ETH OSM Oracle Reward Adjustment

ETH OSM Oracle Reward Adjustment

As our ecosystem evolves, it is essential to continuously assess and optimize our protocol to ensure its efficiency and sustainability. This proposal aims to reduce the rewards paid to keepers, thereby achieving a more cost-effective and sustainable system for maintaining the accuracy of the ETH price oracle.


This post and forum proposal is a continuation of the thematic discussions around protocol income and expenses we’ve had here: LINK

The core DAO team, with engagement from the broader DAO community, continues to investigate how to best optimize these flows. This process formally kicked off last year with Fabio’s fantastic work on optimizing keeper rewards broadly, and more recently with the DAO’s exciting vote to reduce the Safe stability fee to just 0.5%.

The Reflexer Finance protocol relies on an oracle system to obtain accurate and up-to-date price information for the ETH market. Obviously, this is used to value Safe collateral, but the protocol’s debt, RAI, is priced in terms of ETH (via a 24hr TWAP in the Uni v2 RAI-ETH pool) and then translated to dollars. Good oracle data, and thus keepers, play an essential role to keep the protocol orderly. In return for the efforts of these updates, keepers receive rewards paid in RAI from the protocol’s specific sleeve to fund these matters.

However, it’s the opinion of members of the core DAO team that the current reward structure may be unnecessarily high, resulting in overpayment for ETH price updates. Historically, ETH OSM price updates cost the protocol’s balance sheet ~2k RAI per month, representing ~1/3 of all payments.

Proposal: Reduce Keeper Rewards

I propose that we reduce the rewards paid to keepers for updating the ETH price oracle. The goal of this reduction is to find a balanced approach that:

  1. Continues to incentivize keepers to contribute accurate and timely price data, especially during times of volatility
  2. Reduces token payments to keepers, benefitting the protocol’s long-term health and solvency

To achieve this, I propose the following adjustments to the keeper rewards structure:

  1. Increase the timing delay of rewards for the ETH OSM update from 1 hour to 4 hours. (e.g., reward can be paid no more frequently than four hours)


There should be a sober discussion about the drawbacks of potentially stale ETH price data. I think the most important issue is obviously how this could affect liquidations in the event of a large volatility event.

First, should there be a safe that would be eligible for liquidation should the oracle be updated, there exists an incentive for a liquidator to update the oracle even if the cost of updating the oracle exceeds the benefit paid by a keeper, since there is a profit to be made from orderly liquidations.

Further, the protocol already has in place numerous protections against bad debt, including a delay system and a significant chunk of RAI on the protocol’s balance sheet to act as a liquidator of last resort.

I think readers should agree that 99%+ of the time, the ETH oracle update is fairly useless and that the protocol or its users materially benefit from its “currency”, but when you need a good oracle you really need a good oracle. And, I believe that this change doesn’t encroach on security nearly as much as it saves protocol expense.

Next Steps

Please share your thoughts, concerns, and suggestions in the comment section below. Ask questions of our dev team. Let’s make sure this is a practical idea before this goes to DAO vote. Your input is invaluable in shaping the future of Reflexer Finance.


Thank you, @0x-kingfish for opening the path to yet another possible optimization of production costs!

could we have 2 timing delay states : one for normal operation (increased from 1 to 4 hrs) and one for “extreme” market conditions (1 hr as it is now).?

Great question. So we’ve actually thought about this, and the solution is kinda the same thing that happens already.

During periods of high volatility, the pricing of ETH can change dramatically, right?. This change is already not bound by our hourly rewards scheme. Liquidators can (and I believe have) call the oracle update themselves (without keeper fee renumeration) in order to begin processing a safe under the threshold.

The idea is this: 99.95% of the time, the oracle update is useless. It doesn’t actually affect anyone because it does not induce a new behavior from any system participants.

That last 00.05% of the time, however, is critically important. We believe in that small percentage of time, liquidators are going to pay to update the oracle (with or without keeper bribe) in order to push bad safes through the system and (in their hope) receive liquidation rewards for doing so.


So, regarding the system’s most expensive oracle,(ETH OSM) there could be a drop from roughly 33% of current costs to less than 10% if this gets up voted🗳 and still have a safe mechanism in place (so, with minimal disruption on RAI system).

This makes the 0.5% stability fee way more sustainable!

To me, that’s good news any day of the week, ser :handshake:

1 Like

voting happening here