Bias Explorer looks for recurring patterns in an instrument’s behaviour: times of day, days of the week, days of the month or year when, historically, the instrument has tended to move in one direction more than another. A “bias” in this sense is simply the average of the movements recorded at each point in the cycle (every time of day, every weekday…), accumulated over time to make the pattern visible as a curve.
It’s not a ready-to-use strategy: it’s an exploration tool. It helps you discover whether a seasonal or intraday pattern exists on an instrument, how persistent it is over time by comparing different periods, and — with the volatility-regime analysis — whether that pattern behaves differently depending on the market context (low VIX vs high). What to do with that observation — whether to build a filter or a strategy on it, or simply keep it in mind — remains your decision.
An important premise. QANTHOS is an analysis, validation and discovery tool, not a financial advisor. The biases shown are statistical patterns observed in history: they aren’t predictions, they aren’t operational recommendations, and their future persistence isn’t guaranteed. A pattern that’s been consistent in the past can weaken or disappear.
1. What it loads: the analysis data
Bias Explorer reads data files exported from MultiCharts for an instrument. From the left panel:
- Data folder — select the folder containing exported files.
- Ticker dropdown — lists found files alphabetically; Refresh reloads the list if you add a file to an already-open folder.
- Load Data — loads the selected file. The summary shows the available period, number of bars, timeframe and the instrument’s Big Point Value.
The data folder path is remembered across sessions.
2. The concept of bias and how it’s computed
For each analysis type, QANTHOS groups all historical bars by their position in the cycle (for example: all 09:30 bars, all 09:31 bars, etc.), computes the average movement in each group, and then progressively accumulates those averages along the cycle’s axis. The result is a curve making visible whether, on average, the instrument tended to rise or fall at certain recurring moments.
Two calculation modes, selectable with the Perc / Abs buttons:
- Percentage — uses the log-return between consecutive bars: the more correct choice for comparing different periods or instruments with prices that changed substantially over time.
- Absolute — uses the currency price difference: more intuitive for a single instrument over a narrow period.
Five bias types, each with its own grouping:
| Type | Groups by | What it reveals |
|---|---|---|
| Intraday | time of day | recurring patterns within the session (e.g. tendency in the first or last hour) |
| Weekly | day of week + time | systematic differences between weekdays |
| Monthly | day of month | start/end-of-month effects |
| Yearly | day of year | annual seasonality |
| Volume (Intraday) | time of day | the average volume profile within the session (has no “direction”: it’s a level, not a delta) |
3. Comparison periods: From / To / N segments
A pattern observed over a single historical period could be random. Bias Explorer lets you split history into several segments and overlay their respective curves, to see whether the pattern repeats consistently over time or is an artefact of a single interval.
Three controls in the Plot Panel:
- From year / To year — the overall historical interval to consider.
- N segments — how many contiguous segments to split it into (segments are whole-year, non-overlapping buckets; if years don’t divide exactly, the remainder is distributed across the first segments).
- Include Total — if enabled, also adds the curve computed over the whole interval as a reference.
The Plot all years available button automatically sets From/To to the maximum available in the loaded data and one segment per year — useful as a first overview.
Why persistence matters more than a single number. A bias curve over a single period can always hide an isolated event (news, a market shock) that shifted the average. Seeing the same pattern repeat, with similar amplitude, across different, non-overlapping historical segments is far stronger evidence of its recurring nature.
4. Computing and reading a bias
With data loaded and periods set, the five Intraday / Weekly / Monthly / Volume / Yearly buttons compute and show the corresponding bias in the dedicated sub-tab on the right; the Calculate BIAS button recalculates the currently displayed one. Each sub-tab shows the curves of all selected periods overlaid, for direct visual comparison.
On the chart you can place 2 pins (right-click) on two points of the curve: the delta between the two points is shown, and it enables the volatility-regime analysis (§5).
An additional 📊 Price + Volume sub-tab shows the price bias and the volume profile overlaid on the same chart, to visually correlate moments of directional push with moments of higher activity.
5. Volatility-regime (VIX) analysis
The 🔥 Volatility BIAS sub-tab lets you break down a bias by the market’s volatility regime at the moment each bar occurred, using intraday VIX data (not the daily close, to avoid introducing information unavailable at the time of each bar — a VIX value is always the last known one at that instant, never a future one).
The default bands are:
| Band | Threshold |
|---|---|
| Low | VIX < 15 |
| Medium | 15 ≤ VIX < 25 |
| High | 25 ≤ VIX < 35 |
| Extreme | VIX ≥ 35 |
The thresholds are customizable. The bias is recalculated using only the bars falling in each band, letting you answer a specific question: is this pattern stronger, weaker, or even reversed when the market is jittery?
There’s also a more targeted path: after placing 2 pins on a curve segment (in any bias sub-tab, not just Volatility), the 🔍 Analyze Bias by VIX Regime button becomes enabled and opens a dedicated report breaking down that specific segment — not the whole curve — by VIX regime. It’s the way to check whether a pattern you’ve noticed in a particular time window depends on the volatility context.
VIX data for this analysis comes from the dedicated volatility-analysis tab: if it hasn’t been loaded there yet, the regime analysis flags that VIX data isn’t available.
6. Auto-Detect: automatic bias search
Spotting a promising pattern by eye on a curve with dozens or hundreds of points isn’t always practical. The 🤖 Auto-Detect Bias button automatically scans every possible segment of the current curve and returns the most promising ones, ranked by amplitude (how pronounced the movement is) and persistence (how much it repeats across the compared periods).
A few notes on how it works:
- Requires at least 2 active periods (Total + a segment, or multiple segments) to be able to assess persistence — with a single period there’s nothing to compare.
- Not available for the Volume bias: the volume profile has no direction (it’s a level, not a delta), so searching for directional biases wouldn’t make sense on this type — QANTHOS flags this explicitly instead of returning an empty result with no explanation.
- The report lists up to 20 candidates, each with start, end, amplitude and persistence.
- Double-clicking a candidate in the report automatically places the 2 corresponding pins on the active chart, shows the delta and directly enables the VIX-regime analysis — a direct path from automatic discovery to context verification.
7. Common issues and notes
“No file found in the folder.” The selected data folder contains no files in the expected format, or is empty. Check the path or use Refresh after adding a file.
“Invalid segments.” The From/To interval and the number of segments aren’t consistent (for example more segments requested than years available). Reduce N segments or widen the interval.
Auto-Detect not available for Volume. Intended behaviour (see §6): the volume profile has no direction to detect as a bias.
The Analyze by VIX Regime button stays disabled. You need to place 2 pins (right-click) on the active chart: the button only enables with a selected segment.
“VIX data not available.” Load VIX data first in the dedicated volatility-analysis tab: the regime analysis requires it as a source.
Need at least 2 periods for Auto-Detect. Enable Include Total and set at least one additional segment in the Plot Panel.
8. An example workflow
A typical exploration path, not a prescription.
- Select the data folder and load the ticker of interest with Load Data.
- Set From/To to the full available history and a reasonable number of segments (e.g. 4).
- Compute the Intraday bias (or the type of interest) and observe whether the different segments’ curves overlap similarly — a sign of persistence.
- If a pattern emerges, try Auto-Detect Bias for a systematic scan of all possible segments.
- On an interesting candidate, check whether it depends on the market context with Analyze Bias by VIX Regime.
- Document the observation (screenshot, manual export of the pattern’s parameters) for possible further investigation — Bias Explorer remains a discovery tool, not one for formal statistical validation.
A bias observed in history is a starting point for investigation, not a conclusion. Persistence across multiple periods and volatility regimes strengthens confidence in the pattern, but only rigorous out-of-sample testing (or future observation) can confirm its practical usefulness.
9. To learn more
- Surface Explorer — complementary robustness analysis, on numeric parameters instead of time patterns.
- Strategies Tab — the starting point for developing the strategies you then explore here.