multi-agent reasoning
The platform does not make decisions. It handles the observation, measurement, and bookkeeping that no single analyst can sustain across a full coverage universe, which leaves more capacity for the judgment that actually drives returns.