Telling your desk what you want

Open Strategy › Guidance and write, in the one box on the page, what you would tell your portfolio manager on the phone. "Bet the house, make 400% of our book by new years." "Really over expose me to the OAI IPO." "Tomorrow go ham, try and make a few thousand bucks and sell it into cash, then two days after go back to being more conservative." Press Read it back to me. Nothing has happened yet.

Where

Strategy › Guidance, the box called Tell your desk

What it does

Compiles your words into a card of typed pieces: a goal, a tilt, a dated schedule, a hard rule, a standing preference, a note for now, a push to take more risk inside your limits, a limit change or a posture change

What you choose on the card

For a note or a rule: whole book or one symbol, standing or until a date, and Enforce or Keep as guidance when the desk read a rule the Executor can enforce. Use my words exactly keeps the note as you typed it

What each piece says

Enforced (the machine holds every order to it), Instructed (the desk is told, with your own words attached, until it ends) or Watching (an event the desk waits for)

When it applies

Only when you press Confirm. Anything that loosens your protection also asks for your code and the words MORE RISK

How it ends

Every row of the list shows when it ends; End now on the row, or Take it back on the card you confirmed

What comes back

A card, in a second or two. First a quick reading by the built-in reader; when the desk has read it too (usually within a minute) its reading replaces the first one. The card lists what it understood as separate pieces, each with your exact words quoted under it and one line per effect:

• Enforced. The machine enforces this on every order: a limit ("at -$2,000 on a day the desk halts; until then it keeps trading"), a posture, a dated posture step, or a rule that refuses buys in a name. These reach the Executor; the PM cannot talk its way past them.
• Instructed. The PM, the Researcher and the Quant read this at the top of every session until it ends, with your words attached: a goal ("work toward 400% by Dec 31; a target, not a promise"), an overweight ("prefer qualified entries tied to the OAI IPO, size them at the top of the band"), a day's mode ("Wednesday: day-trade, aim for about $3,000, sell the day's positions into cash before the close").
• Watching. An event the desk waits for ("the OAI IPO"). The piece ends when the event has happened or on its end date.

Untick any piece you did not mean. If the words were a question ("why didn't you buy MSFT?"), the card says so and offers nothing to confirm; ask it in Consult instead.

The choices on the card

A goal and a schedule carry their own dates. A note or a rule carries three small choices, right on its piece, and the card re-reads itself every time you change one, so the lines above the choices always say what Confirm will do.

• Rule. Shown when the desk read something the Executor can enforce (no new buys, never buy a list of names, a cap on how much is invested). Enforce is what a rule means: the Executor refuses orders that break it, and confirming asks for your authenticator code. Keep as guidance turns it into a note: the desk reads it and answers for it, and nothing is refused. Nothing becomes a rule on its own.
• Covers. The whole book or one symbol. A rule covers the names it names.
• Lasts. Standing is read at the top of every session until you end it. Until a date is a nudge for now that the desk must acknowledge at its next session; pick 1 day, 3 days, 1 week or 2 weeks, or keep the day the desk read from your words ("this week" ends Friday). Until an event waits for something a headline would name: we watch the news for it, wake the desk when it may have happened, and ask you monthly whether it still matters.

A limit or a posture change is different: it is written once and has no end date. The piece says stays until you change it, and if your words named a time ("keep 30% cash this week") it says so plainly: the change does not end on its own. When the book follows a posture, the piece also says that confirming moves the book off it (its posture becomes Custom). A cap on how much is invested can be said another way: choose Lasts · Until a date on that piece and the same cap becomes a rule the Executor enforces until that day, which leaves your limit and your posture as they are and lifts on its own. Other limits cannot carry an end date.

Until an event, and the Use my words exactly button beside Confirm, switch the card to your words as typed. The desk then reads the note exactly as you wrote it (up to 4,000 characters) instead of the typed pieces, with the same three choices; if the words read as a rule you are offered Enforce or Keep as guidance per rule before anything is saved, and Build the rule yourself spells a rule out by hand. Back to the desk's reading returns to the pieces. Guidance: standing rules and notes for now has the difference between a rule and guidance, with worked examples.

Confirming, and what needs your code

Confirm applies every ticked piece in one step and writes one entry in the audit log per change. When a piece loosens your protection (a wider loss halt, a bigger position, a riskier posture now or on a scheduled day) the card says exactly what loosens, in your numbers, and asks for the words MORE RISK and your authenticator code, the same ceremony as loosening a limit on Risk › Limits. A rule the machine will enforce (no buys in a name) asks for the code too, because it sweeps working orders. Tightening never needs a code. A switch to the Speculative posture also lists its risks and asks you to type the acknowledgement under them; see Speculative mandate: what it means.

What the desk does with it

Under the box, What the desk is working under is the one list of everything in force: the goal first (there is only ever one; a new goal replaces the old one and says so), then schedules with each dated step and whether it has run, tilts, enforced rules, standing notes, notes for now and notes that wait for an event. Every row says its state (in force, awaits the next session, acknowledged), what it covers and how long it lasts; a row the Executor enforces carries the label Rule. A limit or a posture you changed from a card is listed too, for as long as the book still runs on it, with Take it back. Under a note the desk has answered, one sentence says what it did with it ("The desk read it and applied it."), and In the desk's words opens what it wrote. What the desk is told, in the list's header, opens the exact text the desk receives. The PM reads the list, with your words, right under its mandate at the start of every session and must say in its summary how the day's decisions served the goal or the tilt, or why they could not. The Critic grades that use: citing your overweight to justify a trade in something you never named, or leaning on the goal to skip a stop, is a finding. A goal changes what the desk selects and how it sizes inside your limits; it never moves a limit by itself. If you want the envelope wider, say that too ("I'm willing to lose $5k a day") and it comes back as an Enforced piece with the consequence spelled out.

A dated step is applied by the machine before that day's open and undone on the day the card said ("Friday: back to Conservative"); you can see each step flip from coming to done on the entry. "Sell it into cash by the close" is an instruction to the PM's pre-close pass for that day, quoted on the entry as Instructed: the desk proposes the exits, the Executor still checks each one.

Telling the desk to take more risk

A posture is a starting set of limits. Inside them the desk has habits: a first buy at half size, a small share of the book at risk per idea, a target for how much is invested that sits under the limit. When you tell it to take more risk ("get really risky", "swing for the fences", "I need to double my money by June") the card comes back with a piece called Take more risk inside your limits, and it works the same on every posture, Conservative included.

The piece says in plain words, with your book's own dollars, how the desk will behave differently once you confirm:

• Bigger positions. What a typical new position is worth now and what it would be, and which of your limits is what stops it there.
• More names at once, when your posture's daily allowance has room for them.
• More of the book invested. The desk works toward the top of the range under your invested limit.
• No half-size first buys. An entry goes in at full size the day it qualifies.
• Holding through losses. A losing day or week is not a reason to cut back; stops written at entry still sell, and your daily loss stop and drawdown brake still end trading when they are hit.

None of it moves a limit. A Conservative book told to take more risk uses Conservative's limits fully and never goes past them; the risk engine checks every order exactly as before. That is also why this piece needs no code: nothing about your protection loosens.

When the goal needs more room than your limits give, the card says so honestly instead of pretending. Beside the goal it lists the exact limits that would have to change and to what on your book (for example the largest holding from 10% to 20% and the most invested from 60% to 90%), as its own piece you can tick. Ticking it is a posture change like any other: it asks for the words MORE RISK and your code, a switch to Speculative also asks for its typed acknowledgement, and the change is written to the audit log. Leave it unticked and the desk pursues the goal inside the limits you have, and says in its summaries how much of the room it used. Each limit in the list also links to its row on Risk if you would rather move one number than the whole posture. A goal never loosens a limit by itself.

To stop it, End now on its row in the list below, or Take it back on the card. There is only ever one such instruction in force; a new one replaces the old one and says so.

Taking it back

End now on any row asks once, in the row, then stops it: the desk stops reading it from its next session, an enforced rule lifts within seconds (ending a rule that blocked buys asks for your code), and a schedule that had already changed the posture puts the earlier posture back. Reword puts the row's words back in the box so you can say it differently; end the old row once the new one is confirmed. Take it back on a card you confirmed undoes the whole card: its entries end and the limits or posture it changed return to what they were; if putting a number back would loosen your protection today, that asks for the code like any loosening.