PROGRAM / 01Active research
Market Regimes
How can market phases be described as state, transition, and uncertainty without turning them into false certainty?
Research
Questions before products. Methods before claims.
The research index connects each published method, public evidence record, and dataset to the operational system it supports. Agenda entries are research questions—not product announcements.
Published work
A small library with explicit links to evidence.
The current library contains the published method, the continuously updated public evidence record, and its machine-readable dataset.
The BottiAi Regime model
Research question, model logic, evaluation boundary, intended use, and known limitations.
Open →Out-of-sample observations since 2020
The continuously updated, publicly delayed record with complete and open subsequent windows.
Open →Public delayed dataset
Machine-readable SPY prices, regime score, and HIGH observations using one consistent cutoff.
Open →Research programs
One active program. Three questions on the agenda.
Only Market Regimes currently produces an operational system. Other programs describe the direction of inquiry and carry no release commitment.
PROGRAM / 02Research agenda
Liquidity & Stress
Which observable conditions distinguish ordinary volatility from deteriorating liquidity?
PROGRAM / 03Research agenda
Signal Reliability
How can model decay, instability, and regime dependence be measured after release?
PROGRAM / 04Research agenda
Decision Systems
Which interfaces help people use probabilistic evidence without turning it into certainty?
Publication standard
Every useful claim needs context.
Performance-like numbers are published with a denominator, data cutoff, scope, comparison window, and nearby limitation. A system output is not presented as a personal recommendation.
01
Research question
The question comes before the output and states what the work does—and does not—attempt to answer.
02
Data and boundary
Universe, cadence, cutoff, and training versus out-of-sample periods remain distinguishable.
03
Evidence and counter-evidence
Negative, early, incomplete, and contradictory observations stay part of the interpretation.
04
Limitations
Overlapping windows, non-stationarity, missing inputs, and excluded costs are stated where relevant.
05
Operational relationship
Published work links to the system it informs, while paid access never removes the delayed public scrutiny layer.