The Art of Quitting… Or How Expected Value Should Determine Your Career Trajectory

On quitting, expected value, and marking your own seat to market.

 

A senior researcher said something to me a while back that I've thought about a lot since: "I'm optimising this move and basing my decision on how things look over a five-year delta." Not "should I move now?", or "is this offer better than that one?", but which set of moves, including staying exactly where they were, gives the best outcome over the next x years.

Most of the time, my message lands on someone's desk on a random day. If it's a bad day, the reply is as much about the bad day / week as the decision. Sometimes something better comes up and someone moves. Sometimes the shit period passes and it all goes back on hold. What almost never happens is someone sitting down, on a good day, and properly thinking through whether the seat they're in is still the best option against the wider ‘playing board’.

Then, recently I found myself reading Annie Duke's ‘Quit’. The premise of the book, using real life examples and analogies backed by data, is that most of us quit something too late, or make a buy-sell decision weighted far too much by emotion or a knowledge imbalance. This is usually because the things that keep us in place (sunk cost fallacy, the endowment effect, the comfort of the known) all push in the same direction. By the time quitting feels like the right call, she argues, it usually was the right call a while ago.

 

Selling fast and buying slow

There's a chapter in there on a study of institutional portfolio managers (Akepanidtaworn, Di Mascio, Imas and Schmidt, "Selling Fast and Buying Slow") which I found super interesting given my line of work.

These are experienced professionals, and their buying decisions comfortably beat the market. The interesting bit though, was that their selling decisions didn't: in deciding which stocks to exit for liquidity on new ‘buys’, they did worse than if they'd sold a position at random. The explanation the authors land on is attention. Buying got a lot of research, the thesis and the debate; selling got far less of it - which stocks to divest from the portfolio were usually determined by historical performance rather than grounded in research on the future expected value…

Got me to thinking that careers to some extent have a similar arc to this. The role you're in is the position you already hold, and like most holdings it doesn’t get looked at in the same way that you might research/weigh up a move.

You know it in detail, day to day, but how often do you sit and weigh the expected value of the current role/seat, with all of the available data that you have, against the expected value of a potential move to another role/firm, factoring in ALL of the different variables that come into play when considering the delta of a career over a 5-10yr timeframe, including professional, personal, financial, happiness/purpose etc. etc.?

Sure, you think about the factors that keep you there, but only at the point where you’ve actually started to consider a move, do you begin to weigh those factors against the EV of a move... and by the arguments put forward in Quit, by that point you’re probably deciding late against optimal expected value.

I should flag the obvious conflict of interest here: a headhunter finding a book about quitting persuasive is not exactly a shock twist. That said, this isn't one of those "never accept a counteroffer" bullshit pieces. Where what we see says stay, we'll say stay, and plenty of it does.

A note on the numbers: the retention figures come from publicly available data. Everything here is correlation, not cause. The rest is what people have told us over the years about why they move and why they stay.

 

One big trade on thin data

Expected value is just probability multiplied by payoff, summed across the outcomes. In systematic trading, an edge is usually being right a little more than half the time, by stacking lots of small, noisy signals and sizing sensibly.

A career decision is the opposite: one big trade, made rarely, on thin data. And the two sides aren't described with the same quality of data. Your own seat you know in high resolution: a PM's moods, the comp economics, which projects get resources, how long the commute takes on a Tuesday…

The alternative is pieced together from four or five interviews, a careers page, a recruiter who (let's be honest) has a horse in the race, and, if you're lucky, a couple of people you know who can fill in the blanks. Rich data against sparse data, and the rich data usually has more weight. I'm not convinced that's because it's better information, it’s just the more available, and therefore easier to weigh, data.

If you treat it the way you'd treat any other position, you need three things: base rates (how these trades usually play out), signals (how to tell a real change from noise) and a stop-loss (what would make you exit, decided in advance).

 

Base rates: the status quo isn't the risk-free asset

The assumption buried in most stay/go thinking is that staying is the low-variance option. So here's the share of people still at a firm N years after joining, by type of firm (bucketed broadly / generally to account for the 4 typical setups, obviously there is variance in this as well, but for the sake of brevity…):

 

‍

At a pod-shop, half the people who join are gone within three years, and fewer than one in three are still there at five. Staying five years is the minority outcome.

Not all of those exits are choices, either. A meaningful share of pod departures (around one in six) come in clusters from the same firm inside a couple of weeks, outside the usual post-bonus wave. Clusters like that are hard to read as a set of separate decisions.

On these numbers, "stay" at a pod really means a decent chance of moving within a few years anyway, some of it on someone else's timing, with less say over where you land and less leverage on the terms. "Move" just means doing it now, on your own terms.

None of this means you should always be targeting a move (obviously). Look at the prop row: half of people are still there after five years. Some firms are built to keep people a long time with strong cultural / financial retention levers, and for the people they suit, that's a perfectly good reason to stay.

 

The cost of the trade, and why it goes up

The biggest cost of moving is time. When people move between firms, the typical gap between leaving one and appearing at the next is about eight months in the UK and nine to ten in the US. That's the non-compete as it actually plays out (aggregated across ALL employees, so doesn’t factor in the different non-compete variance for seniority / function etc.). It's usually paid, so the cost is less cash and more momentum: the market moves on, and your edge goes a bit stale while you watch it.

(Something we can’t account for in this data is how relevant this specific point is, in the current market conditions, with shifting technology factors in play that have almost zero attributable data to weigh up/against. Something to consider if you’re right now weighing up your own decision…).

The second cost is the one that rises. Unvested deferrals, longer non-competes at the senior end, a buyout that gets harder to fund: the cost of leaving tends to go up with tenure and seniority. If you think of the option to move as something you hold, the price of exercising it keeps creeping up, which is a decent argument against waiting indefinitely for "one more bonus cycle".

Moving isn't free in the other direction either. Seats taken after a move aren't any more durable than anyone else's, and in UK pods, people who'd just moved were noticeably likelier to move again (31% still at the new firm after three years, against 49% generally). Some of that is selection, but serial moving has a compounding cost of its own.

A worked example, with hypothetical numbers. Take a mid-level researcher two years into a pod seat they broadly like. Call their total comp ($/£/€) 400 a year: 150 base and 250 bonus, with half the bonus deferred and vesting over 3 years.

Staying, the pod numbers give them about a 44% chance of still being in that seat three years from now. So "stay" really means a 44% chance of three more years at 400, and a 56% chance of moving anyway at some point, with roughly one-in-ten odds overall that the move happens inside a cluster, on someone else's timing. The same researcher two years into a prop seat has about a two-in-three chance of still being there, which is a very different kind of 'stay'.

Moving now costs (on average) about eight months between seats, and the unvested deferral they walk away from: about 208 after their second award, rising to 250 a year later, where it plateaus as one tranche vests and another lands. Whether a new firm buys that out, and how much of it, is usually where the real negotiation sits.

Which leaves you comparing two things: a 44% chance of three more years in a seat you like (plus a 56% chance of making the move anyway, later, with more on the table and some chance the timing isn't yours), or a move now, on your terms, at today's price. Plenty of people run those numbers and stay, and if the seat is genuinely good, they should. The point is that the stay side has a distribution too, and it's rarely written down anywhere.

 

Signals: what people say, and what they do

Over the years we've always asked people what would make them move and what keeps them where they are. The patterns are consistent.

Scope and progression is the main reason people give for moving, in roughly seven out of ten conversations, and more than four times as often as comp, which comes third behind culture*. A change of leadership or team, and capital or a strategy being cut, sit just behind.

(Being pragmatic, this data is probably skewed to some degree... the unspoken rule generally is "don't lead with comp as the primary factor", but I would prefer we're all honest with ourselves for the sake of our sanity...)

When people explain why they stay, it's mostly about the work: they've built something, they can see progression, the firm is performing etc..

Duke's argument is that the forces keeping us in place mostly work below the level anyone would admit to on a call, so the interesting bit is where what people say and what they do don't line up.

The most common reason for staying, by some distance, is "I've built something here". Sometimes that's entirely fair: if your edge sits on infrastructure, data or relationships that won't come with you, walking away from it is a real cost. But it's also close to a textbook description of sunk cost. And nearly three-quarters of the people who said it also named scope or progression as the thing that would make them move.

 

Progression shows up on both sides.

Of the people staying because they could see progression where they were, four in five named progression as the thing that would make them go. The question people are really asking is when the progression stopped, and most answer that later than they'd like.

Almost nobody tells us they're staying out of fear of the unknown, and hardly anyone mentions unvested deferrals. Taken at face value, the biases Duke writes about barely exist in our industry. But a good share of the people who tell us "not now" have then moved not long afterwards anyway, which isn't what you'd expect if "not now" were a settled decision. (A flat "no" to an approach usually means "not that seat", rather than "not ever"). Biases don't show up in what people tell you about themselves. They show up in what they end up doing.

 

Regime change versus noise.

The most useful distinction I know is between a regime change in your seat (capital pulled, mandate changed, team restructured, a manager gone) and a slow build-up of friction that was always there. A bad quarter, a manager about to rotate, an odd comp cycle: that's noise. Around three in ten people we speak to describe a genuine regime change, and plenty of people who do the diligence conclude their seat is right, which is a perfectly good answer.

    

People like me (you/vous)…?

Scope and progression tops the list at every level. What changes is what comes second. Junior and mid-level people talk about culture and, at the junior end, location: moves at that end are mostly chosen. At the senior end and among PMs, the second reason is more often something that happened to them: capital or a strategy cut (about one in five senior researcher and PM conversations, against one in twenty or fewer below that) or a change of team or leadership.

Nearly four in ten senior people describe a regime change in their seat.

Meanwhile "I've built something here" rises steadily with seniority: barely mentioned by juniors, around one in eight mid-level people, one in five seniors, and more than a third of PMs. So the people with the most reason to stay put are also the people whose moves are most often triggered by something outside their control.

 

The stop-loss

In Quit, Duke's practical tool is the “kill criteria”: decide in advance what would make you leave, and by when. "If capital isn't increased by the review in March, I'll start looking." "If the restructure lands on my desk, I move." It's the difference between drifting and deciding. It's also the homework the PMs in that study did on every buy, but skipped on every sell.

From the conversations, staying tends to look like the better bet when:

  - The firm is built for tenure (prop shops, captive vehicles, pay that ramps with time). The retention numbers back that up.

  - The friction is noise, not signal.

  - Your edge depends on infrastructure that won't travel.

  - The reason you'd move doesn't match what moving would fix. Moving for money rarely cures boredom.

And the other way round when:

 - There's been a regime change in your seat, not just accumulated friction.

  - Several people from your team or platform have left in a short window.

  - You're a pod PM carrying losses under a netting deal. That changes the maths on staying more than most people allow for (if the decision remains in your own control of course!)

  - The cost of leaving is about to step up: the next layer of deferral, or the jump to a longer non-compete (definitely something that’s become more prevalent as firms seek to secure IP / reduce attrition, and increase NC’s out to 18/24 or even 36 months…)

  - You've been saying "not now" for a while.

I can't give you data on kill criteria, only an observation: the people who make these decisions well usually knew their conditions before the moment arrived, and started gathering information while the seat was still good rather than the week they decided.

 

Marking it

Back to that five-year delta. Over five years, the base rates say most people in this industry will change seat at least once, whether they plan to or not. The question is whether you mark the position while you still get to choose the exit.

If you've liked what you read, and would like to talk through your own five-year delta with the future in mind, we're at hello@augmentti.com.

‍

01

[

Related

]

WE'RE Here for

_

Insights. Options. Opportunities. Advice.