phoebe
MARKET WORLD MODEL
PHOEBE / MARKET WORLD MODEL MARKET RESEARCH

READ BETWEEN THE LINES.

Phoebe is building a market world model. It brings market observations and possible price paths together so researchers can see when a trade idea holds and what would invalidate it.

Explore the scenario demo
FIG. 001 ILLUSTRATIVE MARKET TOPOGRAPHY
01 / SCENARIO DEMO INTERACTIVE / LIVE DATA

Explore possible paths.

Choose an asset and a time window. Toggle the scenario overlay to see where the illustrative paths branch from the market price.

READING THE MARKET

Market data is loading.

BTC / USD COINBASE
—

Connecting to Coinbase Exchange…

Price data from Coinbase Exchange. The scenario overlay is illustrative and is not a model forecast.

02 / RESEARCH METHOD FROM OBSERVATION TO REVIEW

READ THE MARKET.
TEST THE IDEA.

The research follows a trade idea from discovery to review. Record the market state, define entry and invalidation conditions, then revisit the idea as the evidence changes.

Align observations
SPOT / PERPETUALS / AVAILABILITY

Align observations across venues using when each record became available. That reconstructs what a researcher could see when the scenario was formed.

Frame scenarios
HORIZONS / QUANTILES / INVALIDATION

Estimate price quantiles over defined horizons. Use the range to ask what would support a trade idea, what would change it, and when to review it.

Evaluate and review
FORWARD TESTS / BASELINES / COVERAGE

Record forecasts before outcomes are known. Compare price errors and interval coverage with simple baselines on the same eligible samples and horizons.

03 / MODEL RESEARCHTHREE RESEARCH DIRECTIONS

Three ways to study the market.

Phoebe brings together complementary approaches to market inference, multi-path modelling, and asset-specific adaptation.

01 / FLASH

Phoebe Beta Flash

Market inference

The default market-inference path studies time-series observations across assets and forecast horizons. It expresses potential price evolution through quantiles, providing a structured basis for assessing scenarios and uncertainty as market conditions change.

02 / PRO

Phoebe Beta Pro

Multi-path OHLCV research

Pro investigates alternative trajectories in OHLCV sequences, treating open, high, low, close and volume as a joint market representation. The research examines how different paths and time horizons can inform a more nuanced view of market dynamics.

03 / TUNE

Phoebe Beta Tune

Asset-specific adaptation

Tune explores asset-specific adaptation through a BONK-focused fitting study. Forecasts are recorded before outcomes and compared with simple baselines over matched horizons, allowing researchers to examine whether adaptation contributes measurable out-of-sample value.

Research in progress. Evaluation covers forecast error, interval coverage and performance against simple baselines.

03 / RESEARCH CURRENT EXPERIMENTS

What we’re testing.

ETH / CROSS-MARKET

Spot and perpetuals

A short-horizon price quantile study aligns an Arbitrum Uniswap spot pool with Hyperliquid ETH perpetual observations.

BONK / SOLANA

Forward evaluation

On a designated BONK/SOL pool, the study records model and baseline forecasts ahead of time and pairs them with outcomes as they arrive.

These are ongoing experiments. It is too early to conclude that the models outperform simple baselines.

PHOEBE / MARKET WORLD MODEL

WHAT
IF?

Start with a market price series. Explore the paths that can branch from it.

Open the scenario demo
PHOEBE / RESEARCH NOTES