NFT Experiment #2 - You could stake on my predictions for Round 257

See the previous forum post for more introduction about buying NFTs of my predictions: NFT Experiment -- You could stake on my predictions for Round 256

For sale this week:
HB_FALCON | Current Rank 3 (last week 4)
HB_CLIPPER | Current Rank 43 (last week 90, with highest 3M return for all of Numerai)
HB_NOVA | Current Rank 404 (last week 719, a newer model since Round 241)
HB | Current Rank 378 (last week 404, OG model and formerly #1 on Leaderboard)


The only thing that I’ll repeat from last week’s post are the reasons not to buy:

  • You can lose up to 25% of your stake without knowing anything about the underlying model.
  • Gas fees alone will likely be >$25.
  • This is not a subscription; you only get predictions for one new round. If you want to use them, you must submit before Round 257 closes on Monday March 29 at 14:30 UTC.

Who am I? Watch my OHwA interview here.


Interesting results so far from Experiment #1 last week:

  • All 4 predictions were purchased:
    • HB_FALCON for 0.103 ETH
    • HB_NOVA for 0.0724 ETH
    • HB_CLIPPER for 0.103 ETH
    • HB for 0.025 ETH
  • No one purchased anything until after I announced on Twitter that the R256 submission files were in the Dropbox on Saturday. I’m not sure if that’s just because the price was too high still or if there was skepticism that I would follow through.
  • None of the predictions were resold or even put back on the market by the buyers.
  • Two of the predictions were purchased by the same ETH address.
  • HB was purchased after R256 was closed. I hope that was on purpose. I intended to have the auctions close with a little buffer until the end of the round so that people didn’t inadvertently buy too late, but I didn’t account for the daylight savings time change!
    • If you bought HB, intended to stake, but didn’t realize that the round had closed, just submit it late anyway. I’ll verify the Corr is the same as my HB model by Saturday and send you your ETH back.
  • I’ll see how much was staked on each model by finding the matching Corr’s to my models on Thursday (or maybe it will take a couple of days to rule out any other models that happen to get nearly identical Corr).
    • Because we’re getting close to the staking cap, the incentives to sell these predictions might diminish if the new stakers are pushing us into the steepest part of the payout curve decline. Note to potential buyers: I plan on reevaluating whether I sell predictions at all once we reach the staking cap which might only be a few weeks away. So this might be a somewhat short-lived experiment.

Changes for Experiment #2 relative to last week:

  • Starting price will be lowered to 0.25 ETH, dropping to 0.025 ETH by Monday March 29th.
    • I think one of the cool capabilities is the option for a buyer to resell, recoup some of their purchase price, and hopefully expand the size and reach of the market. Especially if a buyer has a bigger sway than I do on social media, additional re-sellers could change things dramatically. Hopefully a lower starting price will encourage buyers to purchase a few days earlier and have the opportunity to resell.
  • Auctions will end Monday March 29th at 1200 UTC (5:00 AM PDT) to ensure that there’s at least a couple hours for a potential buyer to submit and stake (even when I forget about daylight savings).
2 Likes

Very interesting. One worry is any potential Numerai Terms of Service issues for the buyer. Do we know if this has been clarified?

I looped in Numerai before I began the first experiment and they didn’t mind.

All predictions for Round 258 are posted in the NFTs in my OpeaSea collection!