Seasonal windows, on a dated board
Stop trading the calendar from memory.
Sell in May. The January effect. The Santa rally. You have heard the folklore for years and never once seen the receipts.
Locuit screens 402,960 seasonal windows across 138 liquid symbols and puts the ones that held on a dated board — the day each opens, the day it closes, how far it has typically moved, and every single year it did not work.
- 402,960
- windows screened
- 138
- liquid symbols
- ≤ 2022
- years allowed to rank
- 6
- year floor before a rate
Descending · shelf
Every figure on this table is illustrative. Real rows are drawn from the same screen the board reads.
The board, four rows of it
Rows carry their dates and their misses.
The timeline is a rotating 10-week view of a 52-week calendar, and every symbol, start date and window length passes through it. Windows not shown today have not been ruled out — their season is another part of the year. Select a row for the record behind it.
| Row | Symbol | Direction | Window | Typical move | Record | Status |
|---|---|---|---|---|---|---|
| 01 | XRT | up | Nov 12 → Dec 5 18 trading days | +4.2% | 8 of 8 · held out 1 of 2 | Opens in 4 days |
| 02 | UNG | down | Mar 2 → Mar 20 13 trading days | −3.8% | 7 of 8 · held out 1 of 2 | Open now |
| 03 | IWM | up | Dec 18 → Jan 8 14 trading days | +2.6% | 7 of 8 · held out 1 of 2 | In 5 weeks |
| 04 | XLE | down | Sep 4 → Sep 22 13 trading days | −2.1% | 6 of 8 · held out 2 of 2 | Resolved |
Descending · open water
Ranked on history through 2022. Everything after is scored separately and printed beside it.
Why the misses stay on the board
A week that repeats is worth knowing about. A week that almost repeats is worth knowing about more.
Both end up on the same board, carrying the same evidence, and you are the one who decides which is which. Every window carries the day it opens and the day it closes, and says whether that is this week or four months out — so you plan around it instead of noticing it late.
Each row draws one bar per year: green where the move went with the window, red where it went against. A record shown without its misses is a sales pitch. This one hands you the whole strip, and it does not shorten when the strip is unflattering.
Windows are ranked on history through 2022 and nothing further. Every year since is scored separately and printed right beside the in-sample record, whether or not the two agree — and no window is ever chosen for looking good there.
Then you log the trade with how sure you were at the time. The journal scores your stated confidence against your own outcomes, so you find out which of your nine-out-of-tens were really fives.
A record is not a forecast, and nothing here is priced as one.
How a week goes
Three moves, in order, every week.
First — read the tide
Windows four weeks back and six ahead, ordered by the day they open. You see what is live right now and what is coming, before anything has to be decided.
Then — check it against what you hold
Your open positions are matched against the board. If something you already own is sitting inside a window pointing the other way, you are told on the same screen, not in a weekly email.
Last — log it, then get graded
Entry, stop, conviction, thesis. Nothing is prefilled and nothing is written until you save. Over enough trades the journal stops being a diary and starts telling you where your judgement is reliable.
Descending · twilight
These four are independent, not a sequence. Nothing here is ordered.
The part nobody puts on a landing page
What this board will never tell you
Every one of these is a limit that was chosen and accepted, not one nobody noticed. They are on the screen where the numbers are, and they are here for the same reason: you should know what you are getting before you get it.
That a window will work
It reports what happened, how often, and by how much. That is a record. A record is not a forecast, and nothing on the board is priced as one.
That a pattern is meaningful
Screening 402,960 windows means some of them look perfectly consistent by chance alone. Rather than hide that behind a test almost nothing would pass, the board states the width of the search on the screen you read it on.
A clean number it cannot support
Under 6 years of history the record is shown and the rate is withheld — you get "4 of 5", never "80%". Percentages read with an authority small samples have not earned.
Anything about the names that failed
The universe holds companies that still exist today. Ones that were liquid a decade ago and later failed or were acquired are absent, which tilts every result on the board optimistically. That is written on the board too.
Descending · abyss
The tide does not wait to be noticed.
Windows open on dates that are already on a calendar. The only question is whether you are looking at them beforehand.