How we built the Diamond Pigs crypto sentiment dashboard

The Diamond Pigs Crypto Sentiment Dashboard started as an internal research tool - not a product. It was built to solve a problem our own team faced: how do you systematically combine on-chain data, macroeconomic signals, derivatives positioning, and market sentiment into a single, reliable view of the crypto market? This case study explains the decisions behind the dashboard, the reasoning for every signal we included, and the AI methodology that turns raw market data into a structured market consensus.

Why we built an intelligence layer before building smarter bots

The origin of the crypto sentiment dashboard was not a marketing decision. It was a research one.

Our existing trading technology relied primarily on technical indicators - RSI, moving averages, trend signals. These worked well in the conditions they were designed for. However, as automated trading grew to account for 70-90% of global trading volume (CoinGecko research), the edge from widely-used technical indicators began to shrink. When thousands of traders use the same signals simultaneously, those signals become predictive of each other's behaviour rather than of the market itself. Investopedia describes this as a core limitation of technical analysis: when a tool becomes too popular, its predictive value erodes (Investopedia).

We needed a broader view. So we began experimenting with a wider range of external indicators - on-chain data, macroeconomic conditions, derivatives positioning, and sentiment measures - to identify which combinations of signals could add genuine, non-redundant information to our trading models.

That research process became the intelligence layer. The Crypto Sentiment Dashboard is the public, simplified version of that layer - a structured market view built on the same foundation that now powers our next-generation AI trading models.

The decision to make it public came once the layer became stable and consistently produced valuable insights. We also recognised that many investors manage all or part of their portfolio themselves. By making a simplified version freely available, those investors can access the same market intelligence that drives our automated investment strategies - without needing to subscribe to the platform.

How we selected the seven signals

Signal selection was one of the most debated parts of the project. The goal was not to display as many indicators as possible. It was to find the smallest set of complementary signals that together provided a balanced, non-redundant view of the market.

We evaluated many more indicators than appear on the dashboard. Some were too highly correlated with each other - they told the same story from a slightly different angle, adding visual complexity without adding information. Others added too much short-term noise, making the view unstable without improving its accuracy. A third group were better suited to our internal AI models than to a public-facing tool.

We also ran a small survey among existing Diamond Pigs customers to understand which signals they considered most useful. Their feedback helped us prioritise clarity over comprehensiveness for the public view.

The seven signals that made the final cut each measure a distinct aspect of the market. The table below summarises what each one tracks and the specific insight it adds:

‍

One insight that emerged clearly from our internal debate: more indicators do not produce better results. Every additional signal had to justify its place by adding unique information. If two indicators told us essentially the same story, we removed one.

It is worth noting that the AI-generated market summary and Bitcoin guidance shown on the dashboard draw on the full intelligence layer - which continuously evaluates many more signals than the seven displayed. The dashboard gives investors a clean, accessible view. The AI behind it operates on a far more comprehensive picture.

How the composite sentiment score works

The Market Sentiment module combines four sub-signals into a single score: social sentiment, volatility, momentum, and Bitcoin dominance.

Each sub-signal captures a different dimension of market psychology and structure:

• Social sentiment reflects how investors are talking and feeling about the market
• Volatility measures the degree of price uncertainty and the intensity of market moves
• Momentum captures the strength and direction of the current trend
• Bitcoin dominance indicates whether capital is concentrating in Bitcoin or flowing into the broader crypto market

Together, these four inputs provide a quick, intuitive snapshot of the forces shaping market consensus - from investor psychology to capital flows.

One of the core design questions was how to handle conflicting signals. What happens when price momentum is bullish but sentiment is in extreme fear? Or when funding rates are elevated but on-chain data shows accumulation?

Our answer was to build a dynamic intelligence layer rather than a fixed formula. Instead of averaging all indicators equally, the AI continuously evaluates how different signals interact and adjusts their importance based on the current market environment. Some indicators become more relevant during trending markets. Others carry more weight during periods of high volatility or uncertainty. A simple average would miss this context entirely.

The result is a market consensus with an associated confidence level. If the evidence is strong and consistent across signals, the confidence is high. If signals are conflicting or inconclusive, the system waits. We believe that avoiding low-conviction decisions is just as important as identifying high-conviction opportunities - a principle that applies equally to our investment strategies and to individual investors reading the dashboard.

How independent AI models reach a market consensus

One of the most distinctive elements of the intelligence layer is the use of several independent AI models that each evaluate the market separately before a combined view is formed.

Rather than relying on a single model to process all signals simultaneously, each AI model first develops its own independent assessment, acting like a specialist analyst with its own area of expertise. The models then effectively "discuss" their conclusions, challenging and validating each other's perspectives before reaching a shared market consensus. This collaborative reasoning process helps reduce bias, increases robustness, and leads to more balanced and informed investment decisions.

This structure was a deliberate design choice. The table below shows why this approach outperforms a single-model design:

‍

Through testing, we found that too few independent models produced consensus too easily - they converged on the same view without sufficient challenge. Too many introduced noise without adding meaningful new perspectives. The number we landed on provided the right balance between independent challenge and coherent synthesis, and is part of the proprietary design of the intelligence layer.

The combined consensus determines one of three positions: long, short, or neutral. If the models disagree significantly, the confidence score stays low and the system holds back. This architecture directly reduces the risk of a single data source or flawed signal driving a poor decision - a meaningful safeguard in a market as volatile as crypto.

Why four-hour updates - and not more frequent

The dashboard updates every four hours. This was not a default choice - it matches the timeframe our AI trading engine uses to evaluate the market.

We tested more frequent updates. The result was increased short-term noise: the dashboard became reactive to brief, low-signal price moves that reversed within hours. More frequent updates created the impression of constant change without adding meaningful information.

Less frequent updates - daily or twice-daily - delayed the detection of genuine shifts in market conditions. In a market that can move 5-10% in a single session, that delay matters.

The table below summarises what we found when testing different update frequencies:

Four hours represents the right balance between responsiveness and stability. It is frequent enough to detect meaningful changes in on-chain flows, derivatives positioning, and macro conditions. It is slow enough to filter out the noise that dominates shorter timeframes.

This alignment between the dashboard and our trading engine is also intentional. The same signals that update the public dashboard feed directly into the AI models driving our automated strategies. The dashboard is not a separate product - it is a window into the same intelligence layer.

A real example: the short signal on 18 June 2026

On 18 June 2026, the intelligence layer generated a short signal on Bitcoin at approximately $64,628.

At the time, the combination of signals across the intelligence layer produced a sufficiently high-confidence bearish assessment. Since then, Bitcoin moved toward $60,000 - representing a return of roughly 7% on the position. As of 29 June, the signal remains open, with the system continuing to evaluate conditions every four hours.

This example illustrates the system working as designed: not reacting to a single indicator, but synthesising multiple data streams into a high-confidence position. It also illustrates the role of the four-hour update cycle - the position has been continuously re-evaluated since opening, and the system has not yet seen the conditions that would trigger an exit.

No strategy is correct all the time. Throughout development, we conducted extensive backtesting to optimise the intelligence layer for long-term, risk-adjusted performance rather than prediction accuracy alone. We track performance across five metrics:

• Sortino ratio - return relative to downside volatility only; the most relevant measure for crypto because it does not penalise upside moves (Investopedia)
• Sharpe ratio - return relative to total volatility; used alongside Sortino for a complete picture (Investopedia)
• Maximum drawdown - the largest peak-to-trough decline; a direct measure of capital protection
• Win/loss ratio - the proportion of signals that close in profit
• Outperformance vs. buy-and-hold Bitcoin - our ultimate benchmark; the intelligence layer must consistently beat passive holding to justify its complexity

When a signal proves wrong, the system is designed to recognise changing market conditions and exit with a controlled loss. Over time, consistent risk management matters more than trying to predict every market move correctly. This is the same principle behind Diamond Pigs' broader risk management approach: protecting capital during adverse conditions is not a secondary goal - it is central to the strategy.

How the dashboard connects to the next-generation AI trading bots

The Crypto Sentiment Dashboard is the first public layer of a broader AI technology project.

The intelligence layer was built as a shared foundation. New trading strategies can be developed on top of it, each benefiting from the same market intelligence while applying different investment objectives, risk profiles, and execution rules. When the intelligence layer identifies a trading opportunity with sufficient confidence, it generates a signal that flows directly to the Diamond Pigs trading infrastructure, which automatically executes the trade in each user's connected exchange account.

For investors who want to understand how this infrastructure translates into active portfolio management, the how it works page explains the full process from signal to trade execution.

We are starting with Bitcoin because it offers the deepest data history, the most liquid derivatives markets, and the strongest correlation with macroeconomic conditions - making it the best environment for training and validating AI models. The Bitcoin-macro correlation is well documented: as explored in our earlier analysis, Bitcoin's behaviour in response to Federal Reserve policy and dollar strength makes it the most predictable starting point for multi-signal AI modelling.

However, the same architecture can be extended to other assets including Ethereum, Solana, XRP, and others as the platform scales. Instead of redesigning each bot around individual indicators, the shared intelligence layer means new strategies inherit the same market understanding from day one - making the platform more scalable, easier to improve, and more consistent across all future AI-powered strategies.

Investors curious about whether the platform suits their approach can use the free Strategy Matching Tool, which asks six questions and recommends the best-fit strategy based on wallet size, risk tolerance, and goals.

Key takeaways

• The Diamond Pigs Crypto Sentiment Dashboard began as an internal research tool, built to give our AI trading models a structured, multi-signal view of market conditions before any trade is placed.
• Seven signals were selected from a much larger set - each chosen because it measures a distinct aspect of the market that the others do not capture. More indicators did not produce better results.
• The composite sentiment score combines four sub-signals (social sentiment, volatility, momentum, BTC dominance) and is weighted dynamically based on market conditions - not averaged equally.
• Several independent AI models each evaluate the market separately before their views are combined into an overall consensus. This architecture surfaces disagreements that a single model would miss.
• Four-hour updates match the trading engine's own evaluation cycle - frequent enough to detect meaningful shifts, slow enough to filter out short-term noise.
• Performance is tracked against five metrics: Sortino ratio, Sharpe ratio, maximum drawdown, win/loss ratio, and outperformance vs. buy-and-hold Bitcoin.
• The dashboard is the public layer of a broader intelligence infrastructure that will power all future Diamond Pigs AI trading strategies across Bitcoin and other assets.

Frequently asked questions

What is the Diamond Pigs Crypto Sentiment Dashboard?

The Diamond Pigs Crypto Sentiment Dashboard is a free market intelligence tool that combines seven key signals - including Bitcoin price, on-chain flows, funding rates, US money supply, and market sentiment - into a single, structured market view updated every four hours. It is designed to help investors understand what is driving current crypto market conditions. Access is free at diamondpigs.com/crypto-sentiment-dashboard.

How does the dashboard decide whether the market is bullish or bearish?

The dashboard uses an AI intelligence layer that evaluates multiple signals simultaneously. Rather than applying a fixed formula, the system dynamically adjusts the weight of each signal based on current market conditions. Several independent AI models each evaluate the market separately before a combined consensus is produced. The result is an overall market view with an associated confidence level - not a simple average of indicators.

Why does the dashboard only show seven signals if the AI evaluates more?

The seven signals shown on the dashboard were selected to give investors a clean, balanced, and easy-to-understand view of the market without overwhelming them with data. The AI-generated market summary and Bitcoin guidance are based on a far more comprehensive intelligence layer that continuously evaluates many more signals. The dashboard is a simplified public window into that broader system.

How often does the crypto sentiment dashboard update?

The dashboard updates every four hours, matching the timeframe used by Diamond Pigs' AI trading engine. This frequency was chosen to balance responsiveness with stability - frequent enough to detect meaningful changes in market conditions, slow enough to filter out short-term noise that reverses quickly.

How does the multi-model AI system work?

Several independent AI models each evaluate the market from a different perspective before their views are combined into an overall consensus. The specific design of the system is proprietary, but the core principle is that each model assesses the market separately first - surfacing contradictions and disagreements before synthesis. This reduces the risk of any single flawed signal driving a poor decision.

Is the Diamond Pigs Sentiment Dashboard connected to its trading bots?

Yes. The dashboard is the public layer of the same intelligence infrastructure that powers Diamond Pigs' AI trading models. When the intelligence layer generates a high-confidence signal, it feeds directly into the trading engine, which executes positions automatically in each user's connected exchange account. The dashboard and the trading bots operate on the same data foundation.

What performance metrics does Diamond Pigs use to evaluate the intelligence layer?

Diamond Pigs tracks five metrics: Sortino ratio, Sharpe ratio, maximum drawdown, win/loss ratio, and outperformance versus a simple buy-and-hold Bitcoin strategy. The Sortino ratio is prioritised because it measures return relative to downside volatility only - more meaningful in a market where large upside moves are expected and should not be penalised.

Glossary

Crypto sentiment dashboard - A market intelligence tool that combines multiple crypto market signals into a single, regularly updated view of current market conditions.

Intelligence layer - Diamond Pigs' term for the AI system that continuously analyses market signals and generates a structured market consensus before any trading decision is made.

Funding rate - A periodic payment between traders in futures markets that reflects the balance of long and short positions. Positive funding rates indicate a long-heavy market; negative rates suggest short positioning dominates.

BTC Netflow - The net movement of Bitcoin between private wallets and exchanges. Outflows (Bitcoin leaving exchanges) typically indicate accumulation; inflows suggest selling pressure may increase.

Sortino ratio - A performance metric that measures investment return relative to downside volatility only. Unlike the Sharpe ratio, it does not penalise upward price moves - making it a more relevant measure for evaluating crypto trading strategies.

Sharpe ratio - A measure of risk-adjusted return that accounts for total volatility (both upside and downside). Used alongside the Sortino ratio to give a complete picture of strategy performance.

VIX - The CBOE Volatility Index, which measures expected near-term volatility in the US stock market. A rising VIX signals growing fear and risk aversion; a falling VIX typically supports appetite for risk assets like crypto.

Maximum drawdown - The largest peak-to-trough decline in a strategy's value over a given period. Used to measure capital protection and downside risk in backtesting.