Expected Value and Bet Sizing

Imagine you’re at a carnival game booth faced with two options: Game A costs $5 to play and you have a 50% chance to win $15. Game B costs $5 and you have a 10% chance to win $100. Which should you play? Game A feels like a safe bet - you win often. Game B is riskier but a bigger prize. To decide, you can compute the expected value (EV) of each. Game A: 0.5 chance of winning $15 yields an average payoff of $7.50, but you pay $5, so net EV = $2.50 gain per play. Game B: 0.1 chance of $100 gives $10 average win, minus $5 cost net EV = $5.00 gain per play. So while B wins rarer, each attempt on average yields $5 value, double that of A’s $2.50. Over many plays, B is the better bet despite longer odds. This concept, expected value, is the cornerstone of rational decision - making under uncertainty. It forces you to look beyond win/lose or best/worst case and consider the weighted average outcome. It’s like taking all possible futures, weighting them by how likely they are, and seeing what the “average future” looks like. If the EV is positive, the bet is, in the long run, beneficial; if negative, it’s a losing proposition. But EV alone isn’t everything - how you size bets relative to your bankroll or risk tolerance matters too. Even a high - EV bet can bankrupt you if you wager too much in one go and hit a bad streak. So let’s break down how to calculate EV and how to decide how big a bet to make given your risk budget.

Calculate expected value explicitly. The formula sounds fancy, but it’s straightforward: sum of (probability of outcome × payoff of outcome) for all outcomes. Include negative payoffs (costs or losses) as negative values. For a simple two - outcome scenario like a project success or failure, it’s EV = P(success) × gain_if_success + P(failure) × loss_if_failure. Suppose you’re considering an optional marketing campaign that costs $10k (that’s a loss if it yields nothing) and has a 30% chance to bring in $50k extra revenue. The other 70% of the time, maybe it does nothing (so you just lose the cost). EV = 0.3 $50k + 0.7 $0 - $10k cost = $15k - $10k = +$5k. That’s a positive EV: on average, a $5k net gain. If you could do this campaign repeatedly, you’d expect profit over time. If EV came out negative, it means the likely payoff doesn’t justify the cost (on average you lose money). When comparing options, compute EV for each, and also consider doing nothing (EV = 0 baseline). It’s enlightening to see the numbers. Sometimes an option with lower probability of success still has higher EV because the payoff is huge, as with our carnival games. Another time, a nearly sure win might have such a minimal reward it’s not worth it. EV cuts through emotional bias - we often overvalue sure things or undervalue long shots - but EV shows the real worth. It’s important to be as honest as possible with probabilities and payoffs in the calculation. If you’re unsure, use ranges: e.g., EV low case vs EV high case. And factor in all components: if there are costs regardless of outcome, subtract them. For example: “EV of hiring this salesperson = 50% chance they boost sales by $200k and 50% chance they don’t, minus their salary.” That’s 0.5$200k + 0.5$0 - $100k salary = $100k - $100k = $0 EV. Not exciting; essentially break - even in expectation, though other intangible benefits or worst - case considerations might swing your choice. Also, EV isn’t just for money. You can assign utility or value to things like time, satisfaction, strategic position, etc., though that’s fuzzier. For instance, “Should we pursue Project X? 20% chance to open a new market (very valuable), 80% chance it fails which is time wasted.” You might qualitatively weigh the value of new market vs lost time, or assign points. The key is to articulate those outcomes and their likelihoods, then see which choice yields more expected “points.” Even without precise numbers, thinking in EV terms (probability impact) clarifies the risk - reward balance.

Mind your risk budget and downside. Expected value tells you on average what you might gain, but it doesn’t tell you about the variability or worst - case scenario. That’s where bet sizing comes in. A positive EV bet is good, but only if you can afford the worst - case loss without ruin. Think of your total capital or “risk budget” as something to protect. A classic rule is: never bet so much on one chance that a loss would wipe you out or put you in a catastrophic position. In personal terms, don’t gamble your entire savings on a stock option - even if EV is positive - because a low probability bad outcome could ruin you. Instead, size the bet small relative to your total bankroll. In business, if an initiative has great EV but if it fails the company dies, it’s essentially betting the company - such one - way door high stakes bets should be deeply scrutinized or broken into smaller pieces. So determine your “downside tolerance”: the maximum loss you could absorb. This might be in money (e.g., “We can afford to lose $50k on experiments this quarter”) or time (“I’m willing to invest 6 months into this venture; if it fails, I can recover”). Use that to cap how much you stake. If an opportunity has high EV, you might still partake in it multiple times, but each individual bet should be small enough that a string of bad luck doesn’t sink you. If it’s a repeatable scenario, one strategy is to bet a fixed small percentage of your bankroll each time (e.g., 5% of investable funds per trade) so losses are limited and wins compound gradually. As you succeed, your bankroll grows and thus your bets sized by percentage can grow too, but you can’t easily lose everything at once. The famous Kelly Criterion from gambling theory suggests an optimal fraction of your bankroll to bet based on edge and odds; you might not compute it exactly, but the spirit is: the bigger your advantage (EV relative to stake), the more you can justify betting, but still remain fractional. If EV is slight and risk high, bet very small if at all. Another nuance: consider correlation of bets. If you have multiple risks that could fail together (correlated), your combined downside is bigger than each alone. For example, don’t invest all in multiple companies that all depend on the same market trend (because if that trend falters, all fail - essentially one big bet). Diversify independent positive - EV bets to smooth out outcomes. That’s why, for instance, venture capital invests in many startups: each is high risk/high EV, but they know some will bust, some boom, and overall they capture the average without one failure destroying the fund.

Prefer frequent small bets over rare big bets. If you have a positive EV opportunity that can be tried multiple times or in smaller pieces, it’s often wise to do that rather than one big swing. Many small bets let you benefit from the law of large numbers - the average outcome will converge to EV with enough trials, reducing variance. It’s like the difference between putting all your savings into one stock versus into an index fund: the index (lots of small bets) is likely to yield close to average returns reliably, whereas one stock might double or go to zero (more volatile). Same principle in projects - if you can run 10 small experiments (with say 60% chance each to succeed) rather than one giant project reliant on everything going right, you increase the chance that at least some succeed and yield benefit. This approach also gives you learning and the ability to adjust (after each small bet, update your base rates and probabilities, improving future EV calculations). It’s not always possible, but think creatively about dividing big decisions into phased decisions or pilot programs. For example, instead of a full product launch to a new market (all - in bet), you might do a test launch in one region (small bet) to gauge response - if it works, expand (like doubling down on a winning bet). In finance terms, this is akin to not investing your whole portfolio at once but dollar - cost averaging over time, or not leverage all your assets - so you can survive downs and keep playing. Many small bets also let you diversify across uncorrelated areas: some might fail, others succeed, the overall result is steadier. Remember, EV multiplies scale: 10 bets with EV $5k each yields $50k EV total. If you rolled them into one big bet, EV is same $50k but risk is lumped - could result in huge win or total loss, more extreme outcomes. Frequent smaller bets align with compound growth as well. If each bet slightly grows your bankroll, subsequent bets become a bit larger in absolute terms, compounding your gains. In career terms, taking many small opportunities to stretch your skills (each with a chance to succeed) is often better than betting it all on one big promotion leap. Each success builds your experience (and even failures likely aren’t fatal and teach lessons). Summing up: think of life/work as a series of bets; structure them so you have many at - bats with positive EV, rather than one do - or - die hero move.

Scale bets with repeatable edges. The concept of an “edge” means having better odds or outcomes than a fair baseline. If you identify an edge - say your marketing strategy yields consistently higher conversion than competitors, or you have insight that a market is undervaluing something - then you have a repeatable positive - EV situation. In these cases, you should bet more aggressively (though still within risk limits) because each bet is in your favor. A gambler analogy: if you somehow know a coin will land heads 60% of the time (edge), you want to bet on heads often and maybe in larger amounts than you would on a fair coin. However, how large? If you bet too high, a few tails in a row (which happens) could wipe you before long - term edge pays off. This is where something like Kelly Criterion guides: bet proportionally to edge. Roughly, Kelly says bet = edge/odds. For a 60% coin (edge 0.2 over 50 - 50 fair) with 1:1 payout odds, it suggests betting 20% of your bankroll each flip to maximize growth without ruin. Betting more than that can actually lower long - term growth or increase risk of hitting zero. You don’t need the formula per se, but the intuition: the greater your advantage, the larger a fraction of resources you can safely allocate, but still keep it within a fraction, not the whole. For example, if your product launches succeed 80% of the time (remarkable edge), you might reinvest a large portion of profits into new launches confidently, but not everything, in case a cluster of failures occur. Conversely, when edges are small, bet small. If you think something is just slightly in your favor (like a 52% win chance in a game), bet tiny amounts because luck will dominate in short run and a bad streak could ruin you. When you have no edge (basically gambling for fun, or a zero - EV speculative project), either don’t play or only use disposable resources for entertainment (like going to casino with a fixed small budget expecting to lose it - view it as paying for fun, not investment). The discipline of bet scaling ensures that when you stumble upon a near - certain or very lucrative opportunity, you capitalize strongly, and when an opportunity is marginal or very uncertain, you tread lightly. It’s a systematic way to allocate efforts or money: push resources toward high - EV, high - confidence bets and pull back from low - EV or wildly uncertain ones.

Don’t forget intangible payoffs and costs. Not everything is captured in immediate dollars. When computing expected value, include qualitative benefits or costs by giving them some weight or at least noting them. For example, a project might have an expected monetary value of - $5k (a slight loss in expected profit terms), but perhaps it yields learning or opens doors for future business. That option value or learning value might be worth more than $5k to you long - term. You could argument “doing this project has an 80% chance to teach us new tech skills which will help in future projects.” It’s hard to number - crunch that, but you can at least acknowledge “non - monetary payoff: potential skill development - likely high,” and factor it into your decision. Possibly you treat that as, say, an equivalent of $X added to payoff. Conversely, some bets might carry hidden costs: stress, reputation risk, opportunity cost of time. A job that pays more but likely demands 60 - hour weeks might have an EV in money, but perhaps a negative factor in health/happiness currency. You can still frame it like: 70% chance I succeed in new high - power job (with extra $20k/year), 30% chance I burn out and leave in a year (maybe hurting career momentum). EV might look good financially, but if burn - out is a hit to your well - being, you weigh that in. The idea is to be holistic in evaluating outcomes. Write out not just “money if win/lose” but other consequences. Give them thought in the decision. In business, explicitly list strategic options unlocked if successful (optional follow - on projects, market credibility, etc.). Those are secondary payoffs that could tilt a borderline EV into positive. Sometimes we pursue break - even or slight loss ventures purely for the intangible upside, which is fine if done knowingly. Just as a good gambler considers not just money but enjoyment or comps they get, you consider learning or networking as part of the bet’s reward. However, keep it specific: “this project could land us our next client” is an option value. Try to estimate how likely and how valuable that next client is, and incorporate maybe a weighted part of that into EV (“20% chance leads to $100k follow - on = +$20k EV intangible”). This prevents fooling yourself that “well it has intangible benefits” for every losing proposition. Force yourself to articulate what intangible, how valuable, how likely. If you can’t, maybe it’s not truly a benefit. This approach keeps the decision grounded.

Compare and allocate across bets with a simple table. When making multiple decisions, it helps to lay out a table: list your current “bets” or initiatives, their cost, their probability of various outcomes, and EV. For example:

| Option | Cost (if any) | Success payoff | Success prob | Failure payoff | EV (net) |

| - - - - - - - - - - - - - - - | - - - - - - - - - - - - - - | - - - - - - - - - - - - - - - - | - - - - - - - - - - - - - - | - - - - - - - - - - - - - - - - | - - - - - - - - - - - - |

| Project A | $20k | +$50k rev | 50% | $0 | $5k |

| Campaign B | $10k | +$30k rev | 40% | $0 | $2k |

| Status Quo (do nothing new) | $0 | $0 | 100% | - | $0 |

In this hypothetical, A and B both have positive EV (5k and 2k respectively). If you have budget to do both, combined EV = 7k. If budget only for one, you’d pick A (higher EV). If A required more funds than you could safely risk, maybe you do B or break A into smaller pilot (like half spend yields half payoff with similar probability perhaps). Also, consider correlation: if A and B outcomes are independent, doing both diversifies risk (maybe A fails but B succeeds, etc.). If they’re correlated (e.g., both bets depend on same market conditions), that combined risk is trickier. The table makes choices explicit. Add a column for downside risk (worst - case impact beyond cost, if any). Also note intangible value if notable, as a footnote or extra column (“learning = high” etc.). Then decide resource allocation: maybe allocate X resources to A, Y to B, ensuring total downside within budget. Perhaps A uses most budget and a bit goes to B as a flier because B’s EV still positive and gives another shot at gain. If you spot any with negative EV (like if Campaign B was actually - $2k EV), you should likely drop those, unless there’s some hidden strategic reason. Sometimes we have resources that must be used or we have to choose among only “less negative” options (like mitigating losses), but generally avoid negative EV bets - they’re expectations to lose value. In personal life, if you find you’re spending effort on something that on balance hurts more than helps (negative EV to your happiness or goals), consider stopping that “bet.” The table approach, even qualitatively, helps clarity. It’s essentially what savvy investors do with portfolios and what product managers do with feature prioritization (weighing benefit probabilities vs costs). Once you choose your bets and sizes, monitor outcomes. If reality diverges from assumption (maybe Project A hitting roadblocks making success less likely), update P(success) downward and re - evaluate - perhaps shift some remaining funds to B or a new option. Treat it dynamically.

Practice shifting effort to the best - EV opportunities. One quick exercise from the outline: list your top three current “bets” (projects, initiatives, personal efforts). Estimate roughly their EV (doesn’t need to be dollars: could be career advancement points, happiness points, etc.). Recognize which is highest EV and which is lowest. Now imagine taking 10% of effort or resources from the lowest and giving it to the highest - would that likely improve overall outcomes? Often yes: you cut your losses on something not promising and back the winner a bit more. Try implementing that: literally allocate a bit more time or budget to the high - EV one this week, and trim the low - EV one (or speed it to conclusion/decision point sooner). This simulates portfolio optimization. Over time, repeatedly trimming the weakest and reinforcing the strongest bets can significantly increase your aggregate success. Be cautious not to chase only one bet though - diversification matters. But if one is clearly floundering with negative EV now (maybe initial optimism not panning out, now likely just drain), don’t be afraid to cut it (the concept of kill criteria from earlier part, applied: if EV goes negative or below alternatives, time to stop that bet). People often have trouble letting go due to sunk cost fallacy; EV thinking helps override that by focusing on future prospects only. It becomes easier to say, “Project C has maybe 20% chance to salvage something and likely to just cost more… its EV is negative now, better to reallocate team to Project A where each hour is far more likely to yield results.” By practicing this thinking, you start instinctively evaluating actions as bets with potential payoffs and using your time/money where it yields the greatest expected good.

By calculating expected value and sizing your bets wisely, you start to play the long game like a savvy investor or poker player, rather than a gambler on tilt. You take opportunities where the odds and stakes make sense and avoid or limit those that don’t, thereby improving your “average outcome” over time. Combined with base rates and clear probabilities, you’re now making decisions with a cool head and a sharp pencil, not just a hopeful heart. Of course, not all decisions are purely numbers; some involve strategy and reversibility. In the next chapter, we’ll add another dimension to decision - making: knowing when a decision is reversible and how that should speed up or slow down your approach. Understanding one - way vs two - way door decisions will further refine how you apply your decision science toolkit in the real world, balancing thoroughness with agility.