(Entries marked with ** were changed within the last 7 days; entries marked with * were changed within the last 30 days.)
Other differences:
Edit this entry / Log info / Last changed on Fri Mar 17 14:02:01 2000 by Peter McCluskey
While there are some reasons to hope that the CFTC will be willing to approve some idea futures trading, everyone I have talked to who seemed to know something about these issues expects it will take substantial investments of time (presumably talking with lawyers) and money. I doubt that I have the patience or skills to accomplish this in the near future.
While real money is a fairly reliable way of limiting the extent to which prices reflect anything other than honest predictions in markets which are widely known and watched, I doubt that this is the only way to accomplish that.
If a play-money market can attract a good enough set of initial participants and if the means of acquiring wealth in that market insure that a large fraction of the wealth in the market is controlled by people who have been successful at predicting the value of claims, then it can potentially be at least as hard under this system to alter market prices for undesirable reasons than with real money (assuming that people value their scores and that the one-account-per-person rule can be enforced well enough - assumptions which are not certain).
Still, there is a clear need for a real-money idea futures market to handle claims which require significant expense to research.
Edit this entry / Log info / Last changed on Mon Jan 24 14:02:01 2000 by Peter McCluskey
Money can also be created if traders achieve a negative account value (such accounts become inactive, possibly permanently).
Money can be removed from the market if traders lose money to the automatic market makers, or if traders have their accounts suspended (e.g. for creating multiple accounts used by a single trader, or for colluding with other traders in such a way as to make a trader's performance misleading).
Also, note that, because the automatic market makers will often lose money, the typical trader in this market will have a positive return on investment.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
All other information (including your email address) will be made public.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
If it generates enough advertising revenue to be a successful business, I plan to distribute some of that revenue to traders who are responsible for making prices more accurate.
Edit this entry / Log info / Last changed on Thu Dec 22 14:02:01 1999 by Peter McCluskey
Buying NO contracts is probably an easier way for a majority of people to initially figure out what is happening. Imagine that when you sell a contract you don't own, that you automatically buy (from the "bank") a pair of YES and NO contracts (the price of the pair always adds up to a fixed amount - in the case of Boolean claims, that value is 1). Your trade then amounts to selling the YES part of that pair and keeping the NO contract.
Selling short a contract is probably an easier way to think of it when you are actually trading, since it avoids the need to mentally convert between NO prices and YES prices. Forget about NO contracts, and simply treat selling a contract you don't own as going into debt that many YES contracts (or owning a negative number of contracts).
Your cash balance looks a bit different under this representation; some of your cash is classified as "restricted cash" to create a result that is equivalent to having bought a pair of YES an NO contracts.
Prices are expressed in terms of YES contracts almost everywhere. The list of holdings page offers an option to switch between a buying (going long)/selling short YES contracts representation (the default) and a YES/NO contract representation.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Other alerts (currently alerts based on Available Credit are the only type in this category) are checked several times per hour.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
They are orders which remain inactive until the last trade for that claim drops below the stop price (for a sell order) or rises above the stop price (for a buy order). A stop limit order becomes a regular limit order when the stop price is hit. A stop order becomes a regular market order when the stop price is hit.
They are typically used to limit your risk by bailing out of your holdings when the market trend goes against you. Beware, though, that if the size of a stop order is large relative to the size of other trader's orders, the stop order will cause the price to move even further against you.
Edit this entry / Log info / Last changed on Thu Dec 23 14:02:01 1999 by Peter McCluskey
Claims in which trading has been halted will normally have an entry in the "comments" field (available on the claim description page) giving a reason for the halt and an estimated of when trading will resume.
Suspending trading upon unexpected news would be a nice feature, but the effort required to do it promptly enough to be of much value appears prohibitive for the forseeable future.
Edit this entry / Log info / Last changed on Thu Dec 23 14:02:01 1999 by Peter McCluskey
Currently, my policy is that claims must be written so that I expect the result will be readily observable on the settlement date.
Claims based on claim templates will probably be approved quickly (maybe instantly in the future). Other proposed claims will be listed for comments for at least 7 days after any change before being approved.
"Readily observable" means that the event must be of the kind that has normally been reported as fact (not opinion) in standard news sources. If the event is not of sufficient interest to the general public that CNN would be expected to report it, the claim must identify a publicly available web site or similar publication (which is widely available at little cost) which has normally reported events of this kind in the past.
Claim authors are strongly encouraged to design claims so that they could be judged by reading a number from a standard site, for example a claim about the unemployment rate would typically reference this unemployment statistics page from the St. Louis Fed. Without a specific reference such as this, it would be easy to overlook ambiguities such as seasonally adjusted versus raw data.
The Iowa Electronic Markets provide good examples of claims that are clear and easy to judge.
Claim authors should not consider the claims on The Foresight Exchange to be good examples of how claims should be worded. A majority of those claims contain ambiguities that would not be accepted here.
Currently, I plan to judge all claims, and I only accept claims which I am confident I will be able to judge with little effort. I am uncertain whether I will make use of other judges in the future.
If there is clear difference between the literal meaning of the claim and the reasonable expectations by most traders concerning the intent of the claim, the expectations of the traders will prevail. The intentions of the claim author are relevant only to the extent they could reasonably have been inferred before a disputes arise.
Edit this entry / Log info / Last changed on Wed Feb 2 14:02:01 2000 by Peter McCluskey
If the bid price is greater than the ask price when trading starts, prices will be determined by averaging the bid and ask price on orders that are matched.
These rules will probably be modified in the future as experience tells us whether many claims that pass this end up being ignored.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Judgement dates more than 5 years in the future are currently prohibited. While I want to encourage long-term claims, I don't want to commit to keeping this market open indefinitely, nor do I want to accept claims which cannot be resolved within the time period for which I'm willing to commit to keeping the market running. If you want claims longer than 5 years, please help to convince me that the market will be in demand further in the future (such as by convincing more people to become addicted to it).
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Claim proposals that are close to being well-worded and appropriate for USIFEX will be posted to the usifex-discuss mailing list. Proposals that aren't close will be rejected without being brought to the attention of other traders.
Edit this entry / Log info / Last changed on Wed Feb 2 14:02:01 2000 by Peter McCluskey
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Symbols with a | in the middle are for conditional claims which result in the value of the symbol before the | if the symbol after the | is true.
Symbols with a ~ in the middle are for conditional claims which result in the value of the symbol before the ~ if the symbol after the ~ is false.
Symbols starting with an @ are templates (used for creating claims with standard wording).
Symbols for recurring claims end with a number indicating the year followed
by a letter indicating the settlement month similar to the way symbols for
commodity futures are created. For example, a symbol for a consumer price
report that is expected to be released in December 2000 might have a symbol
of CPI0z. Note that the settlement month may be different
from the time period which the underlying data describes (e.g. the
CPI0z claim mentioned above would refer to the CPI report about November
prices).
Here are the letters with the months to which they correspond:
Currently only one digit is used to indicate the year, but multiple digits may be used in the future.
Edit this entry / Log info / Last changed on Tue Jan 25 14:02:01 2000 by Peter McCluskey
Conditional claims are bought and sold with contracts of the underlying claim rather than with cash. Prices are quoted in units of the underlying symbol unless otherwise noted (at least it should work that way; I'm not sure whether it is consistent about it yet).
If you don't hold the appropriate contracts of the underlying claim when you trade a conditonal claim, the system will create them as needed by purchasing pairs (units of YES and NO contracts) of the underlying claim.
If the underlying claim is settled before the dependent claim, contracts of the conditional claim which are conditioned on the winning side of the underlying claim are converted into unconditional claims. Conditional claims conditioned on the losing side become worthless, and those contracts will disappear. For example, if X is judged true, Y|X contracts become Y contracts, and Y~X contracts become worthless.
While this approach may make it harder to understand what will happen when you trade a conditional claim than with claims of the form "event X happens and event Y happens", it makes it easier to relate the prices to probabilities of individual events.
Let's suppose we have a claim GORE that says Al Gore will be elected president, a claim WAR that says U.S. soldiers will die in combat, and conditional claims WAR|GORE and WAR~GORE. Here are examples of what would happen if you start without any holdings in any of these claims, and buy or sell one of the conditional claims at a price of 0.65 GORE contracts or NOT GORE contracts.
| Action | New GORE holdings | Change in cash holdings | Profit/Loss (in Quatloos) from trade for different settlements | ||
|---|---|---|---|---|---|
| Gore elected | Gore not elected | ||||
| Buy 1 WAR|GORE | -0.65 GORE (0.65 GORE NOs) | -0.65 Quatloos | soldiers die in combat | +0.35 | 0 |
| soldiers don't die in combat | -0.65 | 0 | |||
| Sell 1 WAR|GORE | 0.35 GORE | -0.35 Quatloos | soldiers die in combat | -0.35 | 0 |
| soldiers don't die in combat | +0.65 | 0 | |||
| Buy 1 WAR~GORE | 0.65 GORE | -0.65 Quatloos | soldiers die in combat | 0 | +0.35 |
| soldiers don't die in combat | 0 | -0.65 | |||
| Sell 1 WAR~GORE | -0.35 GORE (0.35 GORE NOs) | -0.35 Quatloos | soldiers die in combat | 0 | -0.35 |
| soldiers don't die in combat | 0 | +0.65 | |||
Currently claims can only be conditional on unconditional Boolean claims. I don't know whether support will be added for claims conditional on non-Boolean claim types or claims conditional on conditional claims.
Edit this entry / Log info / Last changed on Wed Dec 22 14:02:01 1999 by Peter McCluskey
All claim types must have upper and lower limits on what values are legal, even if the idea involved refers to things with unlimited values. The lower limit must be at least -1000 and at most 0. The upper limit must be at least 1 and at most 1000.
Boolean claims must have limits (for both legal and probable values) of 0 and 1.
Limits below 0 or above 100 are generally discouraged unless they enable a direct mapping between claim values and the standard representation of the idea that the claim is trying to predict. If the natural representation would produce a significant chance of numbers above 1000 or below -1000, the claim terms should scale them by some power of ten to produce numbers that are likely to have absolute values below 100.
Edit this entry / Log info / Last changed on Thu Dec 23 14:02:01 1999 by Peter McCluskey
The data will be available under a license which allows essentially unrestricted use after 24 hours(?). The exchange will reserve the right to charge for real-time access to trading data.
Edit this entry / Log info / Last changed on Fri Mar 10 14:02:01 2000 by Peter McCluskey
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
The system is probably secure enough for a game, but not for real money. Passwords are not encrypted, and anyone who can intercept email sent to your address, or who can gain access to the server's file system, can get your password. I plan to offer before too long an option to encrypt passwords for any traders who want it enough to accept that if they forget the password, it will be expensive to reset it.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey
The idea futures software is unlikely to add any security holes, but people are encouraged to look at the source code to see if I've overlooked anything.
The system designed so that the cgi-scripts should only
be able to affect the system by sending commands to the server daemon.
User input is checked fairly carefully when it reaches the server
(see validate_fn in IFServer.py) to insure that strings
which will be evaluated by the database cannot contain characters that
might be used to execute arbitrary functions.
I believe the only other way that user can affect the filesystem through the idea futures web interface is when creating or modifying a claim, a .html file is written (see GenSearchFiles in IFClaim.py). The file names are strictly limited by the restrictions on what characters can be used in a claim symbol, and I can't see any way in which the files could be executed.
I have not yet done much analysis of the security weaknesses in the Python cpickle module which converts user input to Python objects to pass between cgi-scripts and the server.
Edit this entry / Log info / Last changed on Tue Feb 2 14:02:01 2000 by Peter McCluskey
There may be some value in creating selective indices, such as one that sums up which political party the market expects will do a better job of accomplishing generally agreed upon goals, but it will take a fair amount of thought to do this in an unbiased fashion.
Edit this entry / Log info / Last changed on Thu Dec 16 14:02:01 1999 by Peter McCluskey