{
  "top/10": {
    "cs_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 0.8370121976925251,
      "interval_95": [
        0.6246343180481755,
        1.0315387473051645
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "gold_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 30.335799066993673,
      "interval_95": [
        23.859855014806207,
        36.814210686536796
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "xp_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 6.7106726967529085,
      "interval_95": [
        2.9101422308614615,
        10.306729349566789
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  },
  "top/15": {
    "cs_difference": {
      "rows": 2371,
      "series": 937,
      "metric": "rmse",
      "improvement": 0.6231285531097264,
      "interval_95": [
        0.34151005158871306,
        0.8826687576918386
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "gold_difference": {
      "rows": 2099,
      "series": 908,
      "metric": "rmse",
      "improvement": 40.14451380978562,
      "interval_95": [
        28.740499607254733,
        52.18472609987942
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "xp_difference": {
      "rows": 2276,
      "series": 931,
      "metric": "rmse",
      "improvement": 6.6083784027589445,
      "interval_95": [
        0.9294338929732449,
        11.972229375676273
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  },
  "mid/10": {
    "cs_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 0.6371689069935798,
      "interval_95": [
        0.5183424890036197,
        0.7523889801739917
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "gold_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 7.637765326784688,
      "interval_95": [
        4.2943116341477445,
        11.07759096619709
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "xp_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 4.693690803630659,
      "interval_95": [
        1.9527229694160113,
        7.381268324338621
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  },
  "mid/15": {
    "cs_difference": {
      "rows": 2369,
      "series": 938,
      "metric": "rmse",
      "improvement": 0.7837136766263821,
      "interval_95": [
        0.6467431656834168,
        0.9222816480665147
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "gold_difference": {
      "rows": 2100,
      "series": 907,
      "metric": "rmse",
      "improvement": 9.364093973686181,
      "interval_95": [
        5.189214764529166,
        13.728985295456122
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "xp_difference": {
      "rows": 2265,
      "series": 926,
      "metric": "rmse",
      "improvement": 6.329887324969263,
      "interval_95": [
        2.7269128293544567,
        10.141001985332117
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  },
  "bot_duo/10": {
    "bot_cs_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 0.19646249842904595,
      "interval_95": [
        0.030541581381059127,
        0.37068269400134857
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "duo_gold_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 9.935522827351406,
      "interval_95": [
        6.018035931403157,
        14.058630362192648
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "duo_xp_difference": {
      "rows": 2388,
      "series": 940,
      "metric": "rmse",
      "improvement": 5.726808337798957,
      "interval_95": [
        3.614867476456067,
        7.977947678168421
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  },
  "bot_duo/15": {
    "bot_cs_difference": {
      "rows": 2383,
      "series": 940,
      "metric": "rmse",
      "improvement": 0.0,
      "interval_95": [
        0.0,
        0.0
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "duo_gold_difference": {
      "rows": 2007,
      "series": 897,
      "metric": "rmse",
      "improvement": 0.0,
      "interval_95": [
        0.0,
        0.0
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    },
    "duo_xp_difference": {
      "rows": 2272,
      "series": 929,
      "metric": "rmse",
      "improvement": 0.0,
      "interval_95": [
        0.0,
        0.0
      ],
      "positive_means": "candidate has lower error",
      "scope": "Conditional fixed-model held-series performance difference, not coefficient confidence or post-selection coverage"
    }
  }
}