Introducing 5x Predictoor Boost in Volume Data Farming
Expect increased APYs for Volume DF
Summary
Due to a bug fix, the Volume DF OCEAN payouts shrunk. Previously, under the bug, payouts were about 20,000–25,000 OCEAN. DF81 payout (Mar 21, 2024) was the first time the bug fix was applied. The Volume DF payout for DF81 was 7,500 OCEAN. That is, a 2.67x-3.33x reduction. (APYs for Passive DF and Predictoor DF were not affected.)
While this was a legitimate bug fix, the drop in reward caused community expectations to be violated somewhat.
Many were upset. We get it.
To address this issue, we are proud to announce the 5x Predictoor Boost program. Practically speaking, it means that the Volume DF payout going forward is ≈37,500 OCEAN.
(And can rise with higher DCV, all the way to the 112,5000 OCEAN budget for Volume DF.)
Also, we have retroactively applied this to the DF81 payout by airdropping each account another 4x Volume DF rewards compared to what it got yesterday.
AMA coming soon!
Introduction / Contents of Document
The rest of this document describes DF budgets, then zooms in on Volume DF. It describes the “why” of a DCV bound on Volume DF rewards, what OCEAN rewards it led to, and how those numbers improved under Predictoor DF. It describes the bug and the fix, and how the Volume DF rewards changed due to the fix. It describes the issue of violating community expectations somewhat. Finally, it describes the fix: 5x Predictoor Boost, which restores Volume DF Rewards to ≈37,500 OCEAN.
DF Budgets
Overall DF budget. We are in Data Farming Main 2 phase, as Mar 14, 2024. From DF Main 1 to DF Main 2, the overall DF budget increased 2x.
Now, weekly rewards budget is 300,000 OCEAN (plus ROSE for Predictoor DF).
DF Budgets per Stream. DF is divided into streams. The “Passive DF” stream rewards people for locking OCEAN. The “Active DF” stream rewards doing more work to add value to ecosystem; it has two sub-streams “Volume DF” and “Predictoor DF”.
Here is reward allocation & some detail per each stream, as of DF Main 2.
- Passive DF. 150K OCEAN / week. Lock OCEAN for veOCEAN. The longer you lock for, the more veOCEAN (details below). Rewards are pro-rata to veOCEAN holdings.
- Active DF. 150K OCEAN / week. It has these substreams:
i. Volume DF: 112.5K OCEAN / week. Allocate veOCEAN towards data assets with high DCV, to earn more. Rewards are a function of DCV and veOCEAN stake.
ii. Predictoor DF: 37.5K OCEAN / week + 20K ROSE / week. Predict future price movement, continuously (as a predictoor).
For further details on each stream, please consult the DF docs.
About Volume DF
Volume DF is a substream of Ocean Data Farming. In it, users allocate veOCEAN towards data assets with high DCV, to earn more. Rewards are a function of DCV and veOCEAN stake.
Volume DF is not Active DF per se; it’s a subset of Active DF. Volume DF is not DF: there are several streams / substreams and it is one of them. Each stream is meaningful on its own.
The Wash-Consume Problem, and Fix
Early versions of Volume DF encountered a problem: wash consume. Let’s explain. Volume DF rewards on DCV. People found a way to game this: publish data assets, then buy & consume those assets themselves. A circular loop that drove up DCV. Volumes ballooned to $Ms.
This was not real volume, of course. And it is wasteful to reward OCEAN for this behavior. The aim is to reward OCEAN for things that truly drive value ot the ecosystem.
Here’s how we fixed wash consume: we ensured that rewards paid never exceed $ from fees. This works because fees themselves do not go into the pockets of the wash-consumers. Therefore anyone wash-consuming would lose $. (In Ocean, fees are baseline 0.1%; and baseline 20.1% for Predictoor feeds.)
The specific implementation is:
Volume DF rewards for asset and each veOCEAN holder, are bound by: asset’s
DCV* (% stake in asset) *DCV_multiplier.
The value DCV_multiplier is 20.1% for Predictoor feeds, and 0.1% otherwise. Call this the “DCV bounds” on the Volume DF rewards.
This also means that some of the OCEAN budgeted for Volume DF for a given week may not get deployed as rewards for that week. Any OCEAN not rewarded goes back into the DF pot for future DF rounds.
How much OCEAN is rewarded to Volume DF?
Before Predictoor was rolled out, DCV in Ocean was very small. And DCV_multiplier was 0.1%. So the DCV bounds (small DCV) * (small DCV_multiplier) = negligible number. Therefore OCEAN rewarded for Volume DF was negligible.
Then Predictoor & Predictoor DF was rolled out, which caused both larger DCV and larger DCV_multiplier of 20.1%.
Decent DCV. Predictoor came with Predictoor DF, where 37,500 OCEAN is spent as sales to buy Predictoor feeds. This is a “baseline” that incentivizes people to run Predictoor bots. In practice, it’s working extremely well, driving high activity and stake-to-predict volume. Predictoor feeds are the only meaningful contributor to DCV. So, DCV is basically the Predictoor DF rewards of 37,500 OCEAN.
Decent DCV_multiplier. Where other feeds had 0.1%, Predictoor feeds 20.1%.
Let’s combine those: DCV bounds (decent DCV) * (decent DCV_multiplier) = decent-ish number. So, OCEAN rewarded for Volume DF comes out decent-ish.
Let’s do exact numbers. We apply the formula above: 37,500 OCEAN volume, times DCV_multiplier=20.1% gives a bound of ≈7,500 OCEAN total. Therefore ≈7,500 OCEAN is practically available to Volume DFers.
The implementation described above was the intended behavior.
Discovery: a bug in DCV-bounds code
Thanks to decent DCV and decent DCV_multiplier in Predictoor DF, the Volume DF payout was supposed to be ≈7,500 OCEAN / week.
However the payout ended up between ≈20,000-25,000 OCEAN / week.
That is, 2.67x-3.33x more OCEAN was being paid out for Volume DF, compared to the intended design of Volume DF. (Remember, this is only a bounds based on DCV of 37,500 OCEAN/week. If DCV was high enough, the full budget of Volume DF would be paid out: 112,500 OCEAN/week.)
Something was up in Volume DF payout calculations. We dug deeper. As of Fri Mar 15, 2024, we discovered the specific bug in those calculations.
The bug had applied data consume volume (DCV) bounds too loosely, which over-rewarded each individual account during payout.
Recall that the DCV bound for each asset and each veOCEAN holder: asset’s DCV * (% stake in asset) * DCV_multiplier. Where “% stake in asset” is the amount of veOCEAN that a user has allocated, compared to the all veOCEAN allocated, for a given asset.
The bound was supposed to be on a per-asset, per-staker level.
But, the DCV bound code was written was only on a per-asset level. That was the bug. It missed the per-staker part. That is, “% stake in asset” was missing.
It meant that a user could create 10 separate accounts with 1/10 of her veOCEAN in each account, and get 10x more rewards compared to a single account. Yes, a Sybil attack. Anyone exploiting this flaw would be rewarded unfairly highly.
(To be clear: it wasn’t a security issue per se; there were no funds at risk. But it enabled people to game the system and get unfair rewards .)
Fixing the bug in DCV-bounds code
The core fix was to apply the DCV bound on a per-asset per-staker level. That is, to add the “% stake in asset” term.
We fixed the bug within hours of finding it. The first payout on the fixed code was for DF81, on Thu Mar 21. The DF81–82 blog post provides further information: in Section 5 (highly detailed); and with a complementary explainer in its Appendix.
The OCEAN budgeted for Volume DF never reduced. However, the amount paid out was reduced, because that payout is bounded by DCV, and DCV hasn’t been high enough to ignore this limit.
Bug Fix’s Effect on Volume DF Rewards
As expected, accounts doing Volume DF got reduced rewards in DF81 payouts.
Smaller stakers (those with a small % stake on an asset) had larger APY drops than larger stakers. This is because the same DCV bound was (wrongly) only at the per-asset level. Each account saw the same bound. After the bug fix, the APYs should be identical.
Let us illustrate.
- If an asset has just one staker, where % stake = 100%, their Volume DF payout for that asset doesn’t change at all, because the new term “% stake in asset” = 1.0. So introducing it doesn’t change the numbers.
- If a larger staker had 50% stake in an asset, their payout would be multiplied by 50%, ie a 1/0.5 = 2x reduction for the asset. Reduced APY for Volume DF follows.
- If a smaller staker had 10% stake in an asset, their payout would be multiplied by 10%, ie a 1/0.1 = 10x reduction for the asset. Reduced APY for Volume DF follows.
The above APY discussion is all for Volume DF. The APYs on Passive DF are independent; they’re at ≈20% these days. Passive DF APYs were not affected by this fix, since the fix is for Volume DF. Please don’t confuse Passive DF and Volume DF.
Bug fixed, but a second issue emerged
Like any professional software development team: we identified a bug; we fixed it asap; we deployed it; we communicated the changes.
However this caused a problem: people had locked OCEAN for years on the basis of seeing historical APYs in Passive DF and Volume DF. A long-term commitment which we appreciate.
Those expectations got violated somewhat, due to the reduced rewards for Volume DF.
Many were upset. We get it.
At the heart of the problem: Volume DF with the bug was ≈20,000–25,000 OCEAN / week. After the bug fix, payout was ≈7,500 OCEAN / week (ie payout of DF81). That is, a 2.67x-3.33x reduction. There were corresponding reductions in Volume DF APYs; those vary per-user as described above.
Towards addressing the second issue
How can we fix the problem of: expectations got violated somewhat, due to the reduced rewards for Volume DF?
To make a specific goal: can we get Volume DF rewards back to at least ≈20,000–25,000 OCEAN / week?
We have constraints:
- It doesn’t make sense to “undo” the bug fix, because then we’d still have the bug, with the issues it entails!
- We want to ensure that DF code implementation reflects the intent of its designers (us).
- As always, the design of DF and its sub-streams needs to remain flexible enough to keep evolving. All so that OCEAN rewards deployed most effectively drive traction in Ocean, and especially data consume volume (DCV).
- Related to the last point, some alpha: we do have other Data Farming sub-streams in the works! Those need breathing space. And, Predictoor is growing so the Predictoor DF weekly budget may increase too.
Introducing 5x Predictoor Boost
We are proud to announce the 5x Predictoor Boost program. It is a solution to the problem above.
Recall:
Volume DF rewards for asset and each veOCEAN holder, are bound by: asset’s
DCV* (% stake in asset) *DCV_multiplier.
Before, Predictoor feeds had DCV_multiplier = 20.1%. 5x Predictoor Boost gives a 5x to this multiplier.
With 5x Predictoor boost, Predictoor feeds get DCV_multiplier = 100.5%.
Let’s do exact numbers. We haveDCV of 37,500 OCEAN from (due to Predictoor DF), times DCV_multiplier=100.5% gives a bound of 37,687.5 OCEAN total (≈37,500 OCEAN). Therefore 37,687.5 OCEAN is practically available to Volume DFers.
(And as DCV rises, up to the whole Volume DF budget of 112,500 OCEAN becomes available to Volume DFers.)
The 5x boost only applies to the first 37.5K rewards given out on Volume DF. After that, the usual dcv_multiplier=20.1% applies. This helps to ensure breathing space for new DF substreams or increased Predictoor DF budget.
Implementation of 5x Predictoor Boost
What happens today:
- Retroactively apply this new formula to DF81 payout (which was yesterday — Thu Mar 21, 2024). In practice, we have airdropped each account another 4x Volume DF rewards compared to what it got yesterday. Here’s the tx.
What happens in before next DF payout:
- Update the DF payout code to apply the 5x Predictoor boost, for future weeks. (Applicable to first 37.5K rewards)
- Update documentation, etc as needed.
Conclusion
Due to a bug fix, the Volume DF OCEAN payouts shrunk. Previously, under the bug, payouts were about 20,000–25,000 OCEAN. DF81 payout (Mar 21, 2024) was the first time the bug fix was applied. The Volume DF payout for DF81 was 7,500 OCEAN. That is, a 2.67x-3.33x reduction. (APYs for Passive DF and Predictoor DF were not affected.)
While this was a legitimate bug fix, the drop in reward caused community expectations to be violated somewhat.
Many were upset. We get it.
To address this issue, we are proud to announce the 5x Predictoor Boost program. Practically speaking, it means that the Volume DF payout going forward is ≈37,500 OCEAN.
(And can rise with higher DCV, all the way to the 112,5000 OCEAN budget for Volume DF.)
Also, we have retroactively applied this to the DF81 payout by airdropped each account another 4x Volume DF rewards compared to what it got yesterday.
AMA soon!
About Ocean, DF and Predictoor
Ocean was founded to level the playing field for AI and data. Ocean tools enable people to privately & securely publish, exchange, and consume data. Follow Ocean on Twitter or TG, and chat in Discord.
In Ocean Predictoor, people run AI-powered prediction bots or trading bots on crypto price feeds to earn $. Follow Predictoor on Twitter.
Data Farming is Ocean’s incentives program. In DF, you can earn OCEAN rewards by locking OCEAN, curating data, and making predictions.
