Here are the Winners of the Predict-ETH Round 7 Data Challenge

$5000 in prizes to data scientists that predicted $ETH most accurately

Summary of Outcome

Predict-ETH Round 7 just completed. There were 252 submissions. We’re ready to announce the winners — drumroll, please!

• First place ($2500 in OCEAN) goes to 0xa98e504040b68e6a281fd489fcd1658df9703a91
• Second place ($1500 in OCEAN) goes to 0xca27c95f0eea6a0645dd7dddac46dfa893099fd6
• Third place ($1000 in OCEAN) goes to 0x9cb9a72299f715cbd58bcdf8b7fdc1107631114e

In this round, we were “full anon” so people don’t have to submit their names. That’s why you see addresses (who sent the txs), and that’s all.

Winners will be airdropped OCEAN into their addresses on Eth mainnet.

Each of the winning predictions showcases how to monetize data successfully using Ocean Protocol’s open source tools for the blockchain. Congratulations to all!

From Predict-ETH to _X_

For months, we’ve been quietly working to morph monthly Predict-ETH competitions into something more frequent and more automated. Therefore this is the last Predict-ETH competition. Stay tuned for an announcement in the next couple weeks about what this morphs into. And keep your prediction skills sharp:)

About Predict-ETH Competition

Predict-ETH is a recurring data challenge to predict the price of ETH and win prize money, using the ocean.py library. Accurately predicting ETH can be a valuable tool in buying and selling ETH, DeFi trading, yield farming, or DeFi protocol development. If your predictions are consistently on point, then you can sell them as a data feed on Ocean Market, earning you even more income.

Of course, predicting ETH is no walk in the park. It’s a complex task that requires skill, knowledge, and a bit of luck. But that’s exactly why Ocean Protocol hosts regular Predict-ETH competitions — to help you sharpen your machine learning skills, learn from others, and potentially make some cash along the way.

Predict-ETH Round 7 Setup

Predict-ETH Round 6 was announced in early June, 2023. The Round 7 README guided contestants how to make their submissions.

The submission deadline was Wed Jul 5, 2023 at 23:59 UTC. Competitors had to make predictions at times: Thu Jul 6, 2023 at 00:05 UTC, 00:10, …, 1:00 (12 predictions total).

If an address submits >1 entry: the youngest is kept, and rest get nmse 1.0.

Predict-ETH Round 7 Evaluation

The winner = whoever has lowest prediction error (normalized mean-squared error, or NMSE). That’s all. There is no subjectivity.

To be eligible, competitors must produce the outcomes that the README guides. This includes:

✅ Created an Ocean data NFT

✅ On the data NFT, set a value correctly: correct field label, correct # predictions, prediction values following correct formatting, predictions encrypted with proper encoding on judges’ public key

✅ Transferred data NFT to Ocean judges before the submission deadline

✅ All on Mumbai network, not another network

Predict-ETH Round 6 Statistics

The following table shows the NMSEs for the 25 (of 252) submissions having the lowest NMSE. If an address submitted >1 entry: the youngest was kept, and rest got nmse 1.0.

This round had some new contestant behavior. It appears that at least one contestant submitted >>1 entries across >>1 addresses. We discuss the implications below; let’s review numbers first.

Q: What is the number of submissions in previous rounds and now?

• Round 6 had 22 submissions; prior rounds had similar.
• Round 7 (this round) had 252 submissions, about 10x higher.

Q: How many unique addresses made submissions?

• In Round 6, 15 unique addresses had submissions. Two addresses had >1 submission: one with two submissions and one with five submissions.
• In Round 7, 83 unique addresses had submissions. 46 addresses had >1 submission, each with 2–6 submissions.

Q: how many contestants (unique humans) made these unique addresses? We can’t know for sure! But we can look for similar patterns of behavior across addresses. And we found several addresses clustering into similar behavior. For example, 15 addresses made submissions at 6 specific times: 14:56, 19:53, 18:04, 23:21, 23:52, and 23:56. Here are two such addresses:

Of the 252 submissions and 83 unique addresses, it appears that at least one contestant had >>1 submissions across >>1 addresses.

As submissions are anon-only, we don’t have a way to enforce just one submission per contestant (Sybil behavior). Correspondingly, the rules explicitly don’t forbid this. We’d been wondering when this would happen! It happened. And that’s ok. People probably wrote scripts to submit >>1 entries across >>1 addresses, ie basic automation.

This is good — we want the Ocean community to be thinking of how do automation in a prediction setting, because it’s a stepping stone to automation in trading, or automated defi protocols. And it jives with our own plans on automation in prediction and trading, including the future of Predict-ETH:)

Conclusion & Next Steps

Crypto price prediction can be an interesting and fun task — better yet, it can be monetized immediately!

Congratulations to all the winners and kudos to all participants for their entries.

The Predict-ETH competition is about to morph into something more frequent and automated, which we’ll announce it in the next couple weeks. Stay tuned and keep your prediction skills sharp:)

About Ocean Protocol

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 Protocol on Twitter or Telegram to keep up to date. Chat directly with the Ocean community on Discord. Or, track Ocean progress directly on GitHub.