Pricing tokens created via the UMA CFD contract

[snake-poison]

Pricing of these synthetics is tricky and while it may be outside the scope of the competition, if we want to just improve on the EMP call options design. It's not intuitive and it is not a show stopper, but if the participants don't clearly understand the risks they take on when buying a Short or Long pair than they might be reluctant to try it. Let me illustrate the issue through the example UMA gave here for their LSP example : https://docs.umaproject.org/synthetic-tokens/long-short-pair#lsp-example

Again please refer and read the story of Alice, Troy and Rachel in the link above.

To summarize: Alice in the story seems to be taking the role of a market maker who is long term ETH bullish. She is looking to generate yield on her idle ETH. She gives up gas to set up this position. She set the strike at 3,000 when the price is currently at 2,5000. She sells long tokens to Troy and short tokens to Rachel.

Note that in the story, UMA did not present the pricing of these sales! Let's analyze the position posed here and how Alice might have priced these transactions to make a profit ( which she must as she spent gas to setup the position).

• Collateral Per Pair was set to 1 so we know that before expiry, 1 long token and 1 short token are worth 1 WETH. That would mean before expiry 1 short token is worth .5 WETH and 1 long token is worth .5 WETH to the CFD contract (that anyone can redeem from). However, at expiry with an oracle price of 2,500 the long token is worth 0 and the short token is worth 1 WETH.

Let's extend the story to include how Alice would price her LONG and SHORT tokens.

How Alice should price the SHORT:

• Alice is motivated by the opportunity of making yield. That means that she must receive more than 1 WETH plus the cost of gas for her position to be economical . While the story mentions her approaching Troy first, the more tricky sale is the short token side with Rachel if she wants to take a neutral position.
• She needs to approach someone that is bearish on the price of WETH, but wants to acquire WETH. The buyer of the short tokens she is selling must receive the short tokens at discount, as there would be no upside to buying the tokens that are in the money.
• Alice knows the minimum she could take for the 1 short token in the money is a little more than .5 WETH but the max a rational buyer would pay is <1 WETH. Since cumulatively she must make more than 1 WETH (plus gas) for the pair, the discount given to short buyer must be made up in the sale to the long buyer. The time until expiry and recent volatility will also impact the pricing of her short tokens.
• Let's say expiry was within 3 days and the price has remained stable recently. She finds Rachel who is willing to buy the 1 short token that will be worth 1 WETH if the price stays below 3,000 for a 5% discounted price of 2,375 (.95 WETH @ 2500)

Rachel's story based on his actions( Short Token Buyer):

• Rachel takes Alice up on her offer to purchase 1 SHORT at the price of 2,375(5% discount on current price of 2,500).
• Expiry past $3,150 it would have been more economical for her to just have purchased WETH. If at expiry the price is below $2,375, it would have made more sense for her to wait and buy more WETH directly from the market at the same price she entered his position.
• Rachel with this transaction is in the money between $2,375 and $3,150. She maximally profits at the price point of 3,000 as she would get 1 WETH valued at $3,000 for the price of $2,375. That's $675, a ~28% increase unrealized gains for holding the position 3 days. If at expiry the price were $2000, she would have paid 2,375 to receive 1 WETH an unrealized loss of $375 a -15% decrease on her 2,375.

How Alice should price her LONG tokens:

• Alice feels confident on being able to sell her ITM short tokens for a 5% discount (.95WETH) so the calculation to sell her long tokens must start >.05 WETH. At the current price of 2,500 that's a premium of at least $125 per token. Alice needs to make her yield here so assuming she wants to target ~$60 per token to part with her ETH for 3 days, she would price her long tokens at $185.
• Let's say expiry was within 3 days and the price has remained stable recently. She finds Troy who is willing to buy the 1 long token that will be in the money if the price goes above 3,000 for $185 (.074 WETH @2500 ETHUSD).

Troy's story based on his actions (Long Token Buyer):

• Troy takes Alice up on her offer to buy 1 Long Token at the price of $185. His position is straightforward, a total downside of the $185 at an expiry below or at 3,000 in three days. His breakeven in USD occurs at an expiry $3,185 when his 1 Long token is able to redeem ~0.058 WETH.

• Had troy just bought WETH instead of the Long Token, he would have received 0.074 WETH (185@2500 ETHUSD) at the time Alice sold him the Long token. So the real breakeven would be for someone taking this position would be when they receive more than .074 WETH which is at expiry of $3240. At this price, the value of Troy's position would be $240 (~30% increase on his 185) and the turning at which leverage starts working to his favor.

• By purchasing the long token instead of WETH, troy gives up upside in exchange for leverage. Had he bought 0.074 WETH and the price moved to 3,000 he would have an unrealized gain of $37(~20 percent). Instead he loses %100 of $185.

Results according to the fictitious story if we take this story as canon:

• Alice sells 100 Long tokens to Troy @ $185 per token for a total of $18,500. Alice then sells Rachel 100 short tokens at @2,375 for a total of $237,500

• Alice who starts with 100 WETH (250,000) secures 256,000 USD a 2.4% gain over just selling her WETH. The contract expires at $3600.

• Rachel takes the biggest loss here with her 100 short tokens now being worth 83 WETH, but because of the price movement her value is still up from 237,500 to 300,000 (~26% in unrealized gains). Had she just bought WETH she'd have 100 WETH that would be valued at $360,000 ( a ~51.5% increase on her initial investment)

• Troy makes a killing here. His 100 LONG tokens are now worth 16.666 WETH a value of ~$59,997.6 from his initial investment of $18,500 (a ~224% unrealized gain). Had he just bought WETH with his original 18.5k, he would have had 7.4 WETH that would have been worth ~26,640 (a not as impressive 44% unrealized gain). Be like Troy.

Let me know if this makes sense. If it does, how can generalize the formula for pricing these contracts?

[markoinether]

That would mean before expiry 1 short token is worth .5 WETH and 1 long token is worth .5 WETH to the CFD contract (that anyone can redeem from)

I think this assertion is wrong, I believe that before expiry you you can only "reduce" position, meaning you would have to provide 1 long + 1 short token to the contract to get 1 ETH back (could be fractions of the tokens too, but always in 1:1 ratio), this makes logically sense (and they explain it in the video like this) since this is how you reduce the position in a neutral way.

Counterexample: Since contract has no inherent pricing for short / long tokens before expiry you could theoritacally redeem 1 short token for 0.5 before expiry when there is 99% that the position will expire out of money... making the system unusable.

[ckoopmann]

She needs to approach someone that is bearish on the price of WETH, but wants to acquire WETH. The buyer of the short tokens she is selling must receive the short tokens at discount, as there would be no upside to buying the tokens that are in the money.

This is only due to the fact that we are using WETH as a collateral in this case though right ? If we would use a USD stable coin as collateral this is not the case, I would think.

[markoinether]

Alice is motivated by the opportunity of making yield. That means that she must receive more than 1 WETH plus the cost of gas for her position to be economical.

This assertion does really hold in practice, look at the picture below (binance option market) https://voptions.binance.com/en , you can see short option + long option at the same strike do not add up the price of underlying, there is in fact no reason they should, everyone would be an option market maker then :).

[ckoopmann]

Alice knows the minimum she could take for the 1 short token in the money is a little more than .5 WETH

Why would you say that, if a token is just barely in the money and almost guaranteed to get not more than say 0.1 collateral tokens at expiry, why would anyone pay 0.5 weth for it ?

[ckoopmann]

That would mean before expiry 1 short token is worth .5 WETH and 1 long token is worth .5 WETH to the CFD contract (that anyone can redeem from)

I think this assertion is wrong, I believe that before expiry you you can only "reduce" position, meaning you would have to provide 1 long + 1 short token to the contract to get 1 ETH back (could be fractions of the tokens too, but always in 1:1 ratio), this makes logically sense (and they explain it in the video like this) since this is how you reduce the position in a neutral way.

Counterexample: Since contract has no inherent pricing for short / long tokens before expiry you could theoritacally redeem 1 short token for 0.5 before expiry when there is 99% that the position will expire out of money... making the system unusable.

I agree with Marko. While 1 long + 1 short token are together worth 1 WETH I don't think you can assume that the value is distributed 50 / 50 on them, since you cannot redeem them individually prior to expiry.

[markoinether]

She needs to approach someone that is bearish on the price of WETH, but wants to acquire WETH. The buyer of the short tokens she is selling must receive the short tokens at discount, as there would be no upside to buying the tokens that are in the money.

This is only due to the fact that we are using WETH as a collateral in this case though right ? If we would use a USD stable coin as collateral this is not the case, I would think.

In this case it does not really matter what is the collateral, you are basically "splitting" 1 ETH into a shortETH and LongETH side that together always give you the same exposure as 1 ETH. think about it like this if the strike is 10 000 USD and expiry in 3 months, there is no inherent way how to say, what is the probability that ETH will be in the money (hitting the strike). Basically market decides, if I think chance is 10% I will be willing to pay 0.1 ETH(in fact less than that but lets keep it simple) and bob is ETH bear and thinks the chance is 1% he is willing to sell me the option for 0.05. He will be willing to sell me long option I will be willing to buy it.
Only after expiry the contract knows how to split rewards between short tokens and long tokens based on the price oracle, it does not matter if we are talking about options, range bonds or whatever else. Contract only allocates between these, but it always holds that at the expiry, if you hold both 1 Long token and 1 short token you will back 1 (collateral token, could be anything) back, no matter what is the distribution of ETH between short / long. Could be 50/50 could be 100/0 or anything in between. Since this condition is always met, you can redeem 1short + 1long for 1 underlying at any time during contract's lifetime, not just aftery expiry

[ckoopmann]

In this case it does not really matter what is the collateral, you are basically "splitting" 1 ETH into a shortETH and LongETH side that together always give you the same exposure as 1 ETH.

I don't think that is quite right, the way I understand it the two tokens (long / short) always redeem to one token of whatever you choose as collateral. If you choose USDC as collateral then the two tokens would always equal 1 USDC together. So they would give you no ETH exposure whatsorever if you don't choose WETH as collateral no ?

[markoinether]

In this case it does not really matter what is the collateral, you are basically "splitting" 1 ETH into a shortETH and LongETH side that together always give you the same exposure as 1 ETH.

I don't think that is quite right, the way I understand it the two tokens (long / short) always redeem to one token of whatever you choose as collateral. If you choose USDC as collateral then the two tokens would always equal 1 USDC together.

Correct, incorrect wording on my part, 1 long + 1 short always give you 1 collateral, no matter what is the collateral.

[markoinether]

In this case it does not really matter what is the collateral, you are basically "splitting" 1 ETH into a shortETH and LongETH side that together always give you the same exposure as 1 ETH.

I don't think that is quite right, the way I understand it the two tokens (long / short) always redeem to one token of whatever you choose as collateral. If you choose USDC as collateral then the two tokens would always equal 1 USDC together. So they would give you no ETH exposure whatsorever if you don't choose WETH as collateral no ?

I believe the way the mechanism is currently designed, is that the price oracle always need to be a Token/USD pair and collateral in Token, could be wrong here.

To your question, you would get price exposure if collateral is in USDC, what you might not get, is the actuall delivery ;), in practice contract would be liquidated before expiry because if price of eth goes up at a certain point collateral would not be able to fullfill the obligation of the contract. e.g. Collateral is 140% of the opened ETH/USDC position in USDC. That means if at the expiry eth rises by 20%, you would get 120% of the USDC in collateral, therefore you had price exposure to eth, the same one as the collateral was in eth actually only paid in USDC instead of Eth. If at any point of the contract's life eth rises to 40% or more, contract would have to be immediately liquidated, to prevent the case that at the expiry you should have received 180% from the contract(eth rose 80%) but there is actually only 140% in the contract.

I think about it like this: Every derivative has two aspects.

1. Price exposure / Obligation => only dependent on the oracle price which is quoted and payoff function defined in contract logic. Stablecoins track price of usd, options have assymetric payoff. Whatever contract you can think of it has a payoff function(y axis == your payout, x axis == price of underlying), basically what it is worh at any price point of the underlying (in the video he is showing different payoff functions for different derivative products)
2. Delivery => this is basically the ability for the payoff function to be realized. You could do delivery without underlying if these are trusted parties, you promise your girlfriend you will "deliver" basically whatever she asks for :D. In traditional finance usually contracts are not fully backed and it is based on trust. In defi, there is no hope to get you an anonymous party on the internet to pay you your profit from your call option, therefore delivery is always depended on some bags of money, or eth or whatever else that is locked in the contract as a collateral to honor the agreement and guarantee payout. Any time the collateral is unable to cover the Obligation, the position is liquidated.

These two principles hold for all derivatives DAI, Call options, Range Bonds, ETc. Basically all derivatives are structurally the same in the essence and differ only in payout function.

[ckoopmann]

I believe the way the mechanism is currently designed, is that the price oracle always need to be a Token/USD pair and collateral in Token, could be wrong here.

I would expect that you can use any of the approved Price Feeds of UMA. Or is there a reason why it should be limited to Token/USD pairs ? | https://docs.umaproject.org/uma-tokenholders/approved-price-identifiers

[ckoopmann]

To your question, you would get price exposure if collateral is in USDC, what you might not get, is the actuall delivery ;), in practice contract would be liquidated before expiry because if price of eth goes up at a certain point collateral would not be able to fullfill the obligation of the contract.

Which contract would be liquidated ? The LSP / CFD contracts of UMA have no concept of liquidation as far as I understand. (Thats the main point what is so great about them, and the reason why the payouts have to be capped to the total collateral)

[markoinether]

I am talking specifically about CFD contract in the video. Price feed could be anything, but the collateral needs to be the Token from the price feed e.g. if UMA/USD collateral in UMA, if price feed x/USD collateral in x

[markoinether]

To your question, you would get price exposure if collateral is in USDC, what you might not get, is the actuall delivery ;), in practice contract would be liquidated before expiry because if price of eth goes up at a certain point collateral would not be able to fullfill the obligation of the contract.

Which contract would be liquidated ? The LSP / CFD contracts of UMA have no concept of liquidation as far as I understand. (Thats the main point what is so great about them, and the reason why the payouts have to be capped to the total collateral)

I think you are right that is basically why they keep collateral == to Token from the price feed, this allows you to always honor call option side of the contract. If UMA moons, collaterall moons too so it is always enough :)

[markoinether]

If there is no liquidation system here and you buy short tokens, your upside is limited, since the maximum amount you get is the full collateral, so if the price of token drops more than 50% you will not get the full upside you are entitled to

[ckoopmann]

If there is no liquidation system here and you buy short tokens, your upside is limited, since the maximum amount you get is the full collateral, so if the price of token drops more than 50% you will not get the full upside you are entitled to

Yes your upside on the LSP tokens is always limited to 1*collateralPerPair (a configuration parameter) tokens of the underlying collateral. That's part of the contract definition / payout function though, so I don't see why people would be "entitled" to more.

[ckoopmann]

I am talking specifically about CFD contract in the video. Price feed could be anything, but the collateral needs to be the Token from the price feed e.g. if UMA/USD collateral in UMA, if price feed x/USD collateral in x

Did they state that somewhere in the docs that the collateral has to be a token from the price feed ? Maybe that is true if you want to build this specific use case but generally I would expect you to be able to use any of the whitelisted collaterals.

Some price feeds are not token / blockchain related (like USD/GBP ) or not even actual "prices" at all. (Like SpaceX Launch Success Rate).

[markoinether]

that is basically a nomenclature, normally if you talk about put option it gives you the right to sell at certain price and the other side is obligated to buy at this price. here this is limited to the collateral, but think we have an understanding here... :)

[markoinether]

I am talking specifically about CFD contract in the video. Price feed could be anything, but the collateral needs to be the Token from the price feed e.g. if UMA/USD collateral in UMA, if price feed x/USD collateral in x

Did they state that somewhere in the docs that the collateral has to be a token from the price feed ? Maybe that is true if you want to build this specific use case but generally I would expect you to be able to use any of the whitelisted collaterals.

Some price feeds are not token / blockchain related (like USD/GBP ) or not even actual "prices" at all. (Like SpaceX Launch Success Rate).

I think it is in the video where he talks about CFDs specifically, but yes you are right collateral could be anything in general

[markoinether]

Technically it gets more complicated though, if you have heterogenius assets, you need two price oracles one to track the value of your position, the other the value of collateral.

[markoinether]

or you just get the full collateral in the end..

[markoinether]

I general to summarize my take on the examle:

1. Alice under normal conditions would get strictly less than 1 collateral for 1 long + 1 short if she is selling them on the market.
2. Given the first assertion holds, she would not be able to do the strategy described in your example, it's premise is, that she can get more than 1 from selling both positions.
3. This would make her necessarily opinionated, meaning she needs to decide what is her judgment of ETH price movement and sell only short token if she thinks the price will go up, higher than the strike so that her 1 long token will be worth 1collateral and she will make some extra money on the side from selling short tokens or vice versa.

That being said, I do not think we should be the Market maker (if we cannot be market neutral), but rather try to create a market place where there are both buyers and sellers that are motivated to trade with each other. I see markets as basically places where both sides can express their opposite opinions, for every buyer there is a seller, buyer thinks that eth is worth more than 2500 if he buys it for such amount and seller believes it is worth less if he is willing to sell it.

Also, with the contracts on ethereum in general, there is a problem of a gas cost, you do not want to open a position if you are not sure that there will be other party that will take you up on the offer because of the gass cost. Thinking that perhaps, people could just signal their intention to buy / sell by signing messages and token.approve and the position would be created only if there are two parties that match. Not sure if it is feasable, just an idea...

[snake-poison]

That would mean before expiry 1 short token is worth .5 WETH and 1 long token is worth .5 WETH to the CFD contract (that anyone can redeem from)

I think this assertion is wrong, I believe that before expiry you you can only "reduce" position, meaning you would have to provide 1 long + 1 short token to the contract to get 1 ETH back (could be fractions of the tokens too, but always in 1:1 ratio), this makes logically sense (and they explain it in the video like this) since this is how you reduce the position in a neutral way.

Counterexample: Since contract has no inherent pricing for short / long tokens before expiry you could theoritacally redeem 1 short token for 0.5 before expiry when there is 99% that the position will expire out of money... making the system unusable.

Thanks this was indeed a bad take by me. I agree. I don't think it impacts my line of thinking on pricing.

[snake-poison]

Alice knows the minimum she could take for the 1 short token in the money is a little more than .5 WETH

Why would you say that, if a token is just barely in the money and almost guaranteed to get not more than say 0.1 collateral tokens at expiry, why would anyone pay 0.5 weth for it ?

You are absolutely right. I will modify the post.

[snake-poison]

She needs to approach someone that is bearish on the price of WETH, but wants to acquire WETH. The buyer of the short tokens she is selling must receive the short tokens at discount, as there would be no upside to buying the tokens that are in the money.

This is only due to the fact that we are using WETH as a collateral in this case though right ? If we would use a USD stable coin as collateral this is not the case, I would think.

That's correct. I believe a way to lose that property on the short side but keep everything else the same would be to set the collateral per pair to be the current price of ETH with the collateral being a stable coin. What's cool with that setup (cash settlement) is that Troy and Rachel wouldn't be exposed to movements in price between expiry and the dispute period on the proposed price. If there is a dispute there could be a delay of up to 2 days. The dispute period even with no disputes is 2 hours I believe.