argus
Allora mainnet · live topics

Topics

15 live of 24 slots · data as of 21:07 UTC

Each topic is one question models compete to answer: an asset, a target, a horizon.

Where a model still gets paid
A topic pays only its top scorers. Past that cap a new entrant earns nothing, however well it does.
11 / 15
topics with model slots open
0 / 15
with forecaster slots open
Rounds / day
3,751
Model slots filled
262
one model can fill several
Most contested
32
BTC/USD Volatility 15m
Best per model
1.41%
BTC/USD Log Returns 8h
TopicAnswerLast 48ModelsSharePer modelCadence
SOL/USD15m#23σ 0.0056732/ 327.33%0.229%~5.5 min
BTC/USD8h#14$58,8117/ 327.29%1.042%~5.5 min
ETH/USD15m#21σ 0.0056832/ 327.29%0.228%~5.5 min
XRP/USD15m#22σ 0.0058832/ 327.28%0.228%~5.5 min
BTC/USD15m#20σ 0.0042132/ 327.26%0.227%~5.5 min
ETH/USD8h#9$1,607.9317/ 327.17%0.421%~5.5 min
SOL/USD8h#10$67.118/ 327.16%0.398%~5.5 min
BTC/USD8h#1-0.009%5/ 327.03%1.405%~6.9 min
ETH/USD24h#16-1.644%6/ 327.02%1.170%~5.5 min
NEAR/USD8h#19+1.184%11/ 327.02%0.638%~5.5 min
BTC/USD24h#15-0.782%8/ 327.01%0.876%~5.5 min
SOL/USD24h#17-3.518%10/ 326.97%0.697%~5.5 min
ETH/USD8h#2+0.551%17/ 326.92%0.407%~6.9 min
SOL/USD8h#3-0.937%13/ 326.92%0.532%~6.9 min
BTC/USD20m#18-0.191%22/ 320.33%0.015%~5.5 min
How a round works
1Submit
Models send predictions inside the round's submission window.
2Combine
Forecasters rate each model; the network blends it all into one inference.
3Score
Reputers fetch what actually happened and score everyone against it.
4Pay
Rewards flow by topic weight, then by score.
Reward map · pool share vs models competingsorted by share per model, best first
share of rewardsmodelsper model
BTC/USD Log Returns8h51.405%
ETH/USD Log Returns24h61.170%
BTC/USD Price Prediction8h71.042%
BTC/USD Log Returns24h80.876%
SOL/USD Log Returns24h100.697%
NEAR/USD Log Returns8h110.638%
SOL/USD Log Returns8h130.532%
ETH/USD Price Prediction8h170.421%
ETH/USD Log Returns8h170.407%
SOL/USD Price Prediction8h180.398%
SOL/USD Volatility15m320.229%
ETH/USD Volatility15m320.228%
XRP/USD Volatility15m320.228%
BTC/USD Volatility15m320.227%
BTC/USD Log Returns20m220.015%
BTC/USD Log Returns 8h: 1.41% of the pool per model · BTC/USD Log Returns 20m: 0.015% · a 95x gap, not an earnings forecast: pay also depends on score

Reward share is the topic's slice of emissions; per model divides it by the models competing. Pay also depends on score. Measured by our collector, hourly.

The scoring study

Does the chain's ranking predict who is right?

Allora scores every model and pays by that score. For each settled round we asked one question: did the model the chain rated highest at that moment land closest to what actually happened? Against picking blind, the answer is yes on 14 of 14 topics.

Top-rated model won the roundone observation per round · scores as published at that height
TopicLossRoundsModelsWonChanceLiftVerdict
ETH/USD Log Returns 24hczar23,7195.933.2%17.1%+16.1 ppclear
BTC/USD Log Returns 8hczar18,9825.036.1%20.2%+15.9 ppclear
BTC/USD Volatility 15mmse23,59113.628.6%13.4%+15.2 ppclear
BTC/USD Price Prediction 8hmse23,6845.233.5%19.4%+14.1 ppclear
SOL/USD Log Returns 24hczar23,6749.924.1%10.2%+14.0 ppclear
ETH/USD Volatility 15mmse23,58714.823.4%11.2%+12.1 ppclear
XRP/USD Volatility 15mmse23,57815.922.3%10.7%+11.6 ppclear
ETH/USD Log Returns 8hczar18,94911.718.6%8.7%+9.8 ppclear
SOL/USD Log Returns 8hczar18,94512.817.3%7.8%+9.4 ppclear
SOL/USD Volatility 15mmse23,59016.917.8%9.7%+8.1 ppclear
BTC/USD Log Returns 24hczar23,6847.920.7%12.7%+8.0 ppclear
BTC/USD Log Returns 20mczar23,62921.411.8%4.7%+7.1 ppclear
ETH/USD Price Prediction 8hmse23,65011.415.3%8.9%+6.4 ppclear
SOL/USD Price Prediction 8hmse23,64614.012.7%7.2%+5.5 ppclear

Won is how often the chain's top-rated model finished closest to what happened; chance is the mean of 1 in k over the same rounds; clear means the lift is at least two standard errors. We score by absolute error against our own reconstruction, the chain scores by CZAR, MSE or ZPTAE, so some disagreement is by design. Volatility truth (topics 20 to 23) depends on the estimator, so those rows are directional. Never pooled across topics.