●Intel
Seasonality
/intel?view=seasonalityWhat it answers
Which calendar windows have historically paid on an instrument, how sharp the bias is, and how rare it looks once you account for the size of the search that found it.
When to open it
- Before taking a position over the next few weeks: the Clock tab lists every window opening inside your horizon, sorted by Strength.
- To test a window of your own. A bias you already follow — year-end strength, the September dip, the gas cycle — can be drawn straight onto the year curve and scored on the spot.
- To set a risk calendar. The Risk tab tells you which months carry the most realised volatility, instrument by instrument.
- To check whether there is a physical cause. The Why tab puts the price window next to physical stocks and positioning, where the lake carries them.
- To filter the full register. The Search tab opens every window kept across the whole year.
Anatomy
Four modes, selected in the header: Clock, Risk, Why, Search. To their right, a pill gives the build date — green dot = built today, amber = older, with the number of days behind.
Clock stacks a control strip, the Calendar ribbon, the Windows table, the Consensus block and your pinned windows. Clicking any row opens the drawer, which is the heart of the view: the same year curve is both the display and the selection surface.
How to read it
The two scores
They recur everywhere in the view. Both run 0–100.
| Score | How to read it |
|---|---|
| Strength | How sharp the bias is: the size of the average move, how consistent it is year to year, and how many years back it goes. It is the default sort of every table. |
| Rarity | What the window is worth once measured against the size of the search. Shown in the drawer as Beats N% of random windows. A spectacular window found among tens of thousands of attempts is not the same thing as a spectacular window found on its own. |
| RARE | The badge appears only on top-tier windows. A row without one carries no label at all — the call is yours. |
The Rarity chips — All, 80+, 95+ — set a floor on that score.
Clock · the control strip
| Element | How to read it |
|---|---|
| On the clock · {n}d | How many windows open inside the horizon, the selection's top strength, and {'{n}'} rare when any window carries the badge. The dim line gives coverage: instruments swept, and windows tested on each. |
| Horizon | 14d, 30d, 45d, 90d. Changes the depth of the sweep — and with it how demanding Rarity is: a window is more remarkable inside a six-week search than inside a twelve-month one. |
| Rarity / Class | Rarity floor and asset-class filter. Each chip's counter gives the row count. |
| Strongest now | The three best windows: ticker, average move, Strength, and when it opens. Click to open the drawer. |
| Clusters | Baskets several related instruments agree on, with the member count ({'{n}'}×). |
Clock · the Calendar ribbon
One row per window on a real date axis. bar = duration · fill = rarity · dashed = today.
| Element | How to read it |
|---|---|
| Bar length | The window's duration, from open to close. |
| Colour | Green LONG, Red SHORT. |
| Opacity | Rarity. Solid = clears the search comfortably; washed out = LOW RARITY. |
| The dashed TODAY line | The build date. Everything to its right is still ahead of you. |
| CLUSTER bracket | Joins members of the same basket when they sit close enough together in the list. |
| Values on the right | Average move, hit rate, years of history, and when it opens. |
Clock · the Windows table
| Column | How to read it |
|---|---|
| Instrument | Class pill, ticker, name. A ⌐ marks an instrument that belongs to a cluster. |
| Strength | The score, with its bar. Default sort. |
| Window | Open and close dates, then the length in days. |
| Opens | +{'{n}'}d before it opens, today on the day, day {'{n}'} once the window is already running. |
| Side | LONG or SHORT. The direction of the historical bias, not a recommendation. |
| Mean / Med | Average and median move over the window. A wide gap between them means one or two years are carrying the average. |
| Hit / Yrs | The share of years the window went the right way, and how many years are available. 80% on 9 years and 80% on 20 years do not read the same. |
| Best / Worst | The best and the worst year. Worst is the only column that tells you what the window has already cost. |
| Effect | How sharp the bias is: the larger the absolute value, the further the average year stands out from noise. |
| End-of-row tags | RARE where it is earned, then plain facts: PERFECT 12/12, {'{n}'}Y for a long history, +4.2% AVG for a large move. |
| ☆ / ★ | Pins the window. It joins the Watching block. |
Clock · Consensus and Watching
Consensus groups the windows several related instruments agree on. FX pairs are decomposed into currencies first, so a single row can cover eight pairs. The statistic shown is the equal-weight basket's, not an average of its members. Click the header to expand the list — dissenters included: a member whose Mean runs the other way is information, not an error.
Watching keeps your pinned windows, wherever they fall in the year. Only their coordinates are stored; the statistics are recomputed on every read. no history = the instrument has too few years to score that window.
The drawer
It opens on any row, in Clock, Search or Why.
The curve. The instrument's average year, with its envelope. Everything outside the window recedes rather than disappearing — the rest of the year stays readable. Grab the handles and move them: everything reprices live while you drag (a live note appears under the scores), then the panel settles on the server's own numbers the moment you release. You can also draw a window from scratch by dragging on the plot. The Known chips snap back to the windows the register kept on this instrument.
| Block | How to read it |
|---|---|
| Strength / Rarity | Both scores, large. The line underneath spells out what Rarity means, and where the window sits against every window swept. |
| Mean, Median, Won, Hit, Best, Worst | The window's statistics. Won gives the raw count (9 / 12) that Hit states as a percentage. |
| Effect size / t | Two readings of how sharp the bias is. At comparable values, prefer the one resting on more years. |
| Return by year | One bar per year: paid / lost. Look for consistency, not height — two enormous years and ten flat ones are not a season. |
| Path · day 0 → {n} | The trajectory inside the window, with its envelope. Tells you when the move happens: early, late, or spread out. |
| Excursion | Avg heat and Worst heat = what the position sat through on the way; Avg run = what it gave in the right direction. The average move says nothing about the path to it. |
| Cycle | The current cycle's dates. So far = the move since it opened, against the typical at the same point. On script = how closely this year is following the usual shape. Profit factor and Annualised close the card — Annualised is there for comparison, never for ranking. |
| High-impact events | The high-impact releases landing inside the window, with the calendar load (light, moderate, heavy). Each row gives the date, the UTC time, the name and the country. Releases Hawkline tracks elsewhere are clickable and open their panel in Macro, in a new tab. |
The Risk tab
| Block | How to read it |
|---|---|
| Market calendar | Twelve diverging bars. Each counts the instruments whose month ranks above their own mid-year, out of the total ({'{n}'}/{'{n}'}). Above the line = a restless month across the whole universe. The colour lifts when the count runs well past what chance would give. |
| Realised vol · month of year | One instrument per row, twelve months across. Typical = its usual level. Each cell gives the month's median volatility; the tint shows where the month ranks inside the year. A dot marks a month that stands well clear of the rest for that instrument. |
| Spread | The ratio between the busiest and the quietest month. High = a pronounced risk calendar; close to 1 = flat volatility all year. |
Hovering a cell gives the detail: median volatility, its ratio to the usual level, its rank within the year, and how many years the reading rests on.
The Why tab
| Block | How to read it |
|---|---|
| Driver | The chips pick the instrument. The subtitle gives how many series are available and how far back they go. none in the lake = no driver tracked for that instrument. |
| Series cards | Physical stocks or positioning, month by month, on the same calendar as the price. The header gives the source, the years covered and the number of observations. Sharp months read solid, the rest are dimmed. |
| Card footer | The series' unit, then a comparison between two eras (ρ). High correlation = a profile that has held; low = a driver that changed shape, and should not be extrapolated. |
| Price windows | The same instrument's best price windows, so the price bias can be read against the driver directly above it. |
The Search tab
The full-year register, filterable. The subtitle gives how many windows were kept and how many were tested per instrument.
| Control | Effect |
|---|---|
| Leads | Five one-click entry points: Perfect records, Opening this month, Beats 95% of random, Big movers, Short windows. Reset clears everything. A preset only sets filters — they stay visible and editable below. |
| Rarity / Class / Side | Rarity floor, asset class, direction (Both, Long, Short). |
| Days | Length band: Any, 5–15, 16–30, 31–60, 61–90. |
| Opens | Opening month. |
| Min hit | Hit-rate floor, up to 100%. |
| Sort | Strength, Rarity, Soonest, Hit, Size, Length. |
The table is the one from Clock, plus a Class column.
Controls
| Gesture | Effect |
|---|---|
| Click Clock / Risk / Why / Search | Switch mode. Class filters are shared across modes. |
| Click a Horizon chip | Change the depth of the sweep — and rescore Rarity against it. |
| Click a row in Clock, Search or Why | Opens the drawer. Clicking the same row again closes it. Risk rows are not clickable. |
Esc or ✕ | Closes the drawer. |
| Drag a handle on the curve | Redefines the window. The numbers follow the gesture; the panel settles on release. |
| Drag on the plot, away from a handle | Draws a new window, in either direction. |
| ← / → on a focused handle | Nudges by a day; with Shift, by a week. |
| Click a Known chip | Snaps back to a window from the register. |
| Click ☆ | Pins the window into Watching. Click again to unpin. |
| Click Why in the drawer | Switches to the Why tab for that instrument. The button only appears when a driver is tracked. |
| Click a release in the manifest | Opens Macro on that release, in a new tab. |
| Click a column header | Sort. |
Cross-check with other Hawkline features
- A window opening soon: Trend matrix to check that current momentum runs the same way. A season fighting the regime is not a signal, it is a conflict.
- A currency cluster: COT for non-commercial positioning, FX for present relative strength.
- A heavy release in the manifest: Macro carries its reaction history.
- A window that overlaps a desk call: Theses and Brief.
Keep in mind
- The register is small on purpose. Out of tens of thousands of windows tested per instrument, very few come out carrying the RARE badge. That is the expected result: searching that wide mechanically produces beautiful curves on instruments with no seasonality whatsoever, and Rarity exists to throw them out.
- A window needs at least eight years to be evaluated at all. Median history runs to around ten years, so some instruments never surface, and a window scored on eight years is still a window scored on eight points.
- A window you draw yourself is judged against the whole year. No narrow sweep found it, so its Rarity is measured on the widest scope — and will often land below the same window opened from a horizon. That is the honest number, not a regression.
- Strength ranks, it does not forecast. A window at the top of the table is the best-formed one in today's register, which commits to nothing about the year ahead.
- Everything rests on a daily build. The header pill gives its age; a late build serves yesterday rather than nothing. Only the drawer recomputes against the day's prices.
- The Why tab only covers instruments with a tracked driver. No series ≠ no cause: Hawkline simply doesn't track one for that instrument.
- The event manifest is not exhaustive. It keeps only high-impact releases from the economies attached to the instrument, and not all of them link through to Macro.