Candle Characterization and Probabilistic Volatility Mapping
Overview
At first glance, a candle size characterization study might seem like a trivial exercise.
After all, you can see how large typical candles are by glancing at the chart.
But this study is not about surface-level observation — it’s about quantitative calibration.
It defines the environment in which all other analysis operates, giving you a baseline for interpreting every subsequent signal.
The output becomes a probabilistic volatility map — a statistical reference that lets you define:
- What is normal price movement
- What qualifies as a best-case expansion
- What counts as a worst-case tail event
Purpose
This study creates an objective measure of “how far things usually go.”
When you sit down to trade, you can recall these numbers — almost by muscle memory — and immediately anchor yourself in the current regime.
It becomes a mental calibration process:
“A 1σ move on the 15-minute timeframe is about 0.6%.
We’re already 1.8% off open — that’s roughly a 3σ candle.
Either there’s a strong catalyst here or a liquidity event unfolding.”
Once these values are familiar, you stop guessing.
You know where you are in the distribution, and that awareness stabilizes your judgment.
Methodology
For each timeframe (1m → 1d), compute and store:
- Mean and standard deviation of candle body size
- Distribution of wick lengths (top and bottom)
- Percentile boundaries (90th, 95th, 99th)
- Wick-to-body ratio
- Optional: volume weighting, open-interest normalization, or time-of-day segmentation
These metrics define a volatility fingerprint per timeframe.
They’re not static — they evolve — but the baseline stays remarkably stable during similar regimes.
Scenario Framework
| Scenario | Definition | Approx. Probability | Interpretation |
|---|---|---|---|
| Most likely | Within ±1σ | ~68% | Typical regime — price noise or trend continuation |
| Best case | +2σ to +3σ | ~5% | Strong impulse / breakout |
| Worst case | −2σ to −3σ | ~5% | Capitulation or forced unwinding |
This framing transforms volatility into a probability language.
You can apply it directly to position sizing, stop placement, or target estimation.
Practical Example: Current Context
(Note: this is a live reasoning snapshot, not a signal)
“Right now I’m considering taking a large long position.
Funding has flattened, and we’ve already had two liquidation events — the first on October 10th, and a second, smaller one, just a few days ago.
The second event occurred roughly five days after the Fed announcement and wasn’t directly related to it.
After that liquidation, price retraced from ~145 to 160.
Funding data supports that this was indeed a smaller but cleaner second flush.If I were to take this long, I’d need to define a stop that’s statistically sound — not emotional.
For that, I’d refer back to the candle size characterization study:
- What is the typical 4-hour candle size (1σ)?
- How often do we see 2σ reversals after retesting the Bollinger midline?
- Where does a typical post-liquidation reversal stall within that envelope?”
This illustrates why the study matters.
It doesn’t tell you what to do; it tells you how far things can go before your thesis is wrong.
Why It Matters
When trading setups that blend microstructure, liquidations, and directional bias, you’re always answering the same question:
“Is this continuation, reversal, or transition?”
The candle characterization study grounds that question in statistics instead of intuition:
- It helps you decide how much to risk based on typical move size.
- It gives you context for momentum strength vs. volatility compression.
- It reveals when the structure of volatility itself has changed, e.g. after a liquidation cluster.
It’s a defensive layer — a way to avoid overreacting to random volatility or underreacting to real shifts.
Foundation for Further Studies
Once established, this dataset becomes the base for nearly every higher-order QLIR relation:
- Momentum vs. Candle Size: Are large moves driven by acceleration or participation?
- Volume-to-Volatility Ratios: How much volume is required to produce a 1σ candle?
- Open Interest Elasticity: How does leverage expansion contract through the volatility spectrum?
- Trend Exhaustion Detection: What percent of historical 2σ+ candles result in reversals?
This is the ground truth layer that ties your indicators back to physical price movement.
It ensures that when you ask, “Is this continuation or reversal?”, your answer is based on probability, not emotion.
Summary
Even if it doesn’t produce new visual insight, this study provides something deeper — a numerical sense of realism.
It’s the quiet foundation behind confident decision-making: a safety guarantee, a calibration map, and a reference point for everything that follows.
Scenario Calculation Example
Let’s take the 4-hour timeframe for illustration.
Suppose the candle characterization study yields the following statistics (example values):
| Metric | Value |
|---|---|
| Mean candle body size | 1.05 % |
| Standard deviation (σ) | 0.72 % |
| 95th percentile | 2.45 % |
| 99th percentile | 3.20 % |
From this, we can define probabilistic boundaries:
| Scenario | Range | Approx. Probability | Implication |
|---|---|---|---|
| Most likely | ± 1.05 % ± 0.72 % → 0.3–1.8 % move | 68 % | Normal regime, routine chop or gradual trend |
| Best case | Mean + (2 – 3σ) → 2.5–3.5 % | 5 % | Strong impulse, likely event-driven or high participation |
| Worst case | Mean – (2 – 3σ) → –2.5 % to –3.5 % | 5 % | Liquidation cascade / capitulation |
Now, assume you are considering a long with entry at $160:
- Most likely range: 160 ± (1σ ≈ 0.72 %) → 158.85 – 161.15
- 2σ downside: –1.44 % → 157.7
- 3σ downside: –2.16 % → 156.55
If the study shows that 3σ candles occur roughly once per month on this timeframe, you can size your stop accordingly:
Stop: 156.5 (≈ –2.2 %), representing a 3σ tail event.
Target: 162.5 – 164 (≈ +2 – 2.5 %), mirroring the upper σ-band.
This structure transforms the trade into a probability-anchored proposition instead of an emotional one.
Even if the trade loses, the risk was defined in statistical rather than arbitrary terms.
Integrating With Other Studies
Once the σ-bands are known, you can layer in other dimensions:
- Funding & OI: Are tail events more likely when funding resets and OI contracts?
- Momentum: Do candles that breach 2σ coincide with acceleration in VWAP slope or volume delta?
- Microstructure: During liquidation clusters, how often do 3σ downside candles get retraced within one standard deviation on the next bar?
By grounding each of these secondary questions in your volatility envelope, you effectively turn the candle characterization study into the coordinate system for all future analyses.
Closing Thought
The goal isn’t to predict which direction the next candle will go —
it’s to quantify the space in which it can move.
Once you know that space, every subsequent decision — stop, target, or even when to sit out — becomes clearer, calmer, and statistically defensible.
Here’s your Next Steps section, reformatted cleanly and consistently with your study markdown style:
Next Steps
The baseline study is global — it treats all candles equally. A more refined view comes from understanding how distributions change under specific market conditions.
Examples of Future Slicing Dimensions
- Volatility Regimes — e.g., when realized volatility is above or below a threshold.
- Directional Context — candles within a Bollinger-defined uptrend vs. downtrend.
- Temporal Factors — month of year, day of week, or time of day.
- Composite Conditions — any logical combination of filters (e.g., “high vol + uptrend + Friday”).
A Longer Term Goal
Build a conditional distribution engine that can take N concurrent conditions — (volatility, direction, regime, temporal, etc.) — and compute the nearest matching distribution.
This would allow us to ask questions like:
“Given the current market conditions across 100 overlapping signals, what does the candle-size distribution look like, and how unusual is the latest move relative to that context?”