How Expected Value Can Help You Make Good Decisions (Part 2)
Learning to take calculated risks when the outcome is uncertain is a matter of skill and intelligent calculation. Get-It-Done Guy explains how to make any uncertain decision.
Page 1 of 2
In Part 1 of this series, we looked at using expected value as a tool for decision-making under uncertainty. It’s a way to make sure you’re taking enough risks, and the right risks. Instead of comparing best case and worst case scenarios when making a decision, look at the expected value, which is the probability of an event times the payoff from that event.
Check out Part 1 of this series for a detailed explanation of what expected value is and how to derive it for any decision you're making.
Visit iFixit.com/GetItDone for all the free repair guides you’ll ever need. And when you find the perfect part or tool, use coupon code GETITDONE at checkout to get $10 off your order of $50 or more.
Do Research to Estimate Probabilities
Calculating expected values requires estimating probabilities and figuring out what the payoff is for each branch of the decision.
For example, Melvin saw what seemed to be an amazing deal—a 2-year-old Toyota Corolla for $5,000 on Craigslist, “Priced to move by a desperate seller.” He planned to buy it and sell it at a high price in a more leisurely time frame. The seller wouldn’t let him see the car or get a mechanic’s report prior to purchase. And he must wire the payment directly to a bank account in Nigeria.
Melvin figures there’s a 5% chance the ad is a fraud and worth $0, and a 95% chance it’s worth $125,000. He calculated the expected value at 5% times $0 plus 95% times $125,000 or $118,750, minus the $5,000 purchase price, or $113,750. Who could pass up that expected value?
You and I should pass that up. Melvin’s gut feel is wrong, both for the probabilities and the car value. Even with expected value, making better decisions requires research.
Estimate Probabilities Intelligently
To estimate probabilities, you can often find publicly available statistics. If you’re estimating the chances that a new company will survive its first 5 years, for example, you can use the statistic that roughly 55% of all businesses fail by year 5, which means that there’s a 45% chance of 5-year survival.
You can look also for more specific data. If you’re considering starting a real estate business, 5-year survival rates are higher, at 58%. You could also spend time researching your geographic area to get statistics that are more representative of the business you’re starting.
For Craigslist, Googling “fraud on Craigslist” doesn’t give stats, but it does give 6 million hits. A quick trip to the Craigslist Scams page shows that the ad Melvin got so excited about fits several warning signs for a scam. We won’t reject it out of hand (maybe it really is an overlooked opportunity) but we’ll adjust our expected value percentages, and say there’s a 97% chance of the car being a fraud, and 3% chance of it not.
Estimate Values Intelligently
Research also helps figure out the values of the different outcomes. If you’re deciding whether or not to go to college, you can figure out a dollar value of a college education by Googling “How higher education affects lifetime salary.” Use those numbers to put a specific dollar value on the payoff of a college education.
Melvin says that the non-fraud outcome of the Craigslist purchase is worth $125,000. A quick visit to the Kelly Blue Book web site shows that a 2-year old Corolla sells between $11,000 and $14,000, not $125,000. Plus, Kelly Blue Book gives 4 prices depending on the car’s condition, and the percentages of cars that they see in each condition. That’s enough information to make each car condition a separate branch of the expected value calculation.