LogoLogo
TournamentsHedge FundDiscordGitHubForumDocs
  • Numerai Tournament
    • Overview
    • Data
    • Models
    • Submissions
      • Model Uploads
      • Compute Heavy
      • NGROK Example
      • CRON Example
    • Scoring
      • Definitions
      • Correlation (CORR)
      • Meta Model Contribution (MMC)
      • Feature Neutral Correlation (FNC)
      • Grandmasters & Seasons
    • Staking
    • Bounties
  • Numerai Signals
    • Overview
    • Data
    • Models
    • Submissions
    • Scoring
      • Definitions
    • Staking
    • Signals + QuantConnect
  • Numerai Crypto
    • Overview
    • Data
    • Submissions
    • Scoring
      • Definitions
    • Staking
  • Community
    • Discord
    • Forum
    • Twitter
    • Youtube
    • Apps & Content
      • Office Hours with Arbitrage
        • Office Hours Recaps: Season 1
          • OHwA S01E01
          • OHwA S01E02
          • OHwA S01E03
          • OHwA S01E04
          • OHwA S01E05
          • OHwA S01E06
          • OHwA S01E07
          • OHwA S01E08
          • OHwA S01E09
          • OHwA S01E10
          • OHwA S01E11
          • OHwA S01E12
        • Office Hours Recaps: Season 2
          • OHwA S02E01
          • OHwA S02E02
          • OHwA S02E03
          • OHwA S02E04
          • OHwA S02E05
          • OHwA S02E06
          • OHwA S02E07
          • OHwA S02E08
          • OHwA S02E09
          • OHwA S02E10
          • OHwA S02E11
        • Office Hours Recaps: Season 3
          • OHwA S03E01
          • OHwA S03E02
          • OHwA S03E03
          • OHwA S03E04
        • Office Hours Season 4
        • FAQ in Office Hours
        • Cited resources
      • Structure of Numerai by Wigglemuse
  • NMR
    • Coinbase
    • Uniswap
    • Etherscan
    • Github
  • Connect
    • Index
Powered by GitBook
On this page
  • Targets
  • Meta Models
  • Scores
  1. Numerai Signals
  2. Scoring

Definitions

PreviousScoringNextStaking

Last updated 4 months ago

Please refer to the to understand the basic functions referenced below.

Targets

  • target_20d

    • timeline: 20D2L

    • bins=5, uniformity=10%, 40%, 50%

    • neutralizers: Standard Factors and other factors not listed

  • target_20d_factor_neutral

    • timeline: 20D2L

    • bins=5, uniformity=10%, 40%, 50%

    • neutralizers: Standard Factors and other factors not listed

  • target_20d_factor_feat_neutral

    • timeline: 20D2L

    • bins=5, uniformity=10%, 40%, 50%

    • neutralizers: Standard Factors and other features not listed

Meta Models

  • Signals submissions are cleaned first by:

    • tie-kept-rank each submission

    • fill nans in each submission with 0.5

    • tie-kept-rank each submission

    • gaussianize each submission

  • Signals Stake-Weighted Meta Model w/ minimum stake (SSWMM)

    • Stake-weighted average of cleaned Signals submissions

  • Signals Naive-Weighted Meta Model w/ minimum stake (SNWMM)

    • Average of cleaned Signals submissions

  • Signals Naive-Weighted Meta Model w/ minimum 10 NMR stake (SNWMMmin10)

    • Average of cleaned Signals submissions with at least 10 NMR stake

Scores

  • submissions are cleaned before used in scoring:

    • drop invalid tickers

    • tie-kept rank the submission

    • fill nans with 0.5

  • MMC - Meta Model Contribution

    • correlation contribution of a submission, SNWMMmin10, and target_20d_factor_neutral

    • timeline: 20D2L (+ 2 days data lag)

  • CORRV4 - Correlation v4

    • numerai corr of a submission and target_20d_factor_feat_neutral

    • 20 score days w/ 2 days returns lag (+ 2 days data lag)

  • ICV2 - Information Coefficient V2

    • numerai corr between binned returns and submission

    • 20 score days w/ 2 days returns lag (+ 2 days data lag)

  • RIC - Residual Information Coefficient

    • numerai corr between target_20d_factor_neutral and submission

    • 20 score days w/ 2 days returns lag (+ 2 days data lag)

  • FNCV4 - Feature Neutral Correlation v4

    • submission is tie-kept-ranked then gaussianized then neutralized

    • tie-broken-rank correlation between target_20d_factor_feat_neutral and submission

    • neutralizers: Standard Factors and V4 Medium Safe Features

    • 20 score days w/ 2 days returns lag (+ 2 days data lag)

  • CWSNMM - Corr w/ Signals Naive Meta Model

    • s` = tie-kept rank, then gaussianize, then pow 1.5 a submission s

    • calculate pearson correlation between s` and SNWMM

    • timeline: 4 days data lag / not dependant on returns

  • MCWSM - Max Corr w/ Signals Models

    • Maximum pearson correlation of a submission with any other Signals submission

    • only compared to other submissions made in the same round

    • timeline: 4 days data lag / not dependant on returns

  • APCWSM - Average Pairwise Corr w/ Signals Models

    • Average pearson correlation of a submission with each other Signals submission

    • only compared to other submissions made in the same round

    • timeline: 4 days data lag / not dependant on returns

main definitions docs