A Shortcut to Decision Making: What's there to Win and What's there to Lose

We struggle with making important decisions in this uncertain world. We have to constantly make decisions without really knowing for sure whether they will turn up ok or not. On top of it, we have to appear decisive and act confident while living with this uncertainty. Leaders who appear to waiver are rarely looked up by the subordinates.

Maybe you have tried to take help of "crystal ball" peddling consultants who appear even more confident than you do, and they might have told you exactly what to do. If you are experienced enough, you would know that even they do not know what is going to happen with certainty. There is a small chance that you might even have tried to do some sort of 'decision analysis', a highly nerdy field involving something called 'decision trees' but have found the approach very complicated and sometime unintuitive.

I do not have a crystal ball or magic potion that will let you make the correct decisions every time. That is simply impossible. But what if I told you that there is a simple approach that can lead you to make better decisions every day. Better still, most of the time it will not even require more than a few minutes of conscious efforts.

Decision Tree

While I am trying to make this as less nerdy as possible, unfortunately it is impossible to talk about decision making without a decision tree. However, I will keep the description as simple as possible. A simple decision tree looks like this:

A simple decision tree attempts to make a choice between two options, option 1 & option 2. Let's say if you choose option 1, you will win with a probability Pw and win a certain amount or lose with a probability Pe and lose a certain amount. The basic idea of a decision tree is to find out these probabilities Pw and Pe and find out something called 'expected value' which is given by (Pw* amount won - Pe* amount lost). You chose the option whichever has the higher expected value.

But there is a big issue with this decision tree. In real world, these probabilities are rarely known. The real world is a complicated system with a large number of variables. Further new variable keep appearing every now and then. In my experience, trying to calculate these probabilities for real world problems is not much different from crystal gazing. As far as trying to use 'expected value' for real world problems goes, I would say the scoreboard is Common Sense : 1, Nerd : 0. [even though there are some real world situations like odds of winning a lottery that can be accurately calculated]. If you disagree with this statement, then you have an entirely different world view compared to me and then the rest of the blog is not for you.

What's there to win and what's there to lose

Despite the inherent issues with calculating probabilities, decision tree is not a useless concept. In fact, if used correctly it can be one of the most useful practical tools to know.

While we are extremely bad at calculating real world probabilities. we are much better at calculating something else. With just a little bit of thinking, we can often figure out what is at stakes. If we win then how much can we win and if we lose than how much can we lose. I have found it to be much more constructive than trying to calculate exact probabilities. And it is immensely useful in cutting the clutter of varied information and make a confident decision. What's best is that it can be applied from the simplest to the most complicated decisions. As an example, here are a few scenarios where this simple tool can be used:

Scenario 1: Should you speed?

You are not alone in this world if you have been tempted to speed on an unknown road. It maybe for an urgent meeting or just for thrill, but the temptation to go a little bit easy on the accelerator (gas pedal) is always there. How do you fight the temptation? Just think what is there to win and what is there to lose. If you win (i.e. don't get caught), you might 'win' a few minutes. Hardly very precious. But if you lose you can lose a large amount of money in traffic fines or even meet a traffic accident. Clearly, despite the uncertainty whether you will get caught, it makes no sense to speed. There is a lot to lose and very little to gain. Your odds of getting caught or getting into an accident have to be probably less than 1 in 1000 for you to give in to the temptation of speeding. Similarly if you see a cloudy sky, it is better to carry an umbrella without going in to probability of rain.

Other interesting arguments can be made about having life insurance, health insurance etc. If the insurer is doing its job right, the "expected value" of such expenses is likely to be negative for you. Still it makes sense to take these policies since your family can lose a heft if the outcome is negative. There is so much at stake that it makes sense to lose on an average.

Scenario 2: How much to prepare for the "next wave"

Imagine you are a government official who had invested a large amount of time in preparing for the previous covid wave in the country. However it was much milder compared to the forecasts because of unknown reasons. Furthermore, now vaccines are available and government has been very fast in vaccinating the most vulnerable age group (45 and above). You want to give a much larger order for vaccine. However, 10 years ago there was an H1N1 outbreak where government spend millions of dollars for vaccine. The investment went to waste as the outbreak was contained before the vaccines can be made available. There are priorities related to stimulating the economy. Given this situation, should you spend big in preparing for the next wave?

Again you can use the simple framework. What is there to gain and what is there to lose? There is a chance that the next covid wave may be even milder and number of lives lost may be even lesser compared to previous wave given that the most vulnerable segment has been vaccinated. There is also a chance that the next wave is more lethal and may spread faster. Your options are to invest aggressively in hospital infrastructure, large vaccine orders or not to invest. If you do invest, you will 'win' if there is a more virulent outbreak because you are ready for it. But if the outbreak is milder then you end up spending billions of dollars and manpower that could have gone in stimulating the economy. The investment will then appear to be a "wasteful" expenditure.

On the other hand if you do not invest and there is a mild wave, you save those billions of dollars. But if there is bigger wave, numerous lives will be at stake and economy will lose tens of billions of dollars. Clearly there is much to lose in this case. Despite all the uncertainty and a good chance that you investments will go to 'waste' it makes sense to invest aggressively in preparing for the next wave,

Similar arguments can be made about decisions like should you have strong military if the country has no direct threat of war and a good support of global superpowers?

[Disclaimer: Though the scenario may appear similar to what happened in India, please treat this as a hypothetical scenario. I do not have any reliable information on what preparations were made or not made in India and what were the facts and constraints in front of the decision makers.]

Scenario 3: Should your company invest in a new technology that is not proven

Every now and then many companies come across the question of investing in a still unproven technology/product segment. Most often the scenario is that if the technology/product segment takes up as expected, it will transform the industry but there is always a chance that it will not take up as expected. Of course the most sensible strategy is a "fast follower" strategy where you wait for the course of technology adoption to become clearer and costs to come down and then invest in a big way. But what if such an option is not available. This can happen if the competitor can gain enough "network effect" or intellectual property to make a fast follower strategy less feasible. The entire problem is complicated even more when the new technology/product is going to cannibalize the existing cash cows. Which company in the right mind will want to invest in something that will damage the existing reliable revenue streams?

Such very complicated decisions can be immediately simplified by thinking about what is at stakes? If the new technology/product segment takes up as expected, will your existing product segment survive. Secondly, do you have enough funding available to invest in new product line. And thirdly, if the technology succeeds how much return can you get? If the new technology/product line can potentially replace your current product line completely and there is no option to wait, then clearly the survival of the company is at stake. It makes sense to invest in the new technology/product line even if it cannibalizes the existing product lines and even when the outcome of the technology/product line is uncertain. If you think the potential market for the new technology/product line is 10x the current one, then of course the job becomes much easier. Kodak could have used this decision making approach when they invented digital camera.

Probabilities are still very important

My arguments above may seem to convey that probabilities do not matter. Only important thing is how much there is to win and how much there is to lose. However, that will be an incorrect conclusion. After all you shouldn't be investing in lottery tickets just because you can win a lot but lose only a little.

The "what-is-there-to-win-and-what-is-there-to-lose" approach I have advocated is a short cut which makes job of decision making easier. It works best either when the probabilities of two outcomes have either (i) similar order of magnitude, let's say 0.3 and 0.7 (like in case of estimation of probability of next covid wave being much more intense than the previous one) or (ii) when the possible loss is unacceptable (like in life insurance). When these conditions are not met, it is worthwhile to do at least a very rough cut expected value calculations. (Warning: it gets a bit more complicated from here).

When I said it is difficult to assess probabilities in real world scenario, it applies when probabilities of outcomes are close to each other. If you estimate probability of one real world outcome is 0.6 and of the other is 0.4, then you are most likely wrong. In my opinion, such an accurate estimation cannot be done for most real life problems. However when one event it far less likely compared to other, it is possible to estimate that with a little bit of analytical skills, study of past events/data and some unbiased thinking. Thus, it is possible to distinguish a 90% probable outcome compared to a 10% probable outcome. In such cases expected value should be calculated and should be used in decision making. For example, let us go back to the new technology question. This time the new technology is not related to the current business. But it seems to be a moonshot, where your company can make 100x returns if successful. It is worthwhile to get even a very rough cut estimate of probability of success. If it is estimated to be in range of 1% then it doesn't make sense to invest in it. But if you give even a 10% chances of success, then it is a very lucrative opportunity. Jeff Bezos used a similar logic to invest in cloud technology. In this case also you will most likely to be wrong in estimating the probability. But if the expected value is a large value compared to investments required, it is easy to overlook the inaccuracies in estimation of probabilities. More importantly, if the expected value is not a large value, it is difficult to tell which option is better. In my investment job, I call these situations 'do nothing' situations. And if doing nothing is not an option, probably a coin toss will work as well as decision analysis!!

If you found this interesting and want to dive deeper into decision making, you can read about it more here:

https://www.dhirubhai.net/pulse/decision-making-under-uncertainty-abhishek-chauhan/