Prediction Markets: A Futures Model Demo

[GENERAL] Name and surname of main applicant

Blake Anderton

[GENERAL] Are you delivering this project as an individual or as an entity (whether formally incorporated or not)

Entity (Incorporated)

[GENERAL] Co proposers and Additional applicants

Advisor: Dr. Dhiveshan Govender

[GENERAL] Please specify how many months you expect your project to last (from 2-12 months)

6

[GENERAL] Please indicate if your proposal has been auto-translated into English from another language.

No

[GENERAL] Summarize your solution to the problem (200-character limit including spaces)

Photrek will develop decentralized, open-source risk characterization tools to improve trading strategy effectiveness & customizability.

[GENERAL] Does your project have any dependencies on other organizations, technical or otherwise?

No

[GENERAL] If YES, please describe what the dependency is and why you believe it is essential for your project’s delivery. If NO, please write “No dependencies.”

No dependencies

[GENERAL] Will your project’s output/s be fully open source?

Yes

[GENERAL] Please provide here more information on the open source status of your project outputs

Photrek is developing Communities of Innovation centered around development and utilization of open-source capabilities. The Risk Intelligence Community of Innovation will focus on development of machine intelligence algorithms for managing risk. This effort grew out of an open-source collaboration called Machine Intelligence for Complex Systems which produced the following open source code bases:

Nonlinear Statistical Coupling Library - Apache v2.0 License

Coupled Variational Autoencoder - GNU v3.0 License

[SOLUTION] Please describe your proposed solution.

Probabilities associated with future events may over- or under-predict outcomes. For Decentralized Finance (DeFi) applications, inaccurate probability predictions lead to lost revenue when they are utilized by trading strategies – formulated under misplaced (e.g., presuming a conservative forecast) assumptions – and ultimately produce significant and unintended risk exposure. Also, forecasted probabilities may vary in suitability for particular uses whether in characterization of low-probability events (e.g., tail-risk) or reporting of high-probability outcomes (e.g., to support confident decision making). Since probability forecasts may combine aspects of occurrence frequency and confidence, it is necessary to characterize these forecasts in a risk-informed manner.

Photrek has developed “Risk Profiles” – an innovative characterization of the quality of model-predicted probabilities of future events when compared to an outcome’s observed (or “true”) probability. The Risk Profile characterizes a model’s predictive power according to statistical information theory metrics including Accuracy, Decisiveness, and Robustness. This Risk Profile is now a hosted application on SingularityNET's AI Marketplace and provides valuable insights to make risk-informed decisions. The Risk Profile will empower traders on the Cardano futures market to accurately characterize probability forecasts to effectively create risk-informed investment decisions and portfolios. 

Photrek's Use Case Concept is to apply our expertise in forecast evaluation to develop a Community of Innovation focused on Risk Intelligence. We will start with a demonstration of how data from Cardano futures markets such as Delta Exchange, Kraken, and the anticipated decentralized Axo market can be used to produce a histogram of the expected ADA price at future dates. The prototype will convert the call and put prices for an expiration date into a probability distribution of the expected asset price. As actual strike dates pass, the actual price will be used to assess the Accuracy, Decisiveness, and Robustness of the forecast. This will descriptively characterize each forecast for reliable investment decisions in the future. 

We will host workshops to explore how community members can measure the performance of their own forecasting algorithms. 

[IMPACT] Please define the positive impact your project will have on the wider Cardano community.

Photrek is building unique capabilities in Risk Intelligence for the Cardano Community. Our Predictive Markets: A Futures Models Demo project will demonstrate how high-quality assessments of forecasts can drive improvements in the development of AI algorithms for forecasting.

Photrek seeks to make sure that the Cardano community is educated about how to safely utilize futures markets to manage investment risks. Our Risk Assessment tool will be demonstrated with existing Cardano futures markets and will engage with the Axo community to prepare for their mainnet launch.

Impact will be measured in a variety of manners. Photrek will use the demonstration – risk characterization of ADA futures – to quantitatively assess the quality of futures forecasts on various markets. Subsequently, these results will be presented to community members in workshops, and the solicited feedback will improve the efficacy of communicating the value of the Risk Assessment tool. Capabilities will be shared through the open-source tool databases and publication of workshop materials for the wider community.

[CAPABILITY & FEASIBILITY] What is your capability to deliver your project with high levels of trust and accountability? How do you intend to validate if your approach is feasible?

Photrek has been delivering high-quality research, educational workshops, and risk analysis for the Cardano Community since 2021. Photrek has successfully led and completed three Catalyst projects (the Lido Nation reference is missing our Predicting Cardano Native Tokens project); and contributed to several projects with Community Governance Oversight and other groups.

Photrek was the first SingularityNET Deep Fund team to launch a Dapp on their AI Marketplace. We are currently hosting the Risk-aware Data Generator and Risk Assessment Apps; and we are funded to launch Simulating Risky Worlds. These Risk Intelligence capabilities will provide the foundation for us to prototype the Predictive Markets app.

We will validate the feasibility of our concept by testing the prediction algorithm using futures data from existing derivatives markets such as Delta Exchange and Kraken. This will validate whether the concept can be launched using the Axo DeFi datasets once available.

The methodology of the Risk Assessment tool has been demonstrated in a variety of contexts to underscore its capability and feasibility. The Risk Profile may be illustrated through its use on synthetic data in which model probabilities are simulated from the true (or “source”) probabilities. The following figure illustrates the Risk Profile for three models (which respectively exponentiated source probabilities by powers of 0.5, 1.0, and 1.5 to simulate differences between model and source probabilities). The source probabilities were created for a case in which each of 1000 instances contains 4 possible outcomes (“pick A, B, C, or D”) and uniform-sampled probabilities were assigned to each instance’s outcome, then normalized. True event outcomes were obtained by sampling (“A, B, C, or D”) according to the source probabilities, producing 1000 total observed events.

The Accuracy metric – highest for the 1.0-power case – reflects the model’s calibration to a frequency of occurrence. The Decisiveness metric – highest for the 1.5-power model case – reflects the model’s utility for using probabilities in decisions (e.g., in which probability ordering – not value – is the key need for effective classification performance). The Robustness metric – highest for 0.5-power – reflects a model’s tendency to report low probabilities for the true event. 

The following figure continues this synthetic model illustration by comparing model probabilities to source (or “truth”) probabilities on a calibration plot. These plots illustrate characteristics of under- and over-confident models when source probabilities are known. However, the Risk Profile metrics – being calculated from the model probabilities – do not require knowledge of the source probabilities but only the outcomes observed. This example with synthetic data illustrates Risk Profile metrics offer insight for evaluating the utility of various sources of probability forecasts.

As an additional example, Photrek used the Risk Profile methodology to evaluate the forecast accuracy of Five-Thirty-Eights 2022 US House of Representatives elections. In this case, despite wide-spread news media interpretation of an inaccurately forecasted Red Republican wave, in fact the forecasts were technically quite accurate, though poorly interpreted. See full article. This example illustrates both (1) how Risk Profile analysis is readily applicable to any probabilistic forecast context and (2) how Risk Profile analysis produces additional valuable insights – here, the accuracy in specifying a forecast’s degree of uncertainty – besides exclusively the Robustness, Accuracy, and Decisiveness metrics.

[Project Milestones] What are the key milestones you need to achieve in order to complete your project successfully?

Create baseline predictive algorithm and assessment metric (Month 2)

• Acceptance Criteria: Demonstration of translating futures prices into a histogram of price probabilities
• Deliverables: summary graphics which demonstrate a capability to access and aggregate futures prices from a specified exchange.
• Outputs: summary graphics for both (1) futures outcomes (i.e., the underlying asset value at call/put option expiry) and (2) futures forecasts (i.e., forming forecast probabilities through using ask/bid prices for each strike price associated for call/put options).
• Intended outcome: ability to access and aggregate the input data – probability forecasts and outcomes – required to conduct Risk Profile analysis.

Integrate risk metrics into predicted futures (Month 4)

• Acceptance Criteria: Demonstration of Risk Profile assessment of predictions using historical futures data
• Deliverables: (1) summary graphics which demonstrate a capability to conduct Risk Profile analysis of futures prices and forecasts from a specified exchange, and (2) analysis report illustrating the utility of the Risk Profile in meaningfully evaluating the accuracy and reliability of forecast probabilities.
• Outputs: summary graphics providing visual illustration of the Risk Profile analysis (e.g., metrics overlaid upon histograms) and report illustrating the use of Risk Profile analysis in creating valuable insights for an exchange’s futures market.
• Intended outcome: ability to conduct and meaningfully interpret Risk Profile analysis for providing useful insights on accuracy and reliability of an exchange’s futures market.

N//A

N/A

N/A

Post refined algorithms onto repository along with final report submission (Month 6)

• Acceptance Criteria: Plan for Solutions proposal using the results of the Concept project
• Deliverables: (1) repository-hosted algorithms available to the Cardano community, (2) final summary report. (3) Community workshops demonstrating capabilities.
• Outputs: (see deliverables) 
• Intended outcome: empower and enlighten Cardano members with tools and insights for risk-informed ADA futures trading.

[RESOURCES] Who is in the project team and what are their roles?

• Blake J. Anderton, Ph.D. – Principal Investigator
• Kenric Nelson, Ph.D. - Risk Intelligence algorithm inventor
• Juana Attieh - Customer Discovery Planning
• Dhiveshan Govender, Ph.D. - Photrek Advisor

Biographies of the Photrek Team

Dr. Blake Anderton is Founder and President of Acuition Solutions, LLC which develops machine intelligence solutions tailored for customers in operations research, precision agriculture, and environmental science. Anderton specializes in applied remote sensing, autonomous platforms/UAVs, decision support systems, quantitative finance, and human-centric AI. Anderton has 21 years experience in software development across environments spanning embedded systems to cloud computing. He holds a Ph. D. in Optical Science (U. of Arizona), M.S. in Electrical Engineering (U. of Alabama-Huntsville), and B.S. in both Physics and Engineering Mechanics (Lipscomb U.).

Dr. Kenric Nelson is Founder and President of Photrek, LLC which is developing novel approaches to Complex Decision Systems, including dynamics of cryptocurrency protocols, sensor systems for machine intelligence, robust machine learning methods, and novel estimation methods. He served on the Cardano Catalyst Circle governance council and is contributing to designs of decentralized governance. Prior to launching Photrek, Nelson was a Research Professor with Boston University Electrical & Computer Engineering (2014-2019) and Sr. Principal Systems Engineer with Raytheon Company (2007-2019). He has pioneered novel approaches to measuring and fusing information. His nonlinear statistical coupling methods have been used to improve the accuracy and robustness of radar signal processing, sensor fusion, and machine learning algorithms. His education in electrical engineering includes a B.S. degree Summa Cum Laude from Tulane University, a M.S. degree from Rensselaer Polytechnic Institute, and a Ph.D. degree from Boston University. His management education includes an Executive Certificate from MIT Sloan and participation in NSF’s I-Corp program.

Juana Attieh, a graduate of Management Engineering from the University of Waterloo, is the co-founder of FLUUS, a solution unlocking Instant Settlements for Emerging Markets. Alongside her role at FLUUS, Juana established LALKUL, a Cardano Stake Pool with a mission to integrate blockchain technology in the MENA region. Furthermore, as a co-founder of the Cardano MENA community, Juana is committed to fostering decentralized governance and contributing towards optimal solutions for self-organizing systems. With her work, Juana seeks to reimagine societies, unlock untapped potential, and provide inclusive opportunities to those who need them most.

Dr. Dhiveshan Govender spent the last decade as invited builder and steward to radical open ecosystem programs with projects focused on the intersections of DeFi, ReFi and DeSoc. He worked alongside APG, the biggest pension provider in Europe (2016-2019), complementing them to position as digital pioneer with investment in large-scale use of data, workflow automation and digital analytical platforms to move beyond jurisdictional boundaries with the vision to serve global citizens. He was a founding member and provided strategic advisory to UTRUST (2017-2022), a leading payment infrastructure and virtual asset service provider, which at the time was the only venture to successfully negotiate its central banking license from Banco de Portugal to operate across all categories. Prior to 2014, Govender was recognised for his passion and experiences in delivering better outcomes to emerging and frontier market challenges, particularly across the African continent. In 2019 Govender turned part-time institutional capital allocator - recognising the unmet need to bridge, actively explore and develop relationships around investment regenerative principles and methodologies.

[BUDGET & COSTS] Please provide a cost breakdown of the proposed work and resources.

MS-1: Create baseline predictive algorithm and assessment metric.

Budget: ₳ 40,500

MS-2:Integrate risk metrics into predicted futures.

Budget: ₳ 35,800

MS-3: Post refined algorithms onto repository along with final report submission.

Budget: ₳ 23,700

Total: ₳ 100,000

[VALUE FOR MONEY] How does the cost of the project represent value for money for the Cardano ecosystem?

Photrek is led by a five person Leadership Circle that uses consent-based decision making to make sure that our policies, objectives, and execution are consistent with building a collaborate environment for our team, our partners, and our customers. We work to develop Enlightened Pathways that enhance the intelligence and sustainability of the planet. Our approach to corporate governance is specified in the Photrek Sociocratic Operating Agreement.

Towards that end, we utilize competitive pricing methods that provide the highest value to our customers and support our team members with fulfilling lives. Our rates are based on self-employment in the US & Canada. The rates take into account the employment overheads of the resources contracted. The amounts are calculated for each milestone based on the hours to complete. For example the chart for engineering and scientific salaries in the Commonwealth of Massachusetts is provided here: https://www.mass.gov/guides/salary-and-compensation.

[IMPORTANT NOTE] The Applicant agrees to Fund Rules and also that data in the submission form and other data provided by the project team during the course of the project will be publicly available.

I Accept