Thursday, February 23, 2006

Two Views of Markets

I earlier offered a draft chart diagramming the similarities and differences between three sorts of markets. That garnered some helpful comments, and thus my gratitude. I'd here like to offer two new views comparing prediction markets, futures markets, and securities markets. The first table tracks each market's primary and secondary benefits. The second table shows each market's structural features.



TABLE 1: Markets v. Benefits

\ Benefit | Pool Hedge Express Promote
Market \ | Capital Risks Prices Discovery

Prediction none possible primary secondary
Market tertiary

Futures none primary secondary possible
Market tertiary

Securities primary possible secondary possible
Market tertiary tertiary




TABLE 2: Markets v. Structural Features

\ Feature | Zero-sum Spot Underlying
Market \ | Trading Trades Assets

Prediction yes yes no
Market

Futures yes no usually
Market

Securities no yes yes
Market




Please let me know if, on your browser, those tables appear out-of-line. And, of course, please share your comments. Some of mine follow.

You might note that these tables, in contrast to the prior chart, do not offer the prospect of a Nolan Chart for classifying financial institutions. I have not, in other words, described axes.

I won't bother to here explain the label in each category, as I think you readers will perceive my reasoning.

A quick survey of these charts hints at why prediction markets resemble futures markets more than they do securities markets. Assign numbers to the ordinal rankings in Table 1 like this: primary = 1, secondary = 2, tertiary = 3, none = 4. Now compare the differences between the categories. PM-FM=0, whereas PM-SM=1. A similar exercise with Table 2 gets these results: PM-FM=1.5; PM-SM=2. Prediction markets, by that reckoning, have fewer differences with futures markets than with securities markets.

Where are gambling markets? I haven't dignified them with inclusion because they really do seem very different and, at the least, present some ugly classification problems. If pressed, I suppose I'd say that gambling markets offer a primary benefit of entertainment and a possible secondary benefit of pooling capital (see, e.g., lotteries). That would force me to add a new column to Table 1.

With regard to Table 2, I think I'd say that gambling markets offer zero-sum trading, that they may or may not offer spot trades (compare roulette to a lottery), and that they probably do not deal in underlying assets (though it depends on how you view the jackpot).

(By the way, Chris, I didn't here follow your suggestion, made in the comments to my prior post, to use time limits a distinguishing feature, because "securities," as used in federal law, includes time-limited bonds and notes. But I did adopt your "underlying asset" suggestion.)

7 comments:

Untenured Sociologist (R-CA) said...

I think in discussing gambling markets you should distinguish between games of chance and sports book. Roulette, lotteries, craps, baccarat, and other pure gambling are exactly as you describe them. However sports book is really a special case of a predictions market in that is a prediction about something out there in the universe and the option price usually predicts long-run probabvility. As such sports book follows all of the structural properties and benefits you outlined for prediction markets.

Tom W. Bell said...

Good point, US (R-CA)! So I guess that a sports book market would have a row in revised version of table 1 something like this:

Pooling: none; Hedging: none; Expression: none; Discovery: possible secondary (assuming we want people to develop better models of sports performance); and (adding a new column) Entertainment: Primary.

In Table 2, sports books would look something like this:

Zero-sum: yes; Spot Trades: no; and Underlying assets: no.

Jason Ruspini said...

There's that psychic connection again.. I've been fascinated with taxonomic tables and "schedules" lately, though mine include gambling.

In short, it looks like you're trying to unwrap/backsolve the conclusion that the CFTC should not have authority over prediction markets by making it appear as though that conclusion was handed-down by some natural taxonomy.

My first thought was to eliminate the "Promote Discovery" category and distribute it into "Pool Capital" (renamed "Capitalization/Funding") and "Express Prices" (renamed "Discover Information/Knowledge/Prices"). At that point, the difference of futures/hedging vs. PMs/discovery becomes stark. Likewise with the intention behind the "spot" category of table two.

I'm supportive, and I see what you're trying to do by presenting this "knowledge" — and it's no accident that the "doing" markets (capitalization&hedging) have enjoyed more success than the "knowing" ones up to this point.

Tom W. Bell said...

Jason: Good points, all. Let me think out loud about them.

I see the argument behind going to a more general "capitalization/funding" column. Note, though, that the means of capitalization differ quite markedly between securities markets and prediction markets. The former raises funds by dint luring payments for the traded instruments; the latter by dint of making payments into (the right) claims.

Note that I used "express" prices because I think it important not just that a PM aggregates info, but that it also makes it public. Consider Poolitics.com, by contrast; it does not make prices public, and so does not really do what we want PMs to do.

What do you think about adding a column in Table 2 relating to the potential losses. Bad investments in derivatives can get veeery ugly. In a standard PM, your downside is limited.

Jason Ruspini said...

Regarding the funding question, if rewards are paid to the correct claims, this brings up the "no-trade" idea again.. Why trade against an expert at the forefront of research? What the heck do I know? With equities, you are aligning yourself with the potential innovators. That question of evolutionary "fitness" of markets aside, it seems that the zero-sum column of table two already captures the desired distinction here: funding before ~ positive sum vs. funding/rewarding after ~ zero sum.

PMs are more like options than futures or securities. The popular ones are european binary options. They have time decay and are prone to very sharp moves. While these contracts aren't open-ended, they can still produce huge %-wise wins and losses. Selling enough contracts for pennies that turn-out to be worth $10 would qualify as an ugly blow-up.

But in practice this scenario will be unlikely for two reasons. Intrade and Hedgestreet require full margin to be posted, so selling longshots ties-up a lot of capital. (The fact that they require full margin just underlines how volatile these contracts really are.) More importantly, if you aren't earning interest on your posted margin, there is no reason to sell a 1yr contract for less than r/(1-r) where r is the risk-free rate. For example, if traders can otherwise earn 4.5%/yr but earn nothing on their margin, the "Pigs Fly" contract that expires in one year should only trade down to 4.3! Talk about a longshot bias.. (Jeb Bush last traded near 3 for the 2008 Republican Nomination.. hmmmm)

Chris Hibbert said...

I'm having trouble figuring out what "Spot trades" might mean such that it's a "no" for futures markets. Can't you liquidate a futures position at the current market price?

re: potential losses

Stocks, commodities, and PMs are clear: you can lose your investment. Options are all over the map. Short selling is the worst, but covered calls limit that exposure. Each option or derivative is different, and you have to examine each side separately. Many combinations are intended purely to limit the downside risk of simpler trades. I don't see how you would generalize about this.

Re: formatting

for Futures Market, "tertiary" shows up under "secondary" no matter what width I make the browser. The rest is okay, given the limits of drawing with ascii. Your font declaration is pretty hosed; you might be able to get fixed width fonts if you correct that.

Don't you know how to make a table in HTML? It's not hard.

Tom W. Bell said...

Agreed about the problems of generalizing about losses, Chris. I'll end up just saying "sometimes" or the like.

Yes, I know how to code tables. Check out my homepage at www.tomwbell.com--all hand-coded, too! I just didn't want to mess with it, here.