Proof of Liquidity System

Overview

PreDex is a prediction market powered by Predix Network and its native PRDX that allows for transparent and non-custodial leveraged predicting on Uniswap pairs. PreDex uses the native Uniswap v2 price oracle for prediction settlement. This oracle uses a so-called TWAP (Time Weighted Average Price) to protect against flash-swap attacks and price manipulation.

Prediction Life-cycle

When a user makes a prediction it goes through the following life-cycle:
- The user sets the parameters on the prediction submission form on the UI (price, amount, timestamp, leverage)
- The user approves the PreDex Router.
- The user opens the prediction, transfering the PRDX tokens to the PreDex Router.
- A few minutes before the prediction timestamp the user can start the settlement process by initiating the TWAP (see settlement for more details).
- After the TWAP time has elapsed, the user can finish the settlement by closing the prediction. The rewards are calculated (see risk/reward) and paid to the user depending on the performance of the prediction.

Settlement

PreDex uses a TWAP of length T to settle predictions. As different markets have different amounts of volume, pool liquidity and arbitrage traders, the length T is market-specific. A bigger value for T means a more secure way of price fetching but this naturally causes the process of closing a prediction to extend. Example: the PRDX/ETH market has a TWAP length of 10 minutes, so T = 10 minutes. The user initiates the TWAP (which is available 10 minutes before the predicted timestamp) and 10 minutes later closes the prediction, at which time the full settlement happens.

Theta Decay

Very much like options, predictions can suffer from theta-decay. PreDex uses linear theta decay that starts after a fraction a of the prediction age. Currently, a = 7. This means that for a prediction of 1 week in the future (a prediction age of 1 week), there is a decay-free period of approximately 1 day that starts when the prediction becomes available to close (at the end of the week). The decay-free period is set in place as to give the users more flexibility and ease of mind as to what timestamps to use. The linear theta decay allows the user to not immediately close their prediction when it becomes available, even when it is at a net-loss. Example: let's say you have an open prediction that is about to become available to settle but the prediction is off by a lot, meaning that if you were to settle the prediction at current prices, you'd lose. If you think price will move to your predicted price however, you can decide to wait and see if the price does indeed move closer to your target. Note: the longer you wait, the more your prediction has suffered from time-decay and thus the reward will be lower and the opportunity cost higher. When the theta decay factor has gone to 0, one loses the prediction. This means that instead of winning 0 PRDX you lose and get back a certain amount determined by the refund percentage.

Risk/Reward

Making a prediction naturally comes with a risk/reward ratio. With PreDex, the risk is very simple: there is a preset refund fraction R. This fraction is market-specific and tells you what percentage of the prediction value you get back in the case of a wrong prediction. This means that for every prediction you open, no matter the size, you can only lose (1-R)%. The reward is limitless. To determine whether your prediction was right or wrong, PreDex uses a so-called "invalidation curve". This curve gives the maximum allowed percentage difference between the predicted price and the price at settlement for a prediction before being considered wrong. This curve only depends on the prediction age; the longer the prediction, the more you can be off. The reward depends on multiple variables: the prediction value, the predicted price, the close price, the prediction age, the time of closing the prediction and the leverage. For the exact equations of the reward curve and dependence on the forementioned variables, please refer to the whitepaper: https://predix.network/predex.pdf. PreDex also supports leverage, multiplying the potential rewards by the amount of leverage you use. The invalidation curve gets divided by the amount of leverage used. Example: a prediction with age T having an invalidation percentage of 5% at 1x leverage is invalidated at >1% at 5x leverage.