The offer stage is where power desks lose people

Interview processes at trading firms are usually short and usually good. Traders interviewing traders ask sharp questions, decisions come quickly, and the candidate leaves with a clear picture of the book, the risk lines and the people. The utilities are the known exception, with longer processes and more rounds, but that only extends the point below rather than escaping it. Desks are rightly proud of their interviews. It is also why the losses that follow feel so unfair: the process worked, the person said yes, and six weeks later they are staying where they are.

The failure happens between a verbal yes and a start date, and in power that window has properties that make it the most dangerous phase of the entire hire.

Why the window is worse in power

It is long. Notice periods at senior level run three to six months. A hire agreed in May starts in October, and five months is long enough for a desk's circumstances to change, a market to move, and a person's certainty to erode. Gardening leave, where the current employer negotiates it, changes the window's character rather than closing it: the person is out of the building, the counter-offer loses its daily platform, and the risk shifts from persuasion to drift, a candidate at home for a quarter with no desk pulling them forward. The habits below cover both versions.

It is transparent. Power trading is a small market, and the leaver's value is precisely known. The firm losing the person can calculate what the departure costs in coverage, in knowledge and on the book. Their counter-offer arrives priced against that calculation, not against a salary band. On a desk where one person runs the overnight battery positions or holds the relationships on one border, the replacement cost is visible to everyone, and the counter reflects it.

And it is asymmetric. During notice, the current employer has the person in the building every day. The hiring desk has a signed offer and a start date in the calendar. One side has presence, meals, the familiar screens and colleagues who now know the person is leaving and behave accordingly. The other side has silence, unless it chooses otherwise. Most of the failures in this window are failures of presence, and they were preventable.

The three habits that close the window

The desks that keep their hires between yes and day one are not doing anything elaborate. They have three habits.

The money conversation happened early. If compensation was pinned down honestly in the first two conversations, base, bonus mechanics, buyout of anything unvested, the offer confirms what everyone already agreed, and there is nothing outstanding for a counter to exploit. If the number was left vague until the end, the offer opens a negotiation, and a negotiation is exactly what a counter-offer needs: an open question, a comparison point, a reason to pause. The discipline is uncomfortable because early money conversations feel presumptuous. They are cheaper than the alternative. A candidate whose number was met before offer has been asked, implicitly, whether the number was the point, and their behaviour answers.

There is a stronger version of this, asked before any offer exists: if your current desk matches everything, what happens? The counter-offer piece on this site makes the fuller case for that question; here it is enough to say that candidates answer it honestly while it is hypothetical, and the answer sorts them. The ones moving for the book, the mandate or the build say so, and a counter cannot give them those things. The ones who hesitate are telling you the counter will work, and it is far better to learn that in week one than in month four.

The resignation was planned like a trade. Who the person tells, in what order, on what day, with what in writing, all decided before the offer is signed. Whether they resign to their manager or above them. What they will say when asked to stay a further quarter for handover, which is the soft counter and works more often than the hard one. Whether they want the conversation early on a Monday, so the week absorbs it, or late on a Friday, which hands the current employer a weekend to organise a response and hands the candidate a weekend alone with their doubts. An improvised resignation produces improvised outcomes. A planned one is over in a day, and the plan itself steadies the person, because people who know exactly what they are going to do tend to do it.

Contact never went quiet. The dangerous silence is the stretch mid-notice when the hiring desk has moved on to other problems and the counter-offering desk has the person at the next screen every day. Someone should own the relationship through notice, not chasing, present. A conversation every couple of weeks. The head of desk calling once a month. Anything real to react to: the desk's numbers, a market event, the plan for their first weeks, the name of the person setting up their access. People who unravel in this window almost always describe the same feeling afterwards: the new firm went quiet, and the old one did not.

What the counter looks like when it comes

Assume it comes. On the register's own numbers, most moves in this market are external, which means most employers have been on the losing side of one and have a playbook. The playbook is fast, generous and personal: same week, often same day; money that closes whatever gap exists; and a conversation about loyalty, timing and the desk's plans that no large-firm HR process could deliver. On a small desk the person delivering it owns the P&L and can invent the package in the room.

The defence is not a better counter-counter. By the time you are bidding against the incumbent, the process has already failed; you are negotiating against presence, history and inertia, and paying a premium for the privilege. The defence is that the counter arrives at someone who expected it, has already been paid what they asked, decided in advance what the counter cannot fix, and has heard from their new desk twice since signing. Against that person, the most generous counter in the market is a compliment, not a temptation.

The last mile

One more habit separates the best-run processes: the start itself is managed to the day. Notice served is not the finish line; the finish line is the person at the desk, with access live on the first morning, a named colleague waiting for them at nine, and a written plan for the first month that was shared before they arrived. Power desks lose people between offer and start more often than at interview, and a small number lose them again in a bad first fortnight, which is the same failure wearing a different date. What a well-run first fortnight looks like in full is its own subject, and a later piece will take it up properly. For now the principle is enough: the search ends when the book has them, and a process that behaves that way, visibly, from the first conversation, is itself part of why the person shows up.