# Latest **URL:** https://research.allora.network/latest.md [Latest](https://research.allora.network/latest.md) · [Categories](https://research.allora.network/categories.md) · [Tags](https://research.allora.network/tags.md) --- ## [Read this before posting](https://research.allora.network/t/read-this-before-posting/5)
**Author:** [@system](https://research.allora.network/u/system)\ **Replies:** 0\ **Last updated:** [April 29, 2024, 8:39pm UTC](https://research.allora.network/t/read-this-before-posting/5 "2024-04-29T20:39:05Z")
We are so glad you joined the Allora Research community. Allora Research An semi-public forum for Allora research This is a semi-public forum for participating in Allora’s research efforts, including but not limited … --- ## [Paper "Flawed in Nature, Perfect through Evolution"](https://research.allora.network/t/paper-flawed-in-nature-perfect-through-evolution/158)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 1\ **Last updated:** [September 2, 2026, 1:03pm UTC](https://research.allora.network/t/paper-flawed-in-nature-perfect-through-evolution/158 "2026-09-02T13:03:30Z")
Thread for discussion of Flawed in Nature, Perfect through Evolution. Flawed in Nature, Perfect through Evolution ADI 3, 1-32; September 2, 2026 J. M. Diederik Kruijssen The performance of artificial intelligence (AI)… --- ## [Inference worker promotion and relegation](https://research.allora.network/t/inference-worker-promotion-and-relegation/157)
**Author:** [@joel](https://research.allora.network/u/joel)\ **Replies:** 2\ **Last updated:** [August 7, 2026, 1:53am UTC](https://research.allora.network/t/inference-worker-promotion-and-relegation/157 "2026-08-07T01:53:40Z")
In this post I describe the framework for a promotion/relegation system, where inference workers on the Allora Network can be promoted to and from mainnet topics. This will be the intended pathway for workers to demonstr… --- ## [Research Monitoring Suite: How We Watch the Allora Network](https://research.allora.network/t/research-monitoring-suite-how-we-watch-the-allora-network/156)
**Author:** [@florian](https://research.allora.network/u/florian)\ **Replies:** 0\ **Last updated:** [May 3, 2026, 11:45am UTC](https://research.allora.network/t/research-monitoring-suite-how-we-watch-the-allora-network/156 "2026-05-03T11:45:56Z")
In this post I want to showcase the Research Monitoring Suite (RMS) — our internal infrastructure for understanding what’s actually happening on the Allora Network. From validating theoretical models to catching subtle b… --- ## [Paper "Context-Aware Inference via Performance Forecasting in Decentralized Learning Networks"](https://research.allora.network/t/paper-context-aware-inference-via-performance-forecasting-in-decentralized-learning-networks/150)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 1\ **Last updated:** [March 27, 2026, 12:12pm UTC](https://research.allora.network/t/paper-context-aware-inference-via-performance-forecasting-in-decentralized-learning-networks/150 "2026-03-27T12:12:01Z")
Thread for discussion of Context-Aware Inference via Performance Forecasting in Decentralized Learning Networks. Optimizing Decentralized Online Learning for Supervised Regression and Classification Problems ADI 2, 40-5… --- ## [CZAR loss function for returns prediction topics](https://research.allora.network/t/czar-loss-function-for-returns-prediction-topics/155)
**Author:** [@joel](https://research.allora.network/u/joel)\ **Replies:** 4\ **Last updated:** [February 9, 2026, 2:38am UTC](https://research.allora.network/t/czar-loss-function-for-returns-prediction-topics/155 "2026-02-09T02:38:51Z")
The CZAR (Composite Zero-Agnostic Return) loss function is designed to address limitations observed in previous loss formulations used for inference synthesis in returns prediction topics. This post documents the motivat… --- ## [Handling ground truth granularity](https://research.allora.network/t/handling-ground-truth-granularity/116)
**Author:** [@florian](https://research.allora.network/u/florian)\ **Replies:** 4\ **Last updated:** [November 4, 2025, 6:27pm UTC](https://research.allora.network/t/handling-ground-truth-granularity/116 "2025-11-04T18:27:56Z")
Ground truth data is typically delivered at a fixed granularity, depending on fundamental features of the topic or on the data provider. This means that the timing of data points may be aligned to a specific resolution, … --- ## [Losses in returns prediction topics](https://research.allora.network/t/losses-in-returns-prediction-topics/120)
**Author:** [@joel](https://research.allora.network/u/joel)\ **Replies:** 8\ **Last updated:** [November 4, 2025, 10:12am UTC](https://research.allora.network/t/losses-in-returns-prediction-topics/120 "2025-11-04T10:12:43Z")
I’ve noticed some strange behaviour of the losses in topics predicting log-returns. Some workers occasionally provide extremely large inferences (up to ~ 10^12) compared to typical returns values (\<0.1), but still have r… --- ## [Feature engineering experiments 1: add log-returns-focused features](https://research.allora.network/t/feature-engineering-experiments-1-add-log-returns-focused-features/134)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 8\ **Last updated:** [October 1, 2025, 5:59pm UTC](https://research.allora.network/t/feature-engineering-experiments-1-add-log-returns-focused-features/134 "2025-10-01T17:59:09Z")
In this experiment, we test the impact of adding a returns-focused feature set (all quantities you can calculate for price, but for log-returns). This is a relatively small amount of work (applying the transformations yo… --- ## [Price/returns topic feature engineering](https://research.allora.network/t/price-returns-topic-feature-engineering/123)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 14\ **Last updated:** [July 17, 2025, 12:18pm UTC](https://research.allora.network/t/price-returns-topic-feature-engineering/123 "2025-07-17T12:18:08Z")
Looking at participant performance in Allora’s Forge programme, I have the impression that feature engineering is the bottleneck in several of the current price/return prediction topics. Markets are like physical system… --- ## [Thorough testing of new forecaster model](https://research.allora.network/t/thorough-testing-of-new-forecaster-model/117)
**Author:** [@joel](https://research.allora.network/u/joel)\ **Replies:** 9\ **Last updated:** [May 29, 2025, 11:55am UTC](https://research.allora.network/t/thorough-testing-of-new-forecaster-model/117 "2025-05-29T11:55:49Z")
Forecasting is the component that enables context-awareness in the Allora Network. During inference synthesis, worker inferences are combined using their historical performance (through an exponential moving average of t… --- ## [Paper "Deterministic AI Agent Personality Expression through Standard Psychological Diagnostics"](https://research.allora.network/t/paper-deterministic-ai-agent-personality-expression-through-standard-psychological-diagnostics/114)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 0\ **Last updated:** [March 24, 2025, 1:18pm UTC](https://research.allora.network/t/paper-deterministic-ai-agent-personality-expression-through-standard-psychological-diagnostics/114 "2025-03-24T13:18:26Z")
Thread for discussion of Deterministic AI Agent Personality Expression through Standard Psychological Diagnostics. Optimizing Decentralized Online Learning for Supervised Regression and Classification Problems ADI 2, 15… --- ## [Extend the Allora network to anomaly detection tasks](https://research.allora.network/t/extend-the-allora-network-to-anomaly-detection-tasks/104)
**Author:** [@florian](https://research.allora.network/u/florian)\ **Replies:** 3\ **Last updated:** [March 12, 2025, 10:01am UTC](https://research.allora.network/t/extend-the-allora-network-to-anomaly-detection-tasks/104 "2025-03-12T10:01:08Z")
Similarly to the classification extension, we want to extend the Allora network to another class of topics: anomaly detection. In anomaly detection topics, consumers supply their own data to the network, and then the net… --- ## [Paper "Optimizing Decentralized Online Learning for Supervised Regression and Classification Problems"](https://research.allora.network/t/paper-optimizing-decentralized-online-learning-for-supervised-regression-and-classification-problems/105)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 0\ **Last updated:** [January 27, 2025, 9:07pm UTC](https://research.allora.network/t/paper-optimizing-decentralized-online-learning-for-supervised-regression-and-classification-problems/105 "2025-01-27T21:07:05Z")
Thread for discussion of Optimizing Decentralized Online Learning for Supervised Regression and Classification Problems. Optimizing Decentralized Online Learning for Supervised Regression and Classification Problems ADI… --- ## [Hybrid ML models](https://research.allora.network/t/hybrid-ml-models/103)
**Author:** [@Ryouhei-Deworkhub](https://research.allora.network/u/Ryouhei-Deworkhub)\ **Replies:** 1\ **Last updated:** [January 13, 2025, 2:02pm UTC](https://research.allora.network/t/hybrid-ml-models/103 "2025-01-13T14:02:15Z")
We aim to identify an effective approach to constructing robust predictive models. It is well-recognized in the academic community that, in addition to single models, hybrid models have also garnered significant attentio… --- ## [Standard forecaster model](https://research.allora.network/t/standard-forecaster-model/37)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 9\ **Last updated:** [November 27, 2024, 4:40pm UTC](https://research.allora.network/t/standard-forecaster-model/37 "2024-11-27T16:40:22Z")
We will need to develop an “off-the-shelf” forecaster model that uses a combination of: network data (e.g. historical inference losses, raw inferences, network inferences, worker scores, worker rewards); private data (… --- ## [Outlier-resistant network inferences](https://research.allora.network/t/outlier-resistant-network-inferences/94)
**Author:** [@florian](https://research.allora.network/u/florian)\ **Replies:** 10\ **Last updated:** [December 19, 2024, 2:22pm UTC](https://research.allora.network/t/outlier-resistant-network-inferences/94 "2024-12-19T14:22:23Z")
Inferers who consistently produce good inferences get a high score over time, and therefore get a large weight assigned to them when computing the network inference. But what happens if an inferer who used to give good r… --- ## [Paper "Merit-Based Sortition in Decentralized Systems"](https://research.allora.network/t/paper-merit-based-sortition-in-decentralized-systems/92)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 0\ **Last updated:** [November 19, 2024, 7:10am UTC](https://research.allora.network/t/paper-merit-based-sortition-in-decentralized-systems/92 "2024-11-19T07:10:40Z")
Thread for discussion of https://doi.org/10.70235/allora.0x10020. Merit-Based Sortition in Decentralized Systems ADI 1, 20-27; October 21, 2024 J. M. Diederik Kruijssen, Renata Valieva, Kenneth Peluso, Nicholas Emmons,… --- ## [Paper "Allora: a Self-Improving, Decentralized Machine Intelligence Network"](https://research.allora.network/t/paper-allora-a-self-improving-decentralized-machine-intelligence-network/91)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 0\ **Last updated:** [November 19, 2024, 7:09am UTC](https://research.allora.network/t/paper-allora-a-self-improving-decentralized-machine-intelligence-network/91 "2024-11-19T07:09:13Z")
Thread for discussion of https://doi.org/10.70235/allora.0x10001. Allora: a Self-Improving, Decentralized Machine Intelligence Network ADI 1, 1-19; September 2, 2024 J. M. Diederik Kruijssen, Nicholas Emmons, Kenneth P… --- ## [Using epsilon to determine topic numerical precision](https://research.allora.network/t/using-epsilon-to-determine-topic-numerical-precision/88)
**Author:** [@steve](https://research.allora.network/u/steve)\ **Replies:** 16\ **Last updated:** [November 9, 2024, 9:06pm UTC](https://research.allora.network/t/using-epsilon-to-determine-topic-numerical-precision/88 "2024-11-09T21:06:22Z")
The topic variable, epsilon (called f\_tolerance in the simulator), controls the numerical precision that the network can achieve. Different problems will require different numbers of significant digits. If epsilon is to… --- ## [Monitoring reputer health](https://research.allora.network/t/monitoring-reputer-health/81)
**Author:** [@cal-hawk](https://research.allora.network/u/cal-hawk)\ **Replies:** 7\ **Last updated:** [September 19, 2024, 2:33pm UTC](https://research.allora.network/t/monitoring-reputer-health/81 "2024-09-19T14:33:02Z")
In this post, I will focus on establishing topic health by evaluating the performance of reputers. The idea here is to design an empirical model that can assess how well reputers are functioning within the network. Here … --- ## [Monitoring tokenomics health](https://research.allora.network/t/monitoring-tokenomics-health/82)
**Author:** [@cal-hawk](https://research.allora.network/u/cal-hawk)\ **Replies:** 6\ **Last updated:** [September 19, 2024, 2:32pm UTC](https://research.allora.network/t/monitoring-tokenomics-health/82 "2024-09-19T14:32:20Z")
In this post, I’m going to focus on how we can monitor the health of emissions and inflation in real time. The idea is to make sure that the system is working as intended—basically, ensuring everything is running smoothl… --- ## [Monitoring topic and validator reward health](https://research.allora.network/t/monitoring-topic-and-validator-reward-health/80)
**Author:** [@cal-hawk](https://research.allora.network/u/cal-hawk)\ **Replies:** 6\ **Last updated:** [September 19, 2024, 2:30pm UTC](https://research.allora.network/t/monitoring-topic-and-validator-reward-health/80 "2024-09-19T14:30:52Z")
In this post, I will explore using the Gini coefficient to assess the health of topic and validator rewards. The goal is to ensure that rewards are distributed fairly and that no single validator or topic is dominating. … --- ## [Monitoring inferer health](https://research.allora.network/t/monitoring-inferer-health/78)
**Author:** [@cal-hawk](https://research.allora.network/u/cal-hawk)\ **Replies:** 7\ **Last updated:** [September 19, 2024, 2:07pm UTC](https://research.allora.network/t/monitoring-inferer-health/78 "2024-09-19T14:07:37Z")
This post is part of a broader effort to design a monitoring suite for the Allora Labs research team. Our goal is to design a suite of metrics to track the health of various components in the Allora network, allowing us … --- ## [Monitoring forecaster health](https://research.allora.network/t/monitoring-forecaster-health/79)
**Author:** [@cal-hawk](https://research.allora.network/u/cal-hawk)\ **Replies:** 5\ **Last updated:** [September 18, 2024, 6:47pm UTC](https://research.allora.network/t/monitoring-forecaster-health/79 "2024-09-18T18:47:34Z")
In this post, I’ll extend the work we’ve done on inferer health to focus on the health of forecasters. My goal is to take the strategy we developed for inferers—assessing model diversity—and apply it directly to forecast… --- ## [Systematic survey of network loss vs pnorm](https://research.allora.network/t/systematic-survey-of-network-loss-vs-pnorm/33)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 17\ **Last updated:** [September 18, 2024, 3:55pm UTC](https://research.allora.network/t/systematic-survey-of-network-loss-vs-pnorm/33 "2024-09-18T15:55:15Z")
In equations 6 and 7 of the whitepaper, we specify the potential function that ends up setting the weights of inferences contributed to the network. This potential function features a free parameter p, which has been cho… --- ## [Extend Allora's Inference Synthesis to classification tasks](https://research.allora.network/t/extend-alloras-inference-synthesis-to-classification-tasks/34)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 11\ **Last updated:** [September 18, 2024, 3:49pm UTC](https://research.allora.network/t/extend-alloras-inference-synthesis-to-classification-tasks/34 "2024-09-18T15:49:25Z")
In the whitepaper, we describe how Allora achieves Inference Synthesis using clearly defined losses, regrets, and ground truths, using weighted averages of scalar target variables – all characteristic of supervised regre… --- ## [Merit-based sortition](https://research.allora.network/t/merit-based-sortition/76)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 6\ **Last updated:** [September 10, 2024, 9:37am UTC](https://research.allora.network/t/merit-based-sortition/76 "2024-09-10T09:37:47Z")
I’ve been thinking about the need to define an “active” set of workers or reputers who are participating. During testnet, Allora’s had tens of thousands of workers, and it is computationally infeasible to accommodate all… --- ## [Adaptive alpha\_regret: Objective standard for choosing alpha\_regret to best handle changing worker performance](https://research.allora.network/t/adaptive-alpha-regret-objective-standard-for-choosing-alpha-regret-to-best-handle-changing-worker-performance/74)
**Author:** [@Renata](https://research.allora.network/u/Renata)\ **Replies:** 5\ **Last updated:** [August 14, 2024, 5:50pm UTC](https://research.allora.network/t/adaptive-alpha-regret-objective-standard-for-choosing-alpha-regret-to-best-handle-changing-worker-performance/74 "2024-08-14T17:50:03Z")
A follow-up on the post “Determine guidelines for choosing alpha\_regret” Imagine worker gets much better or much worse. We don’t want to overreact if that is incidental, but we want to catch it early if it is systematic… --- ## [Determine guidelines for choosing alpha\_regret](https://research.allora.network/t/determine-guidelines-for-choosing-alpha-regret/32)
**Author:** [@Apollo11](https://research.allora.network/u/Apollo11)\ **Replies:** 8\ **Last updated:** [August 14, 2024, 2:45pm UTC](https://research.allora.network/t/determine-guidelines-for-choosing-alpha-regret/32 "2024-08-14T14:45:41Z")
alpha\_regret is a free parameter chosen by topic creators. We need to understand how the network loss depends on alpha\_regret so topic creators can choose a sensible, and ideally optimal, value. The first stage of the s… [Next page](https://research.allora.network/latest.md?page=1)