Community / Research Note
WTI $130 in April? Regime shifts and repricing windows
Will WTI Crude Oil (WTI) hit (HIGH) $130 in April?
This post shows a compact workflow for turning a raw trade export into a structured market read. The core idea is to label time into a small number of regimes, then look at price, volume, and the largest prints through that lens.
Outcomes
Yes / No
Model
HMM-style regimes
Extra
Clustering baselines
Regimes (market moods)
Regimes are recurring market environments. Here the trade history is bucketed into time windows, and each window is described using simple features like returns, volatility, and activity. The model then assigns each bucket to one of a few hidden states.
The key value for a trader is isolating windows where pricing behavior changes. When the market switches into a higher-volatility state, slippage and repricing risk typically increase. That can turn “normal” sizing and timing into a bad trade.
Traditional ML view: clustering the market states
In addition to the sequential regime model, a simple k-means clustering model can be applied to the same bucket features. This is a useful diagnostic because it ignores time ordering and asks a simpler question: do the bucket features naturally separate into a few distinct clusters?
For this market, the clustering view helps confirm whether “drift” and “repricing” buckets look meaningfully different in the feature space. If clusters do not separate, it suggests the market is mostly one-mode behavior and regime labels should be treated cautiously.
Price path
The price path shows where belief moved and how quickly it moved. Regimes help separate slow drift from sharp repricing windows.
Volume clusters
Volume tends to arrive in bursts. Those bursts are often the most informative windows to investigate, especially if they coincide with price movement.
Repricing moments
Large single-step moves are a fast way to find the timestamps that matter. They are often tied to information shocks or aggressive positioning.
Largest prints
The largest prints are useful for understanding whether repricing was driven by a few aggressive trades or by broad participation. The chart includes a side legend with maker and taker addresses for context.