A New Type Of Prediction Market: RoosterRed’s Prediction is Payoff (PIP) Markets
Hello Friends,
Today we are going to look at a fantastic new model for prediction markets. I hope this helps people. Prediction is Payoff (PIP) markets will give all participants in the market a fair deal; they will help researchers of prediction and betting markets, potentially answering many unknown questions; and they will make all asset markets more rational, eventually reducing market manipulation (though they might contribute to it initially).
This type of prediction scheme is the fairest to participants in that their payoff is inversely proportional to the probability of their prediction. That may sound complicated, but this is the simplest betting/futures/prediction scheme ever to anyone with basic knowledge of probability. If the market predicts you have a 10% chance of being right, you should get ten times the money you put down if you are right. If the market predicts you have a 50% chance of being right, you should get twice the money you put down if you are right. That is it. That is all a participant does: chooses a prediction, gives the probability it occurs, and puts down collateral. The payoff for a correct prediction is literally the prediction itself: Prediction is Payoff (PIP).
The simplest example is a Yes/No market. Since I think crypto will be the first to adopt PIP schemes, an example would be a Bitcoin Price Prediction market. The market would simply be asked “Do you think Bitcoin will rise in value in US dollars in 24 hours?”. People make predictions as outlined above, saying Yes or No, giving the percentile probability they agree to, and authorizing the Smart Contract to deduct their funds if their prediction is matched. If the prediction can be funded, which just means matched to an opposite prediction, the winner gets a 100% payoff based on their bet. We need to outline some stipulations to this scheme to ensure it both furthers economic knowledge and truly gives everyone a fair deal.
It is absolutely vital that data on every prediction is public, accessible, and archived. The basic user interface would be similar to current cryptocurrency exchanges, and would be constantly updated to provide the participant with the best, most current, and most understandable data. See Cryptowatch for an example, with Bid and Ask prices being replaced with Yes and No probabilities. You would be able to quickly see what bids are out there for both the Yes and No options. If the market has a 50 Bitcoin prediction at 90%, you could see that easily. The data has to be freely accessible to reduce the risk of market manipulation by the house. The house might have an incentive to create misleading or hard to understand user interfaces. Outsiders must be allowed create better user interfaces for participants. Researchers must be getting the best possible data, all of it, and be able to run experiments with live data feeds.
The matching algorithm is very simple. The Smart Contract looks at the highest Yes prediction and the highest No prediction. If they add up to at least 100%, the contract is funded. Predictions of the same probability are ordered by time submitted with the earliest first. The Smart Contract pays all fees! This ensures that every participant’s prediction has the exact payoff it should. This also ensures that the market can be truly efficient.
If you haven’t figured out how the house makes money, that is because you are too used to outdated models. In our simple example, Yes and No predictions are treated separately because they are different predictions. It is the markets job to be efficient, not our math. For instance, it is completely conceivable that Yes predictions turn out to be more accurate than No predictions overtime. Are optimists smarter than pessimists? Let that sink in. We can give researchers incredibly clean data that is easy to interpret and allow them to research the predictive power of markets like never before. The sum of the market prediction of Yes and the market prediction of No is actually an empirical question. It would also be very interesting to look at predictions that weren’t funded in several situations.
Back to how this is possible and how the house makes money. Markets are rarely in perfect equilibrium. Even when a market is in equilibrium, fees in crypto are very low. The only costs to the contract once it is created are network fees. Network fees are very low and are generally adjustable based on priority of closure (higher fees means your transaction is processed faster). But even fast fees are incredibly inexpensive in many networks. A fast network transaction generally costs less than .005% of the total transaction. So even when a market is efficient, and transactions are given a high priority on the network, the house is not losing much money, proportionally.
The scheme can be described a number of ways. You could say the house encourages the creation of arbitrage for itself to exploit. You could say that the spread, a common term in betting, is determined by the market. It doesn’t matter. The point is that every participant gets a fair deal. Each winner of a prediction contract literally gets the payoff of their prediction: Prediction is Payoff (PIP). There are also a indeterminate number of research questions that researchers will be able to ask and study thanks to this scheme. Finally, if done correctly, these markets will reduce market manipulation in the long run, especially pump and dump schemes and insider trading.
In order to reduce market manipulation, these markets would ideally be open to all participants and anonymous. This probably isn’t possible because of gambling laws, but lets look at the current pump and dump problem in the cryptocurrency sphere and see how a truly open market could mitigate the problem. Most pumps are not done by single entities, but groups of dozens of people who coordinate through Discord, IRC, and other chat applications. Every individual participant has the incentive to use Yes/No PIP price prediction markets before the pump happens to make extra money. An uptick in activity on a cryptocurrency PIP market could signal an incoming pump, alerting investors that the pump they are seeing was planned out ahead of time. These markets would be better warning signals of incoming pumps if they were open and anonymous because the pumpers would want to hide their identities to shield themselves from legal repercussions (as crypto becomes regulated), many pumpers are underage (there are some really rich teenagers in crypto right now), and most pumpers probably avoid paying taxes as much as possible.
Market manipulators will of course exploit Yes/No PIP markets as well. I fear that initially they may encourage market manipulation. This will reduce with time, if it does happen. The leaders of pump and dump groups are psychologically savvy, and some have an army of internet trolls that spread false information. Upticks in Yes/No PIP markets could be spun as insider trading. Also, market manipulators could use the PIP markets to send false signals. So, not everything about these PIP markets is good, but more information is always better in the long run.
I hope you found this interesting.
Notes
- PIP markets are also ideally suited for sports predictions (Team A or Team B) and run-off elections (Candidate A or Candidate B). Both of these types of futures markets are hot topics in economics.
- PIP markets theoretically would work with more than two options (N-Option PIP Markets). There are major issues that I have not completely ironed out. I will post an example when I have a real-world workable model down to the user interface for participants.
- These PIP markets do not have to be done through Smart Contracts. In fact, they could have been done years ago, when electronic sports betting was first introduced.
- It is essential that the markets have accurate data on outcomes. This is known as an Oracle problem. Right now, the best solution I know about is Town Crier, and coincidentally Town Crier currently offers authenticated data feeds for cryptocurrency prices. I do think it should use more that one data source and require consensus.
- A completely decentralized PIP market is possible using ERC20 tokens or something similar. Yes and No would be given separate tokens and once the outcome was determined by the Smart Contract, payouts would automatically be deposited to the winners.
- Ideally these markets would accept the major cryptocurrencies and pegged ones like Tether to promote long term bets. Most cryptocurrencies are highly volatile, so participants might avoid bets that were longer than a day unless they could use a currency that was pegged to a relatively stable and predictable fiat currency.
- I do not recommend these markets have their own cryptocurrencies. That is an absolutely terrible idea for many reasons. The easiest example being high volume events like a US Presidential Election. In the run up to the election, the PIP currency would appreciate; after the election, the PIP currency would depreciate. If one candidate was heavily favored and won, it is possible, probable even, that the winners would lose money in terms of US dollars or Bitcoin.
BITCOIN (BTC):: 1jsF3BFYNGRG7rtzjuKbtPURcKzod4etq
ETHEREUM (ETH): 0xa26724b29Ee557960eB9A92d659BB88ddA8d47d1
LITECOIN (LTC): LTEU39XFtD1nj1QqpuVf97ok8F5bFL5VeK
ARK (ARK): ATkSmt2LqAwKm6zmd8aYwZB4cUEZGrcCym
MONERO (XMR): 48nXBajorzkZLeotCBw1XgYaUgENGuxS1hhK75hoGgUSQVHsoVa3EWj86EvqKjeorUY77SVu8fj8TPevfz3dnmX567djac2
Your Friend,
RoosterRed
[email protected]
XMR:
Note: I usually have to correct grammar mistakes in my articles, so this will likely slightly change. It is usually little, inconsequential mistakes like “their” instead of “there” and word substitutions. I may have to do a major update or attach addenda with an article this intricate. Steemit gives you a week to do that.
Addendum 1: Scattered Thoughts On The Matching Algorithm
I am not fully convinced that the matching algorithm above—specifically its first step at simply looking at the top Yes and No bids—is the ideal way to go. I will offer one alternative here, and then explain why I still favor the algorithm in the article.
An alternative matching algorithm would look at all bids and maximize the amount of contract funded (maximum funds in contract). As consequences, the spread would be minimized and and market efficiency would be maximized. Note that we did not maximize the participants: we maximized the funds in the final payouts. This is attractive for all three attributes given (funds maximized, efficiency maximized, spread minimized).
I still favor looking at the top Yes and No bids, seeing if they add up to at least 100%, and funding the contract if they do for three reasons. First, (human) participants aren’t always going to know exactly what is going on with the alternative algorithm, even in a simple Yes/No PIP Market. This problem is compounded in N-Option PIP markets (more than two options). Next, it is the market’s job to be efficient, not our math. Markets should strive to become efficient; we shouldn’t need to help them out; and PIP markets can be perfectly efficient as they are. Last, as currently outlined, PIP markets truly operate marginally, which is a very nice thing for models to do in economics. I think everything can be doubted though, including the sacredness of marginality in economics.
Regardless, I think PIP markets can operate under either algorithm, and it is an open question which is better, though I currently favor the marginal one in the article.
There is a TVM (time value of money) flaw in this model. The markets can only be in perfect equilibrium when the TVM for the participants is one (one bitcoin at the time of participation is worth one bitcoin at payout). This matters little for short term predictions, but a lot for long term predictions. Regardless, I think the model is promising for research in both short and long term cases, but it does have a TVM flaw that would prevent perfect equilibrium in many cases.
I should specify what Perfect Equilibrium is in the market, why it is interesting, and when it can actually be reached. Lay readers are probably confused. And even advanced readers might mistakingly think Perfect Equilibrium is impossible.
Perfect Equilibrium in a PIP market would mean the following conditions are true:
1. 1 BTC at the time the predictions are made is worth 1 BTC at payout.
2. The market predictions add up to 1 or 100% (to some negligible degree).
If the strong form of the Efficient Market Hypothesis (EMH) is true, the percentile market predictions will be the product of all publicly available information about the likelihood of the event in question, which means the payouts will be the market's expected value of each possible outcome. Many economic models controversially assume or conclude the later is true in any competitive market. What PIP markets uniquely give is a very clean real world model to investigate this.
In fact, N-Option PIP markets could help research the plausibility of the EMH by seeing how markets adjust to an increase in outcomes. A four person election shouldn’t be harder for efficient markets to deal with than a runoff election and is still a relatively simple market. If PIP markets either scale easily or breakdown rapidly then how many view the EFH could change. The glaring drawback of the PIP model is Condition 1 above.
When is 1 BTC at prediction worth the same as 1 BTC at payout (Condition 1)? Under a continuous discount rate function, never, but continuous discount rate functions are preposterous for very short period of times. If people truly know that payouts will happen and happen in a fair way, then in very short period of times their discount rates will be effectively zero. But there is a another path to perfect equilibrium most readers will miss.
If people truly know that payouts will happen and happen in a fair way, Condition 1 is effectively reached in all circumstances where participants would have held, not traded or spent, the prediction funds (the funds going into the PIP contract). This is trivial, but can be a little hard to immediately see. It is based on Opportunity Cost. If the next best option a predictor has is to simply hold the funds the duration of the contract, then they don’t have a discount rate in any meaningful way. They might if you ask them, and they would if you modeled using a continuous discount rate function, but, in any meaningful way, they don’t have time preferences for the funds they put into the contract.
I think I will be rewriting this article. But I now realize that this equilibrium--let's call it Outcome Payout Equilibrium--is a lot more interesting on its own. It is just forcing the market to attempt to create Expected Value instead of assuming Expected Value should or will eventually hold. I am sure this isn't an original idea, but I don't know why it isn't talked about. I just saw it as a solution to create a prediction market with the correct payouts that bewilderingly was not being used. The trouble for this very simple market to find equilibrium (fund/find Expected Value) is disconcerting to me.
I deleted a post on informational asymmetry because it was not accurate. Buying stocks is not like buying a used car in an informational asymmetric environment. It is like buying a used car in this environment:
Where the sellers and buyers both could be using different parts of publicly available information with little to no overlap: information does not flow between outcome choices. All this follows from arbitrage. Severe and persistent Outcome Payout Disequilibrium breaks the EMH.
I am working on patenting this. This is IP.
Get your post resteemed over 50000+ followers and get upto $17+ value Upvote. Your post will skyrocket and
give me maximum exposer. See our all pakages at:
Congratulations @roosterred! You have completed some achievement on Steemit and have been rewarded with new badge(s) :
Click on any badge to view your own Board of Honor on SteemitBoard.
For more information about SteemitBoard, click here
If you no longer want to receive notifications, reply to this comment with the word
STOP
Do not miss the last announcement from @steemitboard!