Communicating Complex Ideas

Posted by
“Prediction markets are appealing and useful because they can make predictions about future events, but getting people to use them effectively requires a bit of work.”For those who don’t follow the latest popular economics books, the relatively recent publication of The Wisdom of the Crowds by James Surowiecki has generated a burst of activity and interest in the prediction market arena. Prediction markets–markets which explicitly take advantage of the “wisdom of the crowds,” in that the price of a contract can be interpreted as a prediction for a future event–seem very simple and straightforward. People can buy and sell a contract based on a event, much like they would buy and sell a stock, and the resulting price says something about the probability for that event occurring.

Take, for example, a popular prediction market contract: Hillary Clinton will be the next U.S. president. If I think Hillary will be President, I can buy the contract. If I don’t think Hillary will succeed in her bid, I can sell the contract. Straightforward enough, but things can quickly get complicated.

What if you believe that Hillary has a pretty good shot at winning, say a 70% chance, but the contract is priced as if there is a 90% chance that Hillary will be the next president. Would you buy it then, or sell it? Suddenly you need to think about the strength of your belief. And how would you sell it if you don’t own the contract already?

Now imagine the contract is conditional, so that it is priced based on the outcome of whom the Republicans nominate, say John McCain. Would that increase her chances, or decrease them? And what if there is a market, say the Democratic Party Nominee Market, where Hillary is just one of several contracts trading? And furthermore, what if you were only one of 10 traders in the market, so that every time you bought or sold a contract you moved the market by over 10%?

Now add to this the fact that most people don’t know how to trade. For most people investing involves a 401K and maybe an IRA. Shorting stock is an alien concept to most. While day traders rabidly tapping out buy and sell orders into brokerage sites gripped the imagination for a while in 2000, these types of traders are a very small percentage of the U.S. population. Most people, it turns out, aren’t very good at buying and selling stocks at all.

And finally, why would one even care about buying contracts on Hillary’s presidential aspirations?

“These challenges become even harder when one thinks about implementing a prediction market within the corporate environment. Here, the cost of playing can be high (trading instead of working), and the contracts can be relatively arcane–predictions of future sales, for example. Selecting a large enough group of the right traders appears to be a difficult task.”Making prediction markets work
Prediction markets are appealing and useful because they can make predictions about future events, but getting people to use them effectively requires a bit of work. They need to make themselves relevant, attract the right types of traders, and teach people who are new to trading how to trade.

Several websites have taken on this daunting task with mixed success. These public prediction markets offer markets and contracts that cover a range of issues and events. They each have different functionality, and have designed their markets in different ways, but all are trying to solve the same problem: how to engage users to participate in these markets, and produce meaningful predictions.

What follows are some basic and general rules, gleaned from audits of these sites (with selected examples to help explain), about how to improve the communication and simplification of complex ideas about trading, and how to engage users, particularly when developing a prediction market. Some of these rules will be familiar and appropriate to all websites (but with a different emphasis for prediction markets), and some will be unique to prediction markets.

The following is a brief summary of these rules:

  • Liquidity is important to a market, so make people want to play.
  • Brief examples on the homepage will sometimes be the only learning users will have the patience for. Use these examples widely.
  • The stock market is a good metaphor for prediction markets, but don’t take it too far.
  • Simplify. Allow users to focus on the things that really matter.
  • Use tools that indicate and reveal what is happening. Trading is a nerve racking experience and users want to know what is happening.
  • Use contextual help. More so than elsewhere, contextual help guides users through complicated processes.

Make people want to play
Call it the usefulness of a website, or the hook which gets users to stay, the most essential part for any prediction markets website is to have markets and contracts that people are interested in participating in. As one recent book (1) put it, “As the wonkishness of the contract rises, however, volume and liquidity fall rapidly.” John Delaney, the founder/CEO of Tradesports.com, puts it more directly: “there is very little point in listing a contract that no one is interested in.”

The Hollywood Stock Exchange (HSX) is a master at getting people to play. With well over 50,000 users using play money to buy and sell contracts and derivatives based on the revenue of movies and actors, the site is clearly popular. It is also predictive. As academics seek to understand how well these markets perform they have measured the predictive power of HSX against Hollywood experts in how well they can predict who will win Oscars (2). HSX, it turns out, is superior.

As an indication of how passionate some users are about HSX, top portfolios on HSX can sometimes be found on eBay where people are willing to pay real money for fake money accounts.

Protrade is another public prediction market and has extended the concept from movies to baseball, football and basketball. On this site users can “invest” in players who they think are under-valued, and “sell” players who they think are over-valued.

Sports, movies, and actors make good subjects for prediction markets, not only because they are popular, but because accessing relevant information and forming an informed opinion is easy. Sports statistics are readily found on any number of sites, and talking about movies and actors is virtually a global past-time. Who doesn’t know about Tom Cruise? Who doesn’t have an opinion on him? Who doesn’t think they know better?

Making prediction markets work in the corporate world partly depends on overcoming this hurdle. Unless you work for the Boston Red Sox or at Universal Studios there probably aren’t that many compelling issues that workers can get passionate about. Motivating people in the workplace to take part in prediction markets takes imagination.

Use your examples wisely
As we all know users don’t tend to read much online unless they have to, or they have found the specific content they’re interested in. First impressions are also very important, and communicating the content of the prediction market, and how it works at a high level, are usually accomplished on the home page. Therefore most prediction markets have some sort of example or explanatory text on the home page. Unless the users are diligent enough to read the help text, this is often the only chance a site will get in explaining itself.

Different sites choose different tactics, but perhaps the best and simplest approach is that chosen by the Yahoo Tech Buzz Game.


Yahoo’s Tech Buzz Game homepage example

In four simple bullets it explains how the game works, what you will be trading, and importantly, how prices are moved.

On Trendio.com you buy and sell words on the expectation of their frequency of appearance in English speaking newspapers. For example, buy the word Google just before a big announcement by the company and the contract in that word rises once the announcement hits the newsstand. Trendio tries to drive their point home graphically.


Trendio.com homepage example

But they end up only confusing matters more. First, the example refers to ideas and not words. Second, without labeling and a consistent y axis it’s difficult to know what you’re comparing and how to even understand the charts.

When people are learning, examples are their guideposts. Use them wisely.

The stock market is a good metaphor, but don’t get stuck in it
Prediction markets are almost always compared to stock markets, and it makes sense to do so. Despite the fact that most people aren’t traders, most everyone has a passing familiarity with it. And almost everyone knows the most basic trading strategy of all time: Buy low, sell high!

But this model can also work against you. In the stock market there are many ways to enter an order. You can enter a market order, a limit order, an all or none order, and so on. While heavy traders might be happy to see these features imbedded into prediction markets, most people will be confused by them. Despite this some sites make the attempt.

Smarkets, though, a prediction market game that allows you to buy or sell contracts based on Amazon products and what their sales rank is, has simplified the trading interface to a point where it’s almost ridiculously simple.


Smarkets Trading Widget, Step 1 (contract information is to the left on the site)


Smarkets Trading Widget, Step 2 (contract information is to the left on the site)

The Smarkets trading widget dispenses with many of the complicated aspects of trading stock. Of course it can afford to as a play money game, but why make users deal with bid ask spreads and limit orders when they don’t need to. The majority of users will be just happy with a trading widget like the one above. If there are users for your site who want such functionality, then an advanced trading screen can be created for them, such as on Hollywood Stock Exhange (HSX).

Experts may quibble that a bid ask spread is important to price discovery, and so it is, but if you can make do without exposing it to the user, why not?


HSX Trading Widget

The HSX trading widget is even simpler and allows for even greater flexibility. Mouse over the movie you want to buy and the widget pre-populates. Max quantity allows you to purchase the maximum allowed amount. If you already own shares in this movie, to sell all of them just sell max, and you never have to worry about remembering how many shares you already own.

Both sites also take another idea from stock trading that people may be familiar with and twist it in a way that’s easy to understand and use. Shorting stock is a difficult concept to understand, and in the stock market it requires that you borrow shares from someone else (your broker figures out whom), sell them on the market, and then repurchase them later, hopefully at a cheaper amount. You pocket (or pay) the difference.

There is some complicated accounting involved and the trader who does this must have a margin account. In other words, you must always have a certain amount of money available in case the market turns against you.

Prediction markets handle the desire for traders to bet against an event or movie by allowing traders to buy shares in the “opposite outcome.” But exposing two contracts for the same event (or movie) to neophyte traders can, perhaps, be a bit confusing (both could potentially be bought and sold), so Smarkets and HSX disguise buying an opposite outcome contract as “selling short.”

In this way they’ve jettisoned the complicated accounting and they’ve removed the “opposite outcome” and, as a result, simplified the user experience. Of course, traders aren’t actually “selling short” but it makes no difference to them. They’ve accomplished what they wanted to accomplished – bet against the movie, or event.

Simplify, simplify, simplify
Yahoo’s Tech Buzz Game does something very interesting and quite logical. They’ve dispensed with allowing the user to enter the amount of shares they want to purchase. Why make the user do the calculation at all? The user only cares about how much money they’re going to invest in a certain contract not in how many shares they’re going to buy.


Yahoo Tech Buzz Game Contract page

By only allowing users to enter an amount of money they’ve actually saved the user time and an extra step. However, they have not carried this innovation over to the selling side of things. When selling, you begin with how many shares you’d like to sell, rather than how much money you’d like to cash out (which would be nice, wouldn’t it?). This is entirely appropriate, since a user may only want to sell half their shares instead of all of them.

Use tools that support the trading process and reveal what’s happening
For those unaccustomed to it, trading can be a nerve racking experience, even when trading with play money. While you bank account may be safe in these games, your ego is not. Nothing is quite so dispiriting as seeing your “assets” dwindle because you were wrong. Therefore, tools that help a user feel “safe” when trading, by explaining what is happening, and exactly what everything means, have value.

Inkling, for example, draws a direct connection between the expected probability of the event and the price. It then sets up three possible trades for you based on how different your expectations for the event happening are from the probability implied in the price. It also allows user to enter their own trades, if they so wish, but neophyte traders can allow themselves to be guided through the process of aligning their expectations for the event occurring and the price and number of contracts they purchase.


Inkling Trade Assistant, Step 1


Inkling Trade Assistant, Step 2

This guidance through the trading process is akin to training wheels on a bicycle. Experienced traders will have no need for this level of hand holding but beginning traders will appreciate it.

HedgeStreet, a real money peer to peer trading site includes a neat tool that helps the trader foresee any potential gains and losses.


HedgeStreet Trading Tool

In this case (Euro-US$ futures contracts) the user can enter in how many shares they would like to purchase. The price is automatically entered for them and their potential gain (upside) and loss (downside) is automatically calculated. It immediately tells you the amount of risk and potential reward you are accepting before making the purchase.

Use contextual help … a lot
And finally, as my father used to say, when all else fails read the instructions. Most prediction market websites have relatively robust Help sections, since there is a lot to learn and get help on. Better, though, are the sites that present their help contextually, and in this way support the user while they are in the middle of a task rather than forcing them to leave the page and go to the Help section.

NewsFutures has the humorously (but appropriately) named “I Need Help!” link which opens up a pop-up window with detailed instructions on how to read what is probably (certainly to most new traders), a difficult to understand bid and ask price list.


NewsFutures “I Need Help” link


NewsFutures “I Need Help” pop-up

While the help diagram is busy, the idea is right.

Smarkets makes it even simpler. With the link placed right under the order type drop down, the user can quickly access an explanation of what they should be doing.


Smarkets contextual help

Contextual help, while not the most sexy recommendation, is just one of the little things that public prediction markets should get right, more so than most sites, in order to create a useable and enjoyable experience for users.

Summary
As prediction markets push beyond being just games and interesting experiments, particularly as they expand into the corporate world, simplifying and communicating complex ideas about trading, contracts, and markets will become crucial. These examples show just some of the ways that these complex tasks and ideas can be communicated.

The most successful sites are those that understand the experience range of their users. Some are veteran traders who know what to do, while others can’t tell a bid from an ask price. Accommodating the novice traders is crucial to the success of these markets, as well as moving them along as they gain experience.

Prediction markets hold great promise. Research has shown a well designed prediction market can predict the future better than experts most (if not all) of the time. This promise encourages us to apply these markets to new areas of knowledge and in new environments. As more people use prediction markets, though, the more usable prediction markets must be.


Notes

7 comments

  1. Fascinating, although I would have preferred a better choice of title for this article because at first I thought it was going to use prediction markets as a example of how to simplify complex ideas in general. Instead it’s about how to simplify complex ideas in prediction markets. Never mind, fun read, who cares.

  2. Interesting article. The HedgeStreet Trading Tool screenshot seems to be pointing to the wrong image.

  3. Great article, as the IA at PROTRADE I can definitely attest to the validity of the issues you raised. The issue about following the stock market metaphor is an interesting one and one that sometimes conflicts with the simplify one. Do you use the exact stock terms even though most users will not understand them or do you come up with a more description term but then potentially confuse the users who do understand the stock market based term. I’d love to hear more comments you may have had about PROTRADE you can email me at nhoh (at) protrade.com.

  4. Thanks for your comments everyone. This is a great area of study, and I’m having a lot of fun with it.

  5. Great review of the book and the concept. Even better would be an additional explanation that this is the newest of technology frontiers for the market or design researcher. Using predictive markets can provide incentive driven feedback for products and services – often with more accuracy than surveys and interviews. It can also be very effective at drawing tacit knowledge from an engaged sales force or distribution personnel.

  6. Great article. Would be interesting to capture different personas and their perspectives – e.g. people who come from a trading background, versus people who come from a more mainstream (non-trading) background.

Comments are closed.