You will be able to calculate the expected value of a choice and use it to compare options.
Pick one choice in front of you right now where you pay money or time first and might gain something later. Write down three numbers: what it costs, what you gain if it works, and your honest estimate of the chance it works, as a percentage. With those three numbers you can check whether the choice is worth making on average, and what you lose if it goes badly.
Lesson 7.1, "Decisions are bets", showed you how to state how sure you are of something as a probability. The method below uses those probabilities.
Expected value is the average result a choice would give you if you could make it many times over. For each possible outcome, multiply its value by its probability, then add those figures together. If the choice has a cost, subtract the cost from that total. The result is the expected value of the choice.
A charity raffle sells 1,000 tickets at S$2 each (an example figure), and there is one prize worth S$1,000. With one ticket your chance of winning is 1 in 1,000, so your expected winnings are S$1 per ticket. Take away the S$2 price and the expected value is minus S$1. People don't mind this at a charity raffle, because the S$1 they lose on average is effectively a donation.
Commercial lotteries and casino games are deliberately set up so that players' expected value is negative. The operator pays out less in prizes than it collects from tickets and bets, and the difference covers its costs, makes it a profit or funds its causes. Any one player can win, but the average player cannot. The more you play, the closer your own results get to that average.
Siti works in HR (you met her in lesson 7.1). She is thinking about a professional HR certification. It costs S$3,000 after any subsidy and would take most of her Saturdays for six months. Her manager has hinted that people with the certification tend to move up a grade with a raise. But a colleague of hers did the same course and nothing changed for her. Siti keeps telling herself "it might pay off", and that thought gives her nothing to decide with.
She keeps it to two outcomes: a raise of S$300 a month, or nothing. She puts the chance of the raise at about 50 percent.
If she counts only the first year, the raise comes to S$300 x 12 = S$3,600. Half of that is S$1,800. After the S$3,000 cost, the expected value is minus S$1,200, so on a one-year view the certification loses money.
But a raise usually lasts longer than one year. The time horizon is the period you count the payoff over, and it matters as much as the probability. Base it on how long you would realistically stay in a role where the payoff applies. Fix it before you do any sums, so you aren't tempted to choose the horizon that gives you the answer you wanted. Siti chooses three years.
Over 36 months the raise totals S$300 x 36 = S$10,800. Half of that is S$5,400. After the S$3,000 cost, the expected value is S$2,400.
Expected value puts options on one scale, so Siti can compare the certification with an alternative. A short course (also invented) costs S$800 and has about a 30 percent chance of leading to a S$200 a month raise. Over three years that raise is worth S$7,200. Thirty percent of that is S$2,160, and after the S$800 cost the expected value is S$1,360. On the three-year scale, the certification's S$2,400 beats the short course's S$1,360.
Nobody gave Siti the 50 percent. It is her own best estimate, like most probability inputs in real decisions. Some people won't calculate expected value at all because the inputs are uncertain, and that is a mistake. A rough estimate you write down openly is better than a vague feeling, because you can look at it, question it and change it.
A sensitivity test shows how much the estimate matters: run the numbers again at a higher and a lower probability. At 40 percent, the certification's three-year expected value is S$1,320. At 60 percent, it is S$3,480. Siti also works out the break-even probability, which is the chance below which the expected value drops under zero. For her certification over three years, it is about 28 percent. If the decision comes out the same at every estimate you think is plausible, you don't need to get the probability exactly right.
You can make the estimate better with the outside view from lesson 5.3, "Vivid stories and the planning fallacy". Use base rates, meaning how often the outcome happened across past comparable cases, rather than single stories. For Siti, the base rate is how many past course-takers in her company moved up. Her colleague's experience is one story, and lesson 1.2, "Not all evidence weighs the same", warned against giving one person's experience much weight. HR or her manager may know how past course-takers did.
An average hides how wide the range of outcomes is. You make the choice once and you get one outcome. So after you compare expected values, look at each option's worst case separately. If the certification goes badly, Siti loses S$3,000 and six months of Saturdays. If the short course goes badly, she loses S$800.
If you could recover from every worst case, the higher expected value is a fair guide. Siti could recover from either loss, so she goes with expected value. If one option has a worst case you could not recover from, expected value alone is not enough. Lesson 7.3, "When the average is not enough", covers the different thinking you need then.
Siti decides to ask HR how many past course-takers moved up within a year. If the answer is anywhere near half, she will enrol.
The method, in order:
List the possible outcomes, keeping it simple (two is fine). Estimate a value and a probability for each one. Fix an honest time horizon before you calculate. Multiply each value by its probability, add the results, and subtract the cost. Repeat for each alternative and compare the expected values. Recalculate at a higher and a lower probability, and find the break-even probability. Improve your estimates with base rates rather than single stories. Look at each option's worst case separately. If every worst case is recoverable, follow the higher expected value. If one is not, use the approach in lesson 7.3.
You probably have a choice like hers in front of you, whether it is a course, a certification, a side project or a change of role. Keep it in mind for the activity below, and be honest with yourself about the horizon.
With the three numbers from the start already on paper, the sums take about ten minutes. Doing them now, while the choice is still open and you haven't paid anything, means the result can still decide whether you go ahead.
Pick a real choice, such as a course that might lead to a pay rise, fix your time horizon, and calculate its expected value from your own estimates of cost, probability and gain. Then do the same for one alternative and compare the two.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).