Abstract - Twenty years ago Amazon.com was a humble online bookstore which was powered by the same servers and systems that other companies used. Amazon then developed a new way to power and reinvent its digital infrastructure when expanding beyond book sales, pushing it to become the most important e-commerce upgrade in history. In doing so it developed Amazon Web Services, as well as the entire concept of cloud computing as an industry.
“In the last quarter AWS accounted for 13 percent of Amazon’s overall revenue of $70 billion, and it is still growing faster than the company as a whole. Its operating income of $2.26 billion made up roughly 71 percent of Amazon’s overall operating profits.”
Just like Amazon all those years ago, Harvest.finance finds itself at a fork in the road.
“You have power over your mind, not outside events. Realize this, and you will find strength.” Marcus Aurelius
Thanks to the hard work of the Farm Team and Farm community many new Chads of the Farmland Kingdom have enjoyed remarkable gains throughout the last 8 weeks resulting in ample bread and over 1.1 $billion dollars in AUM deposited in yield bearing ftokens on Harvest. This defied my expectations.
The birth of yield farming alpha
Back in March, when Synthetix.io first launched an incentives program for liquidity providers to deposit sUSD on Curve through the iearn exchange protocol, it changed the industry forever.
Synthetix.io showed that when used correctly yield farming could be used to incentivise and grow the Synthetix.io community and user base rapidly, accelerating growth in network effects, as well as distribution and liquidity of the SNX token. This was copied and improved upon by compound.finance and many other innovative projects in the space. Now this yield farming trend seems to be reversing.
Misaligned yield farming Incentives (MYFI)
MYFI are leading to unfavorable outcomes and free falling APYs for YFarmers around the space. In order for Harvest to continue to grow and yield bountiful bread for the Farmers Collective (FC), Harvest needs to address and solve many of the misaligned incentives that exist between Developers, Users, Speculators, and Farmers in today’s Defi yield farming space.
Distorted tokenomics lead to unsustainable farming price premiums
Distorted tokenomics leads to massive price distortions and unfavorable outcomes for governance token holders due to strong liquidity mining rewards for early LPs, and due to the limited supply of coins right at the beginning of the distribution curve. I refer to this phenomenon as the early “Farming Price Premium” (FPP). It is a price premium on early minted coins. The question arises why a farmer would pay more for a coin at the beginning of the distribution curve than for a coin minted at the end of the distribution curve (all things remaining the same). The genesis coin can be used to acquire inflation dividends rights whilst circulation/competition for staking/farming the coin is low. The FPP leads to a massive early price spike and dump. The more inflationary and scarce a token is (circulating supply in relation to it’s max supply), the greater the corresponding FPP distortion will be and the subsequent pump & dump in the coin price.
As prices start to crash, loyal token holders bear significant losses except for the pump & dumpers that cause the crash. This is becoming more of a problem these days, thereby killing off the community vibes and good will amongst early adopters.
Both Curve.fi and YFI have built remarkable networks despite the coin price, market cap, and tokenomics being completely different.
YFI started low and rose over time as the project had underlying value - so does curve, but the tokenomics are misaligned. Therefore, all the farmers and early speculators of YFI made bank as the price rose in value from a couple of hundred dollars to 42k at one point. At the same time, CRV went from $50 at one point to $0.36…
No real value is created or transferred to the native projects in the current implementation of yield farming
Yield farming in its current implementation does not create any real value for the project or farmers. Instead, it introduces significant risks and Ponzi like tokenomics for all of the participants involved. It will simply cease to exist, just like the ICO boom ran out of steam in early 2018.
Real world example of a yield farm
Currently, $207million of idle cash is sitting in the BOND stable coin farming pool - an unaudited honey pot of sorts. These cash deposits are technically at risk of getting hacked.
As these funds are sitting idle, they are not being used to farm and to create any real value or yield for depositors, forgoing millions in potential alpha for LPs.
Barnbridge are not acquiring capital by distributing their gov token to farmers. In other words, Barnbridge is incurring several costs to try and build a network and distribute their native gov token in exchange for ‘proof of liquidity’.
It remains speculative how proof of liquidity will add significant value to a coin network by creating inflation & exposing underlying deposits to inflation and hackers. In addition, all the liquidity deposited is dependent on APYs remaining high. LPs will not remain loyal, as most will want to leave in search of higher alpha once yields fall for a sustained period of time.
There are several risks that need to be considered by every LP when depositing capital in unaudited farming pool:
Whilst funds are sitting idle in a farm, they are subject to inflation debasing your base capital – money printer goes bbbrrrrrr.
- Gas costs
Rising Gas prices are hurting all farmers, especially smaller farmers’ risk outlay when depositing in and withdrawing from yield farms.
- Onchain risk exposure - unaudited contracts, potentially vulnerabilities and back doors
It seems that time of launching unaudited yield farms and forks is coming to an end. Farmers have been taking on more risk through hidden minter keys and backdoors, whilst fighting over diminishing alpha. Returns usually dictate incentives and therefore have a large effect on investor decisions.
- Unstable and falling APYs
APY is linked to highly inflationary governance tokens rewards during the mining period despite claiming to be deflationary. Exposing farmers to a highly unstable source of alpha even if there is technically no risk of IL - which is the case with Barnbridge’s deposit contract - farmers still need to take on considerable risk to provide proof of liquidity in exchange for the native Gov token. In the majority of cases these tokens are given to LPs in exchange for taking incredible risks, considering that these governance tokens have no real value attached to them. For this reason, I personally decided against mining BOND even though I found the project of some interest and promise. The risk reward of depositing my capital in Barnbride’s unaudited yield farm did not make sense, given the audited alternatives available to me.
- There are many , many more risks to be considered when depositing in yield farms…
I just don’t want to write a book about it.
Compounding different protocol layers exponentially increases risk
The race for first movers’ advantage in the defi space is leading to several incomplete networks being launched in alpha or beta, thereby subjecting all network LPs to several risks. It is helpful to think of these as Guinea Pig networks, rather than test networks, as they are live on the Ethereum mainnet and have real value stored within them. Many of the new yield farming financial instruments in defi expose the LPs to exponential protocol risks because they stack on top of each other and interact with multiple defi protocols. For those who like to spin up large defi lego chains, it can become very risky very quickly.
For this reason more sets of good standards within defi are essential.
Barriers to Fork (BTF) fight off vampire attacks in the defi world
All true defi projects must be open source and decentralized in nature, so that code can be copied freely and at zero marginal cost. Many devs have used forked code to reduce security risks and speed up development. Defi projects that wish to compete long term need to build up barriers to fork (USPs). The more BTF a project has, the harder it is for other projects to vampire attack their network liquidity and its utility (i.e. Sushi on Uni). The best BTFs arguably are:
- Human Capital: devs and community
- Alpha wells: alpha sources belonging to a network (UNI token rewards)
- Distributed cloud server (FaaS) and custodial (Fort Knox) infrastructure
- Brands such as Apple & Coca Cola
- Anything that cannot be copied by code and is inherent to the value of a network
APY Cliff & Alpha Wells
TVL is a function of alpha that is generated on the platform (Uni, Crv)
Why does Uniswap have the largest TVL? Because they have the largest Alpha well. LPs swarm to Alpha wells like flies to shit. In a world where LP can move capital freely within the blockchain space, capital will always be attracted to alpha, wherever is it to be found. LPs are not loyal, uni found out the hard way. Most projects have ‘blown their load way too early’ (e.g. Party Chad). They ran out of their native governance tokens and with it their ability to sustain and develop long term relationships with LPs and users within their network. It was a good move of the Harvest Team and community to spread out Farm rewards over a longer distribution curve (i.e. 1 year +) avoiding a sudden APY cliff.
This has produced higher and more sustainable alpha for ftoken depositors through higher Farm APYs on base vaults deposits.
Where can I find the Farm token distribution curve picture?
Harvest has approximately 30 weeks before the APY cliff begins. This occurs when the majority of Farm token rewards are depleted. This needs to be addressed before it begins to hurt the humble FC.
Development costs, regulatory risks & potential liabilities
– ‘one mistake in the Dark Forest can cost you everything’
The crypto market moves faster than any other market, leaving no time to rest on your laurels if you want to stay ahead of the game. Most projects cannot just spin up multiple types of secure audited yield farms easily, quickly, and cheaply. Creating auxiliary infrastructure for one-time use is expensive, inefficient, and takes away focus and resources from more valuable activities. There is no point in being a ‘Jack of all trades’ in tech (cf. Yahoo). Even if a dev decided to launch a fork, making some short cuts here and there, they are still subject to considerable risk! Developer teams take on significant risks, costs, delays, and potential financial regulatory liabilities when they decide to build their own yield farms, such as:
- Server/infrastructure costs
- Security audits
- Regulatory risks & potential liability
- Maintenance & upgrades
- Marketing & community building
This section really deserves its own article but it’s a lot of work and it ain’t easy Sally…
However, why lose sleep over trying to compete with the FC, a hive mind network of LPs, continuously identifying, arbitraging and generating alpha to fund and deploy further infrastructure upgrades and dividends. “If you can’t beat them, join them” 1932. Senator James E. Watson
In other words, why build auxiliary infrastructure for only one-time use and incur unnecessary costs and hassle.
Launching a successful audited yield farm as efficiently as the Harvest team and community (FC) as a competitor will not be easy. Whilst many devs can launch custom servers, they rarely can do it in a more cost-efficiently manner than Amazon AWS and their existing economies of scale.
Harvest can save the yield farming industry
The FC has the ability and skills to disrupt the entire yield farming space. The actions and achievements of the FC over the last 8 weeks speaks for itself - and so does the $1.1bn in AUM. The FC needs to continuously innovate and address the many current problems (listed above) that are holding back the yield farming industry today. By setting new standards in efficiency, security, alpha, distribution and interoperability this can be achieved.
The FC can create sustainable sources of alpha on Harvest by establishing secure and sustainable funding channels and synergies for the most talented dev teams within the Harvest ecosystem. This was demonstrated when the Harvest team were the first to launch UNI farms, thereby capturing superior alpha for the FC relative to all their peers. Recently, Harvest elegantly migrated to new base vault contracts. This cannot be forked and greatly enhanced the brand and trust among LPs. Instead of a directional change, the FC now needs to double down and leverage this brand, trust, and skills, in order to expand the Harvest ecosystem.
Harvest.Finance’s first mover advantage will not last forever. In crypto there are always plenty of smaller competitors willing to take more risks and tradeoffs. If the FC keeps innovating like Amazon all those years ago, Harvest has the opportunity to reinvent the crypto industry for everyone.
“Never let the future disturb you. You will meet it, if you have to, with the same weapons of reason which today arm you against the present.” Marcus Aurelius
Harvests killer application is that it autonomously crops and locks in higher rates of returns for its farmers, than they would be able to obtain solo farming. Harvests vault leverages economies of pooled liquidity to create significant gas costs savings for its users. A convenient service that has attracted $1.2bn in AUM within 8 weeks from its launch, with no marketing budget (I am aware of). By implementing marginal compounding gains, we can make it, much much better!
If we created a vault strategy for FARM tokens, we could also create a yield bearing fFarm token, by using the Farm in the profit share pool as collateral to borrow and liquidity farm to boost exisiting yields further. Multiple mining strategies reducing slippage and dilution of APYs
Ftokens lock in APYs daily reducing price exposure to highly volatile tokens. The primary aim of the FM and AMM is to build upon this killer app by creating more utility (alpha) and infrastructure (liquidity rails) around ftokens. The proposal is not to propose a change in direction, but a doubling down of what the FC has been good at, generating alpha.
Inspired by the launch of the Farmers Market (FM) in week two, I will now propose an outline of the future evolution of the yield farming industry:
Farming as a Service (FaaS), the Farmer’s Market (FM) & Fort Knox (FK) -
A distributed cloud yield farming infrastructure for the defi age
“Farm and farming as a Service” will be the AWS of yield farming, allowing talented developers to obtain funding in exchange for providing Alpha on the Farmers Market (FM). FaaS will leverage and take advantage of existing network effects and distributed infrastructure by diversifying revenue streams and decreasing the custodial risks for the humble FC.
The FC and project partners have a unique opportunity to define a new set of yield farming standards in security, user experience, and interoperability, consequently reducing costs, risks and thus bringing longer lasting alpha to farmers who are seeking superior and safer risk adjusted yields.
With sufficient infrastructure upgrades (distributed farming cloud), it should be possible to spin up a yield farm in 20mins, just like you can an instance on AWS. Our clients would save time and money by leveraging the technology and infrastructure Harvest already has in place and continues to develop. all All the ftoken deposits will be held within Fort Knox whilst the humble farmers would be able to access additional yield on top of their base vault rewards without exposing themselves to any extra custodial risk.
“Your hard work is about to become easier with Harvest”
Fort Knox (FK) distributed custodial & cloud computing solution
Fort Knox enables Harvest to spread the cost of hardware, development, security and maintenance over a larger network of clients and users (Amazon did this with AWS), creating cost efficiencies and economies of scale. FaaS will be built on top of Fort Knox’s infrastructure, thereby allowing clients to leverage Harvest’s bespoke vault technology, network of LPs, and audited infrastructure, establishing long term relationships and real value for all participants. Remember, harvest is not a fork!
Quality over quantity
It does not make sense for several crypto projects to launch multiple different types of yield farms (honey pots) with inferior security, under par UX/UI and interoperability. If all these resources were funneled into Fort Knox, the FC could create ‘commercial grade enterprise software’. Harvest could become the custodial backbone or “The Humble Farmers’ Fort Knox” for not just our humble FC, but for every village, town, city, and kingdom within Farmland, i.e all the FaaS clients and their users. Most importantly, this allows LPs to boost APYs on top of their existing base rewards without exposing them to additional custodial and protocol risk. The importance of this has been highlighted by the recent attack on Harvest.
Farmers throughout Farmland will have options to add additional yield (on top vault base rewards) by staking their Farm & ftoken deposits within different pools on the FM. This additional yield will come in the form of the native governance token of partner projects, that have passed the FC due diligence requirements and community vote.
The Farmers Market (FM)
The way in which crypto projects have raised finance and distributed their native network token has changed considerably since Mastercoin was first launched.
The FM is the next evolution in capital distribution and project financing for distributed networks. All ftokens deposits staked on the FM are held within Fort Knox at all times. The aim of the FM is to fund and secure the very best developers and projects in the space to build within the Harvest ecosystem, whilst securing non-forkable sources of new Alpha for vault depositors. FaaS will allow promising projects to spin up pre-audited yield farming smart contract templates. This can be regarded as a basic AWS interface initially which allows clients to set up some variables within the smart contract (e.g. number of epochs, rewards, etc).
The FM will give humble farmers (LPs) an additional option to stake their Farm and ftokens to crop and speculate on alternative yields and sources of alpha. Alpha will come onto the FM in the form of different governance token projects, launching yield farms on the farming cloud (FC). They provide the humble farmers with bespoke yield farming opportunities that cannot be forked away, thereby increasing aggregate alpha drops on Harvest whilst accessing Harvests LP network.
Project due diligence
“FaaS is 100% dependent on the inherent value of the governance token we decide to list on the FM.” Super Chad 690420
If the FC replace Farm token rewards with inferior alternatives, ftokens will not move from the stake page to the FM. Projects that apply to list on the FM will have to go through a screening process, set out by the FC. Harvest needs to have a filter in place to prevent malicious listings that could undermine the Harvest brand.
The FM could list multiple different types of yield farms - not just type 1 and 2 pools - if the community agreed that they would add additional yield/alpha to the FM. Even if LPs did not stake on the FM ever, they would still benefit through the reverse dilution that occurs when ftokens are un-staked from the stake page and re-staked on the FM.
Pool 1 farms - Solving the idle funds problem: A real world FaaS case study with Barnbridge
According to my proposal, FaaS would allow new projects to directly attach their yield farms to Harvest’s base vaults (except for the profit share pool). This would allow farmers to generate a real return on their underlying stable coin and crypto deposits, whilst also enjoying the main APY incentive, the native governance token of the project listed on top of Harvest.
As mentioned earlier, I decided against mining BOND because of the extra inflation, gas costs, and custodial risks. If Barnbridge were to launch their USD pool on top of a Harvest Vault via FaaS and Fort Knox, that would change everything. And here is why….
Reducing custodial risks while boosting APYs for every farmer on Harvest
The FM would allow humble farmers to acquire new governance token rewards, boosting aggregate alpha dropped to Farm and ftoken stakers. Therefore, all farmers will benefit from APY boosts, regardless whether they plan to stake on the FM or not.
The base vaults rewards remain the same on the FM. In this case, Barnbridge just replaces the Farm reward incentive with the BOND reward. This enables farmers to generate a real return on the underlying deposits on app.barnbridge.com, whilst also receiving the BOND token reward.
The FM could provide many alternative sources of alpha beyond just farm rewards. The aim of my proposal is to get as many ftokens off the stake page and onto the FM as possible. This means, that all ftokens that a) are staked on the FM and/or b) held on Harvests own Automated Market Maker (AMM/DEX) will be paying 30% of their base rewards into the profit share pool via ftoken exposure. Without receiving any Farm token incentives, the FM reduces Harvest’s dependence on future Farm emissions to keep APYs high.
As explained above, the FM solves the idle fund problem by removing farmer’s pressure to sell their gov token rewards in order to shield their base capital against inflation and gas costs. All FM deposits are funneled directly into Fort Knox via ftokens deposits. This provides LPs and clients with direct exposure to all of the Harvest base vaults rewards (USDC, USDT, DAI, TUSD, WBTC, WETH, UNI Pairs etc.) through the FaaS profit share feature.
FaaS Profit Share
Profit share can tether real value against speculative governance tokens. This could partly solve the speculative value problem explained above, providing that the capital shared ends up in the ‘right hands’. Each FM listing could implement their own profit share terms. Projects that just want to use FaaS to save costs and distribute their native token can spin up yield farms without profit sharing, whilst projects that want to raise capital can raise the profit share percentage.
A profit share example:
- 50% of base vault profits will be distributed back to the LP, shielding them against inflation and gas fees.
- 30% of base vault profits will be deposited in the profit share pool as usual.
- The remaining 20% of base vault profits will be sent to an escrow contract that the FC has authority over. If projects that raise financing on the FM fail to meet objectives that have been agreed upon, the FC will return all funds held in the escrow deposit back to LP providers.
Profits could be used to add real value to governance tokens, establishing a less speculative investment proposal for LPs and a price floor for the token:
- Further protocol development.
- Add an insurance fund via Nexus Mutual.
- Yielding treasury vault that is tethered to ftokens assets (fUSD, fBTC, fETH, fFarm, fGold), thereby establishing a yield bearing price floor for the token.
Additionally, Harvest Finance would handle the custody (Fort Knox) of all assets tethered against ftokens, preventing rug pulls.
Harvest own AMM
Benchmark wrote a good technical proposal on how a Harvest AMM would look like under the hood - check it out.
As ‘everyone’s mother is launching a DEX/AMM right now’, it simply does not make sense to try and compete with UNI at this point. Frankly, it is embarrassing to suggest it. However, it is difficult to think of another way to implement FaaS and FM without one.Let me explain:
The primary function of our AMM is to build out liquidity rails for Farm and ftokens, not to compete with UNI on providing liquidity on WBTC – ETH!
The AMM can reduce slippage and gas fees for everyone by creating several fat direct liquidity rails for capital to enter and exit the Harvest ecosystem efficiently. This means, that it is easier and cheaper to unwind defi lego chains that contain ftokens or Farm within them.
The Harvest community and team have already adopted this strategy, in part by partnering up with Snowswap to provide liquidity rails between ftokens and yusd, ycrv etc.
It is possible to turn Farm and ftokens into an interest bearing ‘must have defi tool’ which allows you to use it as a springboard to farm more yield on top of the base vault rewards. You can only do this by creating ‘highly targeted liquidity bridges or save points’ from Farm and ftokens to more stable assets within cryptospace. I.e. direct paths into BTC,ETH, USDC & all of the best yield farming coins the FC decide to list. We want to avoid competing with UNI as much as possible.
Pool 2 farms - acquire additional yield staking liquidity pairs on the AMM
FaaS should allow new FM listings to launch liquidity pairs against their native gov token (pool 2 farms).
Newb: “Why would someone want to deposit Farm & ftokens in a DEX or on the FM when they are not getting any Farm rewards, idiot super chad!”
Because these pools will be paired against and/or rewarded with a native governance token, they provide the main APY incentive for farmers to stake LP pairs on the AMM and FM. Currently, these pools are incentivised by astronomical high APYs (paid in their native governance token) to attract LPs in order to buy and farm their gov token.
Even though LPs which farm pool 2 rewards also receive AMM trading fees, these AMM fees are minimal in relation to the main APY incentive. Therefore, I propose an adjustment to the distribution of AMM fees from the default UNI pricing model. Traders will incur a 0.3% trading fee using the Harvest AMM, and the fees will be split as follows (example):
- 0.05% goes to Farm token holders via the profit share pool.
- 0.05% goes to the project providing the gov token incentive for Pool 2 farms – which in turn would create additional revenue streams for Harvest clients.
- 0.20% goes to LPs of AMM.
AMM fees are often insignificant in relation to the Gov token incentive for Pool 2 farms. Consequently, a 0.05% drop-in trading fees for LPs will not have a material effect on their overall APY.
The FC should encourage new FM listings to create as many Farm and ftoken liquidity pairs.
Importance of having ftokens on AMMs
After the hack on 26 October, many humble farmers within the FC were calling to implement a 0.5% vault withdrawal fee. If the FC decide to go ahead with this, they need to provide other alternatives for LPs when entering and exiting the Harvest ecosystem. Improved ftoken liquidity rails/pairings will become increasingly important, as the AMM remove ftokens from the stake page, thereby creating another reverse dilution for all Harvest LPs. The USDC-Farm (15+ million in AUM) should be migrated to Harvest’s native AMM, in order to retain the 0.05% UNI trading fee within the Harvest ecosystem, whilst providing a nice base of liquidity to bootstrap the initial launch of the AMM.
Pool 3 (optional) - double mining pools & triple mining reward pools
Double mining pools should be optional and need only be set up by our clients. This prevents a conflicts of interest between Harvest and its FaaS and FM clients and partners.
Harvest have banned vampire protocols and will continue to ban other malicious actors trying to suck out liquidity from the ecosystem. However, even if Harvest ban all vampire protocols, many solo farmers will still want to lock in their APY. It does not matter how good someone’s project is, they cannot prevent LPs from selling their tokens. Knowing this fact, clients can spin up capped and/or ‘limited time only’ double mining pools on the FM. This allows the FC & clients to implement another 30% profit share - on top of base vault profits. Let me demonstrate…
Double mining pools will autonomously sell and lock in alpha for ftoken and Farm deposits (Uni & Crv) and additionally, autonomously farm and sell the gov token rewarded as the main APY incentive.
Newb: “Super Chad you idiot, why would a project want to create a vault on the FM, that farms and sell its own native token causing downwards price pressure”
30% of all double token rewards - rewards given out on the FM, not on the base vaults - would be converted into Farm, using the Farm AMM (Pool 2). This would continuously generate more volume and exchange fees for Farm token holders and LPs as rewards are harvested daily. Gov tokens harvested and sold for Farm could be split between Harvest and its FM partners:
- 30% will go into profit share - 9% of total profits
- 70% governance token project - 21% of total profits
This gives Harvest clients an additional revenue stream, share, and say within the Harvest ecosystem, baptizing many new humble farmers and developers, creating long-lasting synergies, network effects, and alpha for everyone involved.
These double mining pools can be used to effectively reduce and counterbalance inherent FPP (Farmers Price Premium) during the initial launch phase of a coin when the distribution curve and liquidity mining incentives start. This reduces the risk of a ‘pump and dump’ on early loyal project backers buying in at inflated prices due to larger FPP, thus, helping our clients to more accurately price their token at launch.
FaaS would allow Pool 3s to put a time and TVL cap in place for these vaults, so that they could just unwind them as soon as they wish.
Pool 0 farms – custom and debt pools
Pool 0 farms would encourage any humble farmer within the FC to suggest a new custom vault strategy. If voted in by the FC, these custom strategies will be listed as a base vault strategy (ftoken strategy) on the main page.
Debt based vaults allow LPs to leverage ftoken deposits to acquire additional yield
Debt based Vaults empower LPs to borrow/leverage their interest bearing Ftoken deposits.
Imagine humble farmers being able to borrow stablecoins against the WETH-BTC UNI LP Pair on Harvest? These stablecoins could be used acquire and to stake more LP tokens, in order to seek additional yield.
More leverage = more TVL = more profit share = more alpha
LPs underlying collateral could grow at a faster rate than their underlying debt payments/leverage. In effect, the returns of the vault could be used to pay off the debt over time, without ever having to repay or service the debt yourself! If farmers were able to use ftokens as collateral, farmers could create custom debt-based vault strategies. Think of it as a leveraged vault that pays itself off over time or borrows money when interest differentials can be arbitraged efficiently.