Hedge Funds interview preparation

Questions

100

Traced to a firm

39

Firms

16

Updated

September 2026

Asked at

All firmsMan Group10Balyasny Asset Management7Bridgewater Associates3DED.E. Shaw3Apollo Global Management2KKR2Oaktree Capital Management2Point722SCSquarepoint Capital2ACAQR Capital Management1BGBaupost Group1Coatue Management1HPS Investment Partners1Northern Trust1Viking Global Investors1Wolverine Trading1

Topic

All topicsStrategy taxonomy8Stock pitch10Short selling6Portfolio construction8Risk and drawdown8Performance and alpha7Event-driven and merger arb8Distressed and credit5Fund structure and economics7Financing, NAV and operations6Compliance and research process5Quant and systematic6India and Category III AIFs5Career and fit11

Level

AnyCoreIntermediateHard

Type

AnyTechnicalMarket viewBrainteaserCaseFit

1. 002Explain the difference between discretionary and systematic trading.Strategy taxonomyIntermediatetechnicalMan GroupGeneralist · London · 2023

   Say this

   A discretionary manager makes the decision on each trade, using a rules-guided but human judgement. A systematic manager makes the decision once, in code, and then the rules trade without intervention. The real difference is where the human judgement sits: in the position or in the process.

   Then walk it

   1. Discretionary: deep work on few positions. A macro PM might run fifteen expressions of four themes. Breadth is low, so the edge has to be depth of insight.
   2. Systematic: shallow work on many positions. A trend or stat arb book may hold thousands of instruments, each with a small expected edge, and the edge is breadth plus discipline.
   3. The fundamental law of active management is the clean way to say it: information ratio is roughly skill times the square root of breadth. Discretionary buys the skill term, systematic buys the breadth term.
   4. Capacity differs. Systematic strategies hit capacity limits in the market microstructure and can be measured; discretionary capacity is limited by how many names one human can genuinely know.
   5. Failure modes differ too. Discretionary fails through anchoring, averaging down and story-telling. Systematic fails through overfitting, regime change and everyone crowding the same signal.
   6. The blurred middle is where most large firms live now: quantamental. Human thesis, systematic screening, portfolio construction and risk done by the machine. Man Group itself runs both AHL on the systematic side and discretionary equity books, which is worth naming if you are sitting there.

   Where candidates lose it

   Framing it as 'humans versus computers'. Discretionary PMs use enormous amounts of quantitative tooling, and systematic researchers make thousands of judgement calls when specifying a model. Say where the judgement sits instead, and mention the fundamental law if you want to sound like you have thought about it rather than read about it.

   Expect next

   • Which has more capacity, and why?
   • How would you know a systematic strategy had stopped working rather than just having a bad month?
   • Which would you rather work in?

   Reported by candidates at Man Group (Generalist, London, 2023). Source: Wall Street Oasis.

2. 004What is global macro, and how is it different from long-short equity?Strategy taxonomyIntermediatetechnicalGlobal macro

   Say this

   Macro trades the price of money and the economy rather than individual companies: rates, currencies, sovereign credit, commodities and index-level equity. The difference from long-short equity is breadth and depth reversed. Macro takes a few large views expressed in liquid derivatives; long-short takes many small company-level views.

   Then walk it

   1. The instruments give it away. Macro lives in futures, swaps, FX forwards and options because those give you enormous notional exposure with a small cash outlay and can be exited in a day.
   2. The unit of analysis is a policy path, not an earnings number. A typical trade is 'the market is pricing three cuts, I think there will be one, so I am short the front end of the curve'.
   3. Expression matters as much as the view in macro. If you think a currency is overvalued, you can short it outright, own a put, or express it through the rate differential. Each has a different carry and a different way of being right on the view and losing money.
   4. Hit rates are low and honest macro managers say so. You might be right on four trades in ten and still have a good year if the winners run and the losers get cut small. That is why stop discipline is cultural in macro and thesis discipline is cultural in equity.
   5. Carry is the silent factor. Many macro trades pay you to wait or cost you to wait, and a trade with negative carry has to be right quickly.
   6. The limitation: because positions are big and liquid, macro books can be fine on the view and get stopped out by positioning and flow. The 2022 gilt episode is the clean example. The direction was right and the path killed people.

   Where candidates lose it

   Talking about macro views with no instrument attached. 'I think inflation stays sticky' is not a trade. The interviewer wants to hear the expression, the carry and the stop. Name the instrument in the same breath as the view.

   Expect next

   • Give me a macro trade you would put on today and how you would express it.
   • What is the carry on that trade?
   • How would you size it, and where would you stop out?
3. 005What is relative value, and give me an example of a relative value trade.Strategy taxonomyIntermediatetechnicalRelative valueFixed income arbitrage

   Say this

   Relative value is long one instrument and short a closely related one, betting that the price relationship between them converges rather than that either one goes up. Because the two legs are similar, the expected return per unit of notional is tiny, so the strategy only pays after leverage.

   Then walk it

   1. Classic example: the cash-futures basis in government bonds. Own the cash bond, short the future, earn the small mispricing as it converges into delivery. Levered ten or twenty times, a few basis points becomes a real return.
   2. Other standard ones: on-the-run versus off-the-run Treasuries, swap spreads, index arbitrage against the basket, convertible bond arb long the convert and short the equity, and capital structure arb long the bond and short the stock.
   3. The economic function is real. These trades supply liquidity and enforce pricing consistency between related markets, which is why the spreads exist at all.
   4. What you are actually short is liquidity and funding. The trade works while you can hold it; it fails when margin calls force you out at the widest point. LTCM in 1998 and the March 2020 Treasury basis unwind are the same story twice.
   5. So the real risk metric is not volatility of the spread, it is how much the spread can widen before your financing is pulled. Haircut, repo term and counterparty diversification are the actual risk controls.
   6. The honest caveat to say out loud: a relative value P&L looks like a beautiful straight line right up until the day it does not. Sharpe ratios computed on the calm period are meaningless without a stress assumption.

   Where candidates lose it

   Calling it arbitrage without leverage or funding in the answer. Unlevered, these spreads are not worth trading. The moment you mention leverage you have to mention repo haircuts and forced unwinds, and a candidate who volunteers that is immediately more credible than one who says 'risk-free'.

   Expect next

   • How much leverage would that basis trade need to be interesting?
   • What happened to the Treasury basis trade in March 2020?
   • How would you size a trade whose volatility is understated by its history?
4. 006What is statistical arbitrage, and why does it need so much capital and infrastructure?Strategy taxonomyIntermediatetechnicalStatistical arbitrageQuantitative hedge funds

   Say this

   Stat arb takes thousands of small, statistically estimated bets on relative price moves, holds them for days to weeks, and relies on breadth rather than conviction. It needs scale because each position's edge is a few basis points, so you only get a reliable return by running many of them at once with costs under tight control.

   Then walk it

   1. The canonical version is cross-sectional mean reversion: rank a universe on a residual signal, go long the bottom decile and short the top decile, rebalance frequently, hold nothing idiosyncratic enough to matter on its own.
   2. The maths is the fundamental law again. If your per-bet edge is small but your bets are numerous and roughly independent, the portfolio Sharpe scales with the square root of the number of bets. Two thousand weak signals beat ten strong ones on a risk-adjusted basis.
   3. That is why infrastructure is the moat: you need clean point-in-time data, survivorship-free universes, a risk model to neutralise factors, a cost model, and an execution stack that can turn over a large book without paying away the edge.
   4. Transaction cost is not a detail, it is the constraint. A signal with 8 basis points of gross edge and 6 basis points of round-trip cost is not a strategy. Most stat arb research time goes into cost and capacity, not into finding signals.
   5. It is also crowded. Many desks run correlated versions of the same value, momentum and reversal residuals, which is why quant equity has coordinated drawdowns. August 2007 is the textbook one: deleveraging in one corner forced liquidation across the whole cohort.
   6. The limitation to state plainly: the alpha decays. A signal that worked for five years will be arbitraged, so the business is really a research pipeline that retires signals as fast as it adds them.

   Where candidates lose it

   Describing it as pairs trading and stopping. Pairs trading is the toy version. What makes it a real strategy is the factor-neutral portfolio construction, the cost model and the research pipeline, and those are what the interviewer wants to hear you mention.

   Expect next

   • How would you test whether a signal is just a repackaged momentum factor?
   • What happened in the August 2007 quant quake?
   • How do you estimate capacity for a signal?
5. 015How do you think about the valuation drivers of a name?Stock pitchIntermediatetechnicalBalyasny Asset ManagementEquity Hedge · Chicago · 2021

   Say this

   I reduce the multiple to its drivers rather than treating it as a given: growth, return on incremental capital, and risk. Two companies on the same multiple with different reinvestment economics are not priced the same, and that gap is usually where the trade is.

   Then walk it

   1. Start from the identity. Value is this year's cash flow, grown at g, discounted at r. So the multiple is a function of growth, the cost of capital and how much capital the growth consumes.
   2. Reinvestment is the part people skip. Growth is only valuable if the return on incremental invested capital exceeds the cost of capital. A company growing 15 percent at a 6 percent return on capital is destroying value while looking exciting.
   3. So I run three numbers on every name: organic growth, return on incremental capital, and free cash conversion. Those three explain most of the cross-sectional multiple dispersion inside a sector.
   4. Then I use a reverse DCF to make the multiple concrete. At today's price, what growth and margin does the market require? That converts an abstract multiple into a testable forecast I can agree or disagree with.
   5. Then the risk side: earnings duration, cyclicality, customer concentration, and leverage. A levered cyclical deserves a lower multiple on trough earnings, and mechanical peer-multiple comparisons miss that entirely.
   6. The limitation I would state: multiples embed the market's view of duration, which is unobservable. That is why I use the reverse DCF to find the implied assumption rather than arguing that 14 times is cheap because the peer is on 17.

   Where candidates lose it

   Answering with a list of valuation methodologies. The question asks what drives value, not which spreadsheet you build. Growth, return on incremental capital and risk, then a reverse DCF to make it concrete. A candidate who says 'DCF, comps and precedent transactions' has answered a banking question in a hedge fund interview.

   Expect next

   • Two companies in the same industry trade at 12 and 22 times. What could justify that?
   • How do you use a reverse DCF?
   • When is a low multiple a trap?

   Reported by candidates at Balyasny Asset Management (Equity Hedge, Chicago, 2021). Source: Wall Street Oasis.

6. 020How is a short thesis different from a long thesis?Short sellingIntermediatetechnicalLong-short equity

   Say this

   The payoff is inverted and the clock runs the other way. A long can compound while you wait and your loss is capped at 100 percent; a short bleeds carry while you wait, your loss is unbounded, and the position grows as it moves against you. So a short thesis needs a catalyst where a long thesis can survive on patience.

   Then walk it

   1. Asymmetry first. A short that halves makes you 50 and the position shrinks. A short that triples loses you 200 and the position has tripled in size. Risk management is therefore built into the thesis, not bolted on.
   2. Time is a cost. You pay borrow, you owe the dividends, and equity markets drift upwards, so a short has a negative expected return from the market factor alone. That is why the market drift is roughly a 7 to 9 percent annual headwind you have to beat.
   3. Reflexivity is against you. A falling stock can be rescued by a buyback, an equity raise, an activist, a takeout or a short squeeze. A rising stock has no equivalent mechanism working against a long.
   4. Information dynamics differ. Company access is worse, management will not help you, sell-side coverage is almost uniformly positive, and you are arguing against the promotional side of the market.
   5. So the thesis has to be harder edged: fraud or accounting distortion, a genuine structural decline, a funding wall, or a specific dated event. Vague overvaluation is a long thesis in reverse and it does not survive.
   6. And sizing discipline is different in kind. Most disciplined books cap single-name shorts well below the maximum long, and many will not short a name with heavy retail ownership at all regardless of the thesis.

   Where candidates lose it

   Treating a short as 'the opposite of a long'. It is not symmetric in payoff, in carry, in information access or in position growth. If you can only describe it as a mirror image, a PM will assume you have never run one and will not trust you with the short book.

   Expect next

   • How much smaller would you size a short than a long of the same conviction?
   • Would you ever short a name with 20 percent of the float short?
   • How do you deal with the market drift working against you?
7. 022What is the borrow cost, and how does it change the hurdle on a short?Short sellingIntermediatetechnicalLong-short equityPrime brokerage

   Say this

   Borrow cost is the annualised fee you pay to keep a stock on loan, and it is the carry on the trade. It tells you how fast the thesis has to work: a general-collateral name at 40 basis points is essentially free to hold, while a 20 percent borrow means you lose the argument if you are right in eighteen months instead of six.

   Then walk it

   1. The range in practice is wide. Easy-to-borrow large caps are 25 to 75 basis points a year. Crowded shorts run 5 to 20 percent. Genuinely restricted names, small floats, recent IPOs, situations in a squeeze, can go past 50 percent and occasionally past 100.
   2. Do the hurdle arithmetic out loud. At a 15 percent borrow, a short that takes two years to work has cost you roughly 30 points of carry. If your target is 40 percent downside you have given away most of the trade to financing.
   3. Add the dividend. You owe it, so a 4 percent yielder at a 3 percent borrow is a 7 percent annual drag before the stock moves at all.
   4. Borrow is also information. A fee that is rising fast tells you the crowd is arriving, which raises squeeze risk exactly when your thesis feels most obvious. I would check the fee trend and days-to-cover before adding, not just the level.
   5. It also changes how you express the trade. Above a certain borrow it is cheaper to buy puts or use a swap, because the borrow is embedded in the derivative price and at least the loss is capped and the position cannot be recalled.
   6. The limitation: borrow is not contractual term funding. A cheap borrow today can reprice or disappear tomorrow, so a thesis that requires a specific borrow cost to work is a fragile thesis.

   Where candidates lose it

   Quoting a borrow cost with no view on how it changes the trade. The number is not the answer; the hurdle is. Multiply the fee by the expected holding period, add the dividend, and compare it to your target return. Candidates who skip that step are pitching shorts they could not actually hold.

   Expect next

   • At what borrow level would you use puts instead?
   • What does a rapidly rising borrow fee tell you?
   • How do you think about the short rebate on the proceeds?
8. 023What is a short squeeze, and how do you manage the risk of one?Short sellingIntermediatetechnicalLong-short equityMulti-manager platforms

   Say this

   A squeeze is a self-reinforcing rally driven by shorts being forced to buy. Price rises, margin and risk limits bite, shorts cover, that buying pushes the price higher, which forces the next cohort out. The defence is almost entirely position sizing and instrument choice, because once it starts you cannot argue with it.

   Then walk it

   1. The mechanics need two ingredients: crowded short interest and constrained supply of stock. Small float, high short interest as a percent of float, and high days to cover are the measurable warning signs.
   2. Then the accelerants. Borrow recalls that force buy-ins, option market makers hedging calls by buying stock, which is the gamma squeeze layer, and index or passive holders who cannot lend more.
   3. Numbers I would actually look at before shorting: short interest above roughly 15 percent of float, days to cover above five, borrow fee rising week on week, and whether retail and options volume are unusually high relative to the market cap.
   4. Management during the event is mostly pre-committed. Size the position so a 50 percent adverse move is survivable, set the stop before you enter, and never average into a squeeze on the argument that the thesis is now better.
   5. Instrument choice is the cleanest structural defence. A put caps your loss, cannot be recalled, and lets you keep the view through the violence, at the cost of paying premium and needing to be right on timing.
   6. GameStop in 2021 is the reference case and the lesson is not 'retail beat hedge funds'. It is that a correct fundamental view on a business can be irrelevant to the outcome of a position when the funding and float mechanics turn against you.

   Where candidates lose it

   Answering with 'I would cover'. Everybody covers; the question is what you did before the squeeze started. The content of a good answer is the pre-trade checks, the size cap, and the choice to use options in crowded names. Also do not cite GameStop without saying what it actually taught you about position sizing.

   Expect next

   • What short interest level would stop you putting the trade on?
   • How does option market maker hedging contribute?
   • Would you rather be short the stock or long a put in a crowded name?
9. 024Would you ever short on valuation alone?Short sellingIntermediatetechnicalLong-short equity

   Say this

   Almost never as a standalone position. Expensive can stay expensive for years while you pay borrow and the market drifts up, and multiples are the slowest thing in the market to mean-revert. Valuation is the size of the prize, not the reason the trade works.

   Then walk it

   1. The structural problem is that a high multiple is usually a statement about expected growth, and the way to make money short is for the growth to disappoint, not for the multiple to be high. So the thesis has to be about the numerator.
   2. Carry makes the timing problem expensive. At a 5 percent borrow and a 7 percent market drift, you need roughly 12 percent of downside a year just to break even on a valuation short.
   3. Expensive names are also the most reflexive. They can raise equity cheaply, buy growth, get taken out, or get repriced higher by one more year of delivery. Every one of those is a loss for a valuation short.
   4. Where valuation does earn its place is as a multiplier on a real thesis. If you already believe the unit economics break, a 40 times multiple means the derating on top of the earnings miss gives you a much larger move. That is a valuation-assisted short, not a valuation short.
   5. It also works better in relative form. Long the cheap compounder, short the expensive peer with the same end market, sized to be sector neutral. Now you are trading the spread rather than betting on the market's willingness to pay.
   6. The honest exception: in a clear liquidity-driven bubble with a dated funding event, valuation shorts do work. But you need the funding wall or the lock-up expiry, and that is a catalyst, which means you were not shorting valuation alone after all.

   Where candidates lose it

   Saying yes with enthusiasm. It is the most common way junior candidates reveal they have never held a short through a rally. The credible answer names the carry, the drift and the reflexivity, and then explains how valuation is used as a multiplier on a fundamental thesis rather than as the thesis.

   Expect next

   • So what does make a good short?
   • How would you express it in relative form?
   • When has a valuation short actually worked?
10. 026What is the Kelly criterion, and why does nobody run full Kelly?Portfolio constructionIntermediatesuperdayQuantitative hedge fundsMulti-manager platforms

    Say this

    Kelly gives the bet size that maximises the long-run growth rate of capital: you bet your edge divided by the odds. For a simple even-money bet it reduces to twice your win probability minus one. Nobody runs it full because it assumes you know your edge precisely, and overestimating it is punished geometrically.

    Then walk it

    1. The formula for an even-money bet: f equals 2p minus 1. With a 55 percent chance of winning, Kelly says bet 10 percent of capital. With a 60 percent chance it says 20 percent, which already feels reckless to anyone who has run a book.
    2. It maximises the expected log of wealth, which is the geometric mean, not the arithmetic mean. That is the right objective if you are compounding one pot of capital forever, and it is why the answer is a fraction of wealth rather than a fixed amount.
    3. Full Kelly produces brutal drawdowns by design. The expected maximum drawdown on a full Kelly strategy is roughly 50 percent. No fund with outside capital survives that, because investors redeem long before the long run arrives.
    4. The estimation problem is worse than the volatility problem. Kelly is very sensitive to the edge, and betting twice the Kelly fraction takes your expected growth rate to zero. Since we estimate edge from short, noisy samples, err low.
    5. So in practice people run a quarter to a half Kelly, which gives up a modest amount of growth for a large reduction in drawdown. That trade is almost always worth it with client money.
    6. The other limitation for an equity book: Kelly is single-bet. Real portfolios have correlated positions, so the useful version is a mean-variance or risk-parity construction with a Kelly-style scaling on the whole book, not per position.

    Where candidates lose it

    Reciting the formula and stopping, or claiming you would use it. The interesting part is the estimation error and the drawdown profile. A candidate who says 'half Kelly because I cannot estimate my edge to two decimal places' has answered the question properly; one who says 'bet edge over odds' has quoted a Wikipedia line.

    Expect next

    • What happens if you bet twice Kelly?
    • How would you extend it to correlated positions?
    • How would you estimate your edge in the first place?

Firm tags come from public, anonymous candidate reports on Wall Street Oasis: strong signal, not sworn testimony. Firms are named as the places a question was reported, not as partners of Fin Maverick. Answers are written for this page to show how to think out loud; they are not scripts to recite.