This release introduces Gold (XAU) and Silver (XAG) trading agents alongside BTC; bringing real-world assets into the agentic trading stack.

What’s New in V4

New Markets: Gold (XAU) & Silver (XAG)

New Models: xAI Grok 4.1, Claude Opus 4.5 & GPT 5.1

Mode Trade AI Agents now support:

• BTC

• Gold (XAU)

• Silver (XAG)

Real-world assets. Real-time signals. Fully onchain execution.

A Single Strategy, Multiple Perspectives

Every Mode Trade AI Agent uses the same fixed trading strategy, powered by x402 indicators. What changes between agents is how the signals are interpreted, not the signals themselves. Each agent:

• Reads identical market data

• Applies the same indicator framework

• Produces a directional trade decision

How the Agents Work

All V4 agents are powered by the same x402-driven indicator set:

Donchian Channels

Used to identify:

• Breakouts

• Breakdowns

• Overall market structure

MACD

Used to assess:

• Momentum

• Trend direction

• Confirmation strength

These indicators form a deterministic, rule-based strategy that runs continuously and consistently — without emotion or discretion.

Model Layer: Different Reasoning, Same Inputs

While the strategy is fixed, the reasoning layer varies by agent.

V4 agents are currently powered by:

• xAI Grok 4.1

• Claude Opus 4.5

• ChatGPT 5.1

Each model interprets the same x402 signals and produces its own directional bias. This allows users to observe, compare, trust or counter different models; all within the same framework.

Trust or Counter

For each agent, users can select one of two execution modes:

• Trust Mode - The trade is executed in the same direction as the agent’s decision.

• Counter Mode - The trade is executed in the opposite direction to the agent’s decision.

Both modes use the same position sizing and risk controls, only the direction differs.

Why V4 Matters

Mode Trade AI Agents V4 aren’t about chasing predictions.

They’re about consistent execution, transparent logic and user-controlled automation.

• Same signals

• Same strategy

• Different reasoning

• Clear control

As more markets go onchain and more trading becomes automated, agents need to be verifiable, deterministic and flexible. That’s what V4 delivers.

Get started creating your own V4 AI Trading Agents HERE

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website:  Mode

AI Trading Agents: Agents

Mode Trade: Trade

Mode Terminal: AI Quant

Community TG Groups: Telegram

Introduction

What are Mode Trade AI Agents?

Mode Trade AI Agents are automated trading assistants that help you trade more consistently using a fixed, rule-based strategy powered by our x402 indicator framework.

Agents handle execution, risk management, and trade monitoring so that you do not need to make decisions in real time. Instead of watching markets continuously or reacting emotionally, an agent applies the same logic on every decision cycle.

What they are not

• Agents do not guarantee profits

• Agents are not substitutes for understanding basic trading concepts

• Agents are not immune to market risk - losses are possible and should be expected

Why use them

• Reduce emotional decision-making

• Maintain consistent execution

• Trade even when you are offline

• Observe and test structured decision-making in live markets
  

Getting Started

Step 1. Create a Sub-Account

Each agent must run in its own sub-account.

• One agent per sub-account

• Sub-accounts must be funded before activation

Example:

To test an agent with $50, create a sub-account named “Test Agent”, transfer $50, and launch the agent using that account.

Agent Settings

AI Trading Agents on Mode automate crypto trading using large language models (LLMs) combined with deterministic market signals from x402.

You choose:

• The AI model

• The sub-account

• The asset to trade

• Your maximum leverage

• Whether to trust or counter the agent’s decisions

Once activated, the agent runs continuously using live market data and the same strategy on every cycle.

How AI Agents Work

Every few minutes, the agent:

• Reads live market data and x402 indicator values

• Evaluates market structure using Donchian Channels

• Assesses momentum using MACD

• Forms a directional trade decision (long or short)

• Applies your selected execution mode (trust or counter)

• Selects position size within your leverage limits

• Monitors open positions using predefined risk logic

You can pause or delete the agent at any time.

Creating an Agent (Step-by-Step)

Create an AI Trading Agent V3 HERE

The agent creation flow consists of 5 steps.

Step 1: Choose Your AI Model and Execution Mode

Select which LLM will power the agent’s reasoning layer:

• GPT

• DeepSeek

• EigenAI

All models follow the same trading strategy and see the same x402 data. Differences arise only from how each model interprets the signals.

Select how you want to interact with the agent’s decisions:

• Trust Mode - Trades are executed in the same direction as the agent’s signal.

• Counter Mode - Trades are executed in the opposite direction to the agent’s signal.

This setting only affects trade direction. Risk controls and sizing logic remain the same.

Tap Next.

Step 2: Choose a Sub-Account

Agents trade exclusively from a dedicated sub-account.

You can:

• Select an existing sub-account, or

• Create a new one

Your available balance is displayed so you know how much capital the agent can use.

Tap Next.

Step 3: Set Max Leverage

Choose the maximum leverage the agent is allowed to use.

• Range: 0.5X to 5X

• Actual leverage varies dynamically based on signal strength

• The agent will never exceed your selected maximum

The UI shows estimated position size for transparency.

Tap Next.

Step 4: Choose Your Asset

Select the asset the agent will trade:

• BTC

• ETH (coming soon)

• SOL (coming soon)

• PUMP (coming soon)

• HYPE (coming soon)

Each agent trades one asset only.

Tap Next.

Step 5: Review & Activate

You’ll see a full summary including:

• Sub-account

• AI model

• Execution mode (trust or counter)

• Max leverage

• Asset

• Estimated position size

Confirm by signing a message.

Your agent is now live.

Viewing Your Agent’s Performance

Use the Portfolio tab to monitor:

• Portfolio value

• Total PnL

• Realized PnL

• Unrealized PnL

• Open and closed positions

Performance updates in real time.

Reviewing LLM Logs

Every agent decision is logged.

You can review:

• The model’s reasoning

• Indicator values from x402

• Why a trade was taken or avoided

• Risk and leverage considerations
  
  This provides transparency into agent behaviour over time.

Best Practices

• Start with lower leverage

• Keep sufficient USDC in the sub-account

• Review logs periodically

• Pause the agent before withdrawing funds
  

Pausing or Deleting an Agent

You can pause or delete an agent at any time.

When paused or deleted:

• No new trades will be opened

• Existing positions remain open unless manually closed or exited via SL/TP
  

Risks & Important Notes

• Crypto markets are volatile

• Leverage amplifies gains and losses

• Past performance does not guarantee future results

• AI models do not predict the future

• Always monitor your positions
  

FAQ

Can I run multiple agents?

Yes - one per sub-account.

Can I edit an agent after activation?

Yes. Agents can be paused or deleted and recreated.

What indicators are used?

Donchian Channels and MACD, delivered via x402.

Does the agent use stop-losses?

Yes - agents apply predefined risk logic where appropriate.

How often does the agent trade?

Agents always maintain a directional bias but only add or change positions when conditions align.

Summary

Mode Trade Agents provide a structured, observable way to automate crypto trading using x402 indicators and modern LLMs.

You choose the model.

You choose how to use the signal.

The agent handles execution and monitoring.

Create an AI Trading Agent V3 HERE

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website:  Mode

AI Trading Agents: Agents

Mode Trade: Trade

Mode Terminal: AI Quant

Community TG Groups: Telegram

x402 was developed by Coinbase and provides a secure, verifiable way for agents to request indicators, prices and analytics from registered data providers.

This means agents can now call x402 to fetch precomputed trading indicator values instead of needing to compute these manually. This reduces the compute load for agents and increases speed in data collection and therefore trade execution. 

What this means for trading agents

Agents can now:

• Call the Mode x402 HTTP API (using x402.org wrapper library)

• Request indicator values for a given pair and timeframe

• Use those values as inputs to their trading logic in real time

Today, the implementation is deliberately simple:

• Three indicators per call

• Series of values, for recent indicator values

This already covers a lot of use cases: agents can cheaply ask “What does the market look like right now?” using a rich set of indicators.

The indicator universe

Agents can query a wide range of indicators via x402, including trend, momentum, volatility, volume, and pattern tools.

With x402, agents can pull indicator data whenever they need it. Each request is priced at $0.001 on Base, allowing agents to run frequent queries autonomously without adding meaningful cost to their strategy.

Responses are structured so that agents can compose multiple indicators into a single decision step, enabling richer state representations and more intelligent behavior.

Supported Indicators (Highlights)

x402 now supports technical indicators across trend, momentum, volatility, volume, and pattern categories.

Below is a shortlist of the most widely used indicators that agents can query today:

Trend

• EMA (Exponential Moving Average)

• SMA (Simple Moving Average)

• SuperTrend

• Ichimoku Cloud

• Linear Regression

• Hull MA (HMA)

• VWMA / VWAP
  

Momentum

• RSI

• MACD

• Stoch / StochRSI

• ROC (Rate of Change)

• Momentum (MOM)
  

Volatility

• ATR (Average True Range)

• Bollinger Bands

• Standard Deviation

• True Range
  

 Volume

• OBV (On-Balance Volume)

• MFI (Money Flow Index)

• CMF (Chaikin Money Flow)
  

 Structure / Price Transforms

• Z-Score

• HL2 / HLC3 / OHLC4

• ZigZag

• Pivot Points
  

The full list can be found here https://x402.mode.network/ta/indicators/list

Why this matters for the AI Agent ecosystem

By standardising indicator computation in x402:

• Agents don’t need to ship bulky TA libraries

• Indicator calculation becomes consistent across agents

• We reduce duplication, cost, and compute overhead

• Developers can focus on strategy logic, not data plumbing

Keep tabs on Mode x402 HERE

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website:  Mode

AI Trading Agents: Agents

Mode Trade: Trade

Mode Terminal: AI Quant

Community TG Groups: Telegram

1. The longer-term signals used by our early agents meant that trades could sometimes take several days to trigger, which some users interpreted as inactivity. 

2. Users wanted a deeper understanding of why agents made certain decisions so they could better assess risk and choose configurations aligned with their strategy.

With the launch of our V2 agents, we’ve addressed both areas directly; transforming the agent experience from a black box into a transparent, data-rich, user-controlled system.

What's New in V2 AI Trading Agents

1. Live and Historical Performance Tracking

From the moment you begin, you’ll be able to see how each LLM performs over time, compare different models and understand their trading behaviour before deploying your own agent. This lets you choose models based on actual results, not guesswork.

2. Choice of AI Model

You now have full control over which intelligence powers your agent. Select from:

• EigenAI

• DeepSeek-V3

• Qwen3-Max

• OpenAI GPT-5

Each model brings its own reasoning style, giving you the ability to match trading behaviour to your goals.

3. Real-Time Decision Visibility

Every few minutes, decisions stream into a live feed across the platform. You also get a dedicated decision log, updated every 5 minutes once your agent is deployed. You’ll be able to see:

• What signals the agent processed

• How it reasoned

• Why it took or avoided a trade

• How it managed risk

This level of transparency is designed to build trust and give you confidence in your agent’s logic.

Coming at Launch

Initially, trading will be limited to BTC, with more assets coming shortly after launch. Your early configuration options include:

• Choosing your AI agent (LLM)

• Selecting your preferred leverage (.5x to 2x)

• Defining indicator allocations (RSI, EMA, MACD, Bollinger Bands)

These indicators directly influence your agent’s trading decisions.

We’ve intentionally capped leverage at 2× maximum, ensuring agents avoid overleveraging and maintain sustainable risk profiles within Mode Trade.

How Mode V2 AI Trading Agents Work

AI Trading Agents allow you to automate your crypto trades using advanced LLM decision-making. You configure:

• AI model

• Sub-account

• Indicator weights

• Maximum leverage

• Asset

Once activated, the agent trades continuously using real-time data and logical, rule-based reasoning.

Every cycle (5 minutes), each AI Trading Agent:

• Reads live market data and ATR volatility

• Applies your indicator weights

• Determines directional bias

• Chooses position size based on your max leverage

• Opens a trade only when signals strongly align

• Manages stop-loss & take-profit logic

• Monitors positions 24/7

You remain fully in control and can pause or delete the agent at any time.

You can follow a step by step guide on how to create an agent HERE

Monitoring Your Agent

In the Portfolio tab, you can see:

• Total portfolio value

• Realized and unrealized PnL

• Open and closed positions

• Overall performance

You also have access to complete LLM reasoning logs for transparency.

FAQ

Can I run multiple AI Trading Agents?

Yes. One per sub-account.

Can I edit an agent after activation?

You can pause and modify it with new settings.

What indicators are used?

RSI, EMA, MACD, Bollinger Bands; weighted based on your sliders.

Does the agent monitor stop-losses?

Yes - all agents apply logical SL/TP rules where deemed appropriate.

How often does the agent trade?

Agents are designed to always have a bias, but will only add to positions when the evidence aligns.

Summary

Mode’s new AI Trading Agents deliver a fully transparent, configurable, and data-driven approach to automated crypto trading; powered by the latest LLM technology.

You set the rules. Your agent handles the rest: 24/7, systematically and with complete visibility into every decision it makes.

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website:  Mode

AI Trading Agents: Agents

Mode Trade: Trade

Mode Terminal: AI Quant

Community TG Groups: Telegram

TLDR:

• OP incentives will continue in Season 6 for veMODE holders to direct to ecosystem apps and AI Trading Agents

• Stake Mode to earn OP rewards (approx 20 - 70% APY)

• Voting power carries over from Season 5 to Season 6

• Weekly Perps Comps up to $15000 in rewards to the top traders on Mode Trade

• Growing Mode: OP rewards aim to fuel tangible growth for Mode and the Superchain

Introduction

Season 6 marks a pivotal step in Mode’s journey; from enabling AI-powered trading to building the infrastructure for verifiable, autonomous agents. As artificial intelligence moves from analysis to action, traders are no longer limited to passive signal analysis. Instead, AI can now decide when to act, how much to risk, and what to ignore.

Mode’s mission remains clear: empower traders to define intent while intelligent agents handle execution, fast, consistent, and free from bias. Building on Season 5 which saw AI assisted trading, season 6 will focus on incentivizing the use of AI Trading Agents.

Key Dates

• Start: October 15th, 2025

• Two-Week Epochs: Voting and incentive distribution follow a bi-weekly cadence throughout Season 6.

• Weekly Perps Comp: From Tuesday at 00:00 CET to Monday at 23:59 CET; top traders rewarded every Wednesday

New In Season 6

1. AI Trading Agents

The first generation of Mode AI Trading Agents includes two powerful archetypes:

• Breakout Agent: Detects volatility and expansion, entering positions as prices break from consolidation.

• Trend Agent: Tracks and participates in sustained moves, staying in position until trend strength fades.

Both operate under controlled risk parameters, applying consistent execution strategies without emotional interference. More AI Trading Agents will be launched throughout Season 6.

2. Verifiable Agent Infrastructure

Agents can now leverage ERC-8004-compatible verification, ensuring that execution logic and trading history can be proven and attributed on-chain. This sets a new standard for trust and transparency in autonomous systems.

Mode will soon be launching a verification framework for trading agents, enabling users to verify that the actions Trading Agents make are fully backed by on-chain proofs, ensuring every trade, strategy, and performance claim is transparently verifiable in real time.

3. Expanded Ecosystem Participation

Traders, strategists and researchers will be able to design and deploy specialised AI Trading Agents within Mode’s environment, blending human intuition with algorithmic precision.

4. Weekly Perps Competition

What's on the Table?

• Total Rewards Pool: up to $15,000/week

• Duration: Until the end of season 6

• Rewards: $1500/week per $5M in volume up to $50M

How It Works

Every week, from Tuesday at 00:00 CET to Monday at 23:59 CET, the trading competition will be live.

At the end of each trading week, $1500 per $5M in trading volume (up to $50M) will be distributed pro rata to eligible traders based on their weekly trading volume.

“Pro rata” simply means you earn rewards in proportion to your share of the total trading volume. Ex: If the total weekly trading volume is $25M and you contributed $5M, you traded 20% of the total. You would therefore receive 20% of the $7500, which is $1500.

Weekly Volume - Rewards in OP

$5M - $1,500

$10M - $3,000

$15M - $4,500

$20M - $6,000

$25M - $7,500

$30M - $9,000

$35M - $10,500

$40M - $12,000

$45M - $13,500

$50M - $15,000

Op rewards will be sent directly to the winning wallets on Wednesday each week.

Who’s Eligible?

To qualify for weekly rewards, you must generate at least $20,000 in trading volume.

• Minimum volume: $20,000/week

• Only eligible traders share the weekly pool

This filter ensures that rewards go to high-signal, high-intent users actively contributing to Mode. Volume rewards are distributed pro rata to eligible traders. The more volume you generate, the greater your share.

Note: The volume on the leaderboard is the sum of all subaccounts + the main account; so the volume created by your AI Trading Agents is included in your total weekly volume.

Conclusion

Season 6 represents a major leap for Mode; from AI-assisted trading to verifiable, autonomous execution.

With the introduction of AI Trading Agents, on-chain verification, and continued veMODE incentives, Mode is building a transparent network of intelligent systems that act on human intent with speed, consistency, and accountability.

Whether you’re a trader deploying agents, a researcher designing AI logic, or a veMODE holder directing votes, Season 6 invites you to participate in the next evolution of onchain intelligence.

About Mode

Autonomous Execution

In the coming decade, we won’t just use AI to process information, we’ll use it to decide when to act, how much to risk and what to ignore.

This transition is already visible in trading. Markets generate endless signals, but the real edge lies in identifying which ones matter and executing them without hesitation. The combination of AI-driven signal detection and autonomous execution is transforming how modern systems trade.

At Mode, we’ve been building the foundations for this future - where traders define intent and intelligent agents handle execution. Our aim isn’t to replace human judgment, but to give it a faster, more consistent form.

From Signals to Execution

Every trader knows that good ideas are rarely the problem. The difficulty lies in applying them the same way, every time, under pressure.You can recognise a breakout but hesitate to act. You can see a trend forming but exit too soon. In fast markets, hesitation is costly. In volatile ones, emotion is fatal.

Mode AI Trading Agents were built to close that gap. Using AI to identify and respond to market structure in real time, then execute trades with mechanical precision.

Each AI Trading Agent lives inside its own Mode sub-account. It observes the same price data you do but without fatigue, distraction or bias. It scans continuously for signals generated by Mode’s proprietary AI models; systems trained to recognise momentum shifting into breakouts and the formation of trends across timeframes.

This blend of AI signal detection and rule-based risk management allows agents to operate as autonomous traders; consistent, explainable and aligned with your intent.

The First Generation

We’re starting with two core archetypes that reflect some of the most universal behaviours in markets:

• Breakout Agent - trained to detect volatility and expansion, positioning as prices break from consolidation zones and closing quickly after momentum drops.

• Trend Agent -  designed to participate in slow persistent moves once trend strength exceeds statistical thresholds and gets out once it’s sure the trend’s over.

Both AI Trading Agents trade under controlled limits, whilst taking care not to overfit them, as not to fall short in live trading. If there’s a breakout, it’ll catch it. If there’s a trend, you’ll be in it until it ends.

For step by step instructions on how to launch your own Agents check out our AI Trading Agents Quickstart Guide **HERE **

Performance vs. Buy and Hold

To evaluate performance, we compared these agents against the most neutral benchmark available: a simple buy-and-hold position.

Buy and hold represents continuous exposure, a strategy that performs best when volatility trends upward but can leave you in deep, long drawdowns. Its risk profile is linear: the same exposure, through all conditions.

Our agents operate differently. They allocate capital conditionally, entering markets when volatility structure supports an edge and withdrawing when conditions deteriorate.This conditional participation produces a different kind of equity curve, one that looks less dramatic but far more stable.

When tested across BTC,  ETH, and top volume coins both the Breakout and Trend Agents demonstrated lower drawdowns and a smoother equity path compared with buy and hold. In bull phases, they captured substantial upside; in sideways regimes, they often sat flat avoiding the grind of directionless volatility.

The backtests below show the performance of our Trend Agent and Breakout Agent applied to the five highest-volume trading pairs during each entry period, capital divided among positions using 0.5×, 1×, and 2× leverage with 0.1% trading costs. This approach highlights how allowing gains to compound can meaningfully affect long-term performance. The results illustrate how higher leverage amplifies returns and drawdowns, while diversified, volume-based selection improves capital efficiency across market cycles. Whilst there were two large outliers in July, it goes to show what can be achieved by bringing your intent to pair selection.

Getting the most out of Mode AI Trading Agents

Automation only reaches its potential when used deliberately. The best results come not from betting big, but from deploying many small, independent systems each trained to specialise, each contributing its own edge. Here’s how to get the most from your agents:

1. Trade a broad universe Deploy agents across multiple markets, majors and liquid altcoins alike. Strategies perform best when they can diversify signal detection. Broader coverage means more opportunities for structured participation without increasing risk.

2. Maintain independence Each agent operates in a separate sub-account, executing without interference. That independence makes it possible to measure real performance comparing behaviour across assets and strategies without cross-contamination.

3. Focus on distributions, not single trades The strength of AI-driven execution lies in repetition. You’re not aiming to predict each outcome but to shape the statistical distribution of many. Over time, consistency compounds in ways intuition never can.

4. Let the agents run manual overrides - closing trades early or changing size mid-run destroy the integrity of the process. Treat agents like instruments, not experiments. Give them time to show their edge. When approached this way, Mode Trade Agents become a way to observe markets through behaviour, not belief.

What’s Next

This launch marks the foundation of a broader ecosystem, one that combines human insight, AI signal generation and verifiable execution.

In the coming months we’ll:

• Onboard leading traders and researchers to design more specialised AI Trading Agents built on our infrastructure; agents that blend human intuition with algorithmic precision.

• Collaborate with Traders and professional strategists to share their trading logic in transparent, measurable ways, allowing users to deploy trusted systems through Mode.

• Move toward verifiability using open standards like ERC-8004, ensuring that every agent’s logic and performance can be audited, proven and attributed on-chain.

The long-term vision is clear:

A network of autonomous, verifiable agents that trade on behalf of human intent, combining the interpretive power of AI with the transparency of open systems.

Final Thoughts

The future of trading isn’t about eliminating risk; it’s about understanding it. AI will never make markets completely predictable but it can make your participation more adaptive, consistent and potentially pick up on things you can’t.

Mode AI Trading Agents are our contribution to that future: intelligent systems that use AI to detect structure, manage volatility and execute your intent with precision. In a world where attention is scarce and reaction times are measured in milliseconds, consistency is the real edge.

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website: mode.network

AI Trading Agents: Agents

Mode Trade: Trade

Mode Terminal: AI Quant

Community TG Groups: Telegram groups

How Tiers Work

There are 6 Tiers in total. Each Tier comes with reduced fees, so the higher your Tier the more you save.

You can unlock a Tier in two different ways:

1. Staking Mode Tokens: The more $MODE tokens you stake, the higher the Tier you qualify for.

2. Trading Volume: Active traders unlock higher Tiers by reaching 30 day trading milestones.

This means you can climb the Tiers whether you’re a long-term Mode holder, a high-volume trader or both. Tiers can be found by navigating to “portfolio”, then “Fee Tiers” on Mode Trade.

What happens if your trading volume puts you in one Tier, but your staked MODE puts you in another?

You always qualify for the highest Tier you meet; whether through trading volume or staked MODE.

Staked Mode = stable Tier qualification (as long as tokens are staked)

Volume = dynamic Tier qualification (can move up or down as your 30D volume changes)

How Volume-Based Tiering Works:

• Tiers are based on your rolling 30-day trading volume.

• That means the system is always looking back at the past 30 days of activity.

• If at any point your 30D trading volume drops below the threshold of your current Tier, you’ll automatically fall into the highest Tier you still qualify for (either from volume or staked MODE).

Example:

• You traded $6M last month, which put you in Tier 3.

• But in the new 30-day window, your rolling volume falls to $2M.

• Since $2M is under the Tier 3 minimum ($2.5M), you would fall back to Tier 2, unless your MODE holdings qualify you for a higher Tier.

Why Tiers Matter

• Lower Fees: Higher Tiers = lower trading fees.

• Flexible Paths: You choose how to qualify; by holding or trading.

• Fair & Transparent: Everyone can see exactly how the system works.

Together with the AI Quant and Synth, Tiers make Mode Trade not just powerful but also economically optimized for traders of all levels.

About Mode

Mode is an Ethereum L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website: mode.network

Community TG Groups: Telegram groups

Mode Terminal: AI Quant

Mode Trade: Trade

What's on the Table?

• Total Rewards Pool: $4500/week

• Duration: Until season 6 begins

• Base Rewards: $3000 per week

• Bonus Rewards: +50% weekly if the total 7-day volume reaches $20M

How It Works

Every week, from Tuesday at 00:00 CET to Monday at 23:59 CET, the trading competition will be live.

Volume

At the end of each trading week, $2,500 will be distributed pro rata to eligible traders based on their weekly trading volume. If traders hit $20M in 7-day combined volume, a 50% bonus is unlocked and added to that week’s rewards.

PnL

The top 3 traders with the highest weekly PnL will receive:

1. 250$

2. 150$

3. 100$

These rewards will increase by 50% if the total 7D trading volume exceeds $20M weekly.

Op rewards will be sent directly to the winning wallets on Wednesday each week.

Who’s Eligible?

To qualify for weekly rewards, you must generate at least $30,000 in trading volume.

• Minimum volume: $30,000/week

• Only eligible traders share the weekly pool

This filter ensures that rewards go to high-signal, high-intent users actively contributing to Mode. Volume rewards are distributed pro rata to eligible traders. The more volume you generate, the greater your share.

About Mode

Mode is building Full Stack DeFAI, the L2 and perpetuals dex scaling DeFi through AI agents and AI-driven financial applications.

Website: mode.network

Community TG Groups: Telegram groups

Mode Terminal: AI Quant

Mode Trade: Trade

What's on the Table?

• Total Rewards Pool: $36,000

• Duration: 8 weeks

• Base Rewards: $24,000 ($3000 per week)

• Bonus Rewards: +50% weekly if the total 7-day volume reaches $20M

How It Works

Every week, from Tuesday at 00:00 CET to Monday at 23:59 CET, the trading competition will be live.

Volume

At the end of each trading week, $2,500 will be distributed pro rata to eligible traders based on their weekly trading volume. If traders hit $20M in 7-day combined volume, a 50% bonus is unlocked and added to that week’s rewards.

Example: If total weekly volume = $10M and you traded $2M, that’s 20% share of the pool. Your reward: $500. If the bonus is activated and volume reaches $20M, that 20% would earn you $750.

PnL

The top 3 traders with the highest weekly PnL will receive:

1. 250$

2. 150$

3. 100$

These rewards will increase by 50% if the total 7D trading volume exceeds $20M weekly.

Op rewards will be sent directly to the winning wallets on Wednesday each week.

Who’s Eligible?

To qualify for weekly rewards, you must generate at least $30,000 in trading volume.

• Minimum volume: $30,000/week

• Only eligible traders share the weekly pool

This filter ensures that rewards go to high-signal, high-intent users actively contributing to Mode. Volume rewards are distributed pro rata to eligible traders. The more volume you generate, the greater your share.

What about AI Terminal?

Everyone holding or staking 400,000 MODE tokens is eligible to use the AI Terminal. However there are a few other options on how you can get your hands on it:

1. Scout for invite codes

We will be doing a few giveaways on our socials, so keep your eyes peeled for that and don’t be late; these will be given out on an FCFS basis. Ask your fav community lead to reach out, we love traders. Mode Streamers will also have some invite codes to give away.

2. Trade

You can trade however you like, reach $500,000 cumulative volume on the Mode Trade account, share proof, and get an invite. Applicable only for new joiners who started trading after the Mode Trade Summer Vibe announcement.

Wrap-up

Whether you're scalping, swing trading, or building long-term positions, Mode Trade is built to reward your activity. Jump in, trade onchain, and claim your share.

Stay updated by following us on X https://x.com/modenetwork

Trade with AI. Trade onchain. Trade with Mode.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi through AI agents and AI-driven financial applications.

Website: mode.network

Community TG Groups: Telegram groups

Mode Terminal: AI Terminal

Mode Trade: Trade

Synth: Homepage

Synth Subnet: Miners Live Scores

TLDR: AI-powered trading is here.

To learn more about the Mode Trade platform and the vision behind it, read our full introduction here.

Mode Trade Perps Competition

After a successful Round 1, our weekly trading competition continues and it’s bigger, better, and open to more winners. Round 2 has increased weekly rewards; now $3,000 in OP prizes.

Whether you’re an experienced perps degen or just getting started, this is your chance to compete, climb the leaderboard, and earn real rewards.

Format

Every week, from Monday at 00:00 CET to Sunday at 23:59 CET, the trading competition will be live.

Trade on Mode Trade during the week, battle for volume, make successful trades, and if you rank among the top 15 traders by volume or top 5 by weekly PnL, you’ll win.

Main Leaderboard Prizes

The top 15 traders by volume each week will receive:

1st - $500

2nd - $300

3rd - $200

4th to 15th - $100

The top 5 traders with the highest PnL each week will receive:

1st - $300

2nd - $200

3rd to 5th - $100

Minimum requirements:

Volume: You must hit at least $50,000 per week in trading volume to qualify for prizes.

PnL: You must hit at least $100 per week positive PnL to qualify for prizes

Prizes

Op rewards will be sent directly to the winning wallets on Tuesday each week.

Track Your Progress:

Check the live leaderboard to monitor your ranking in real time and stay competitive throughout the week:

Mode Trade Leaderboard here.

How to Join:

Mode Trade is free to use, but if you want to unlock advanced AI-powered features like:

The AI Terminal for command-based trading

Synth predictive market data

…you’ll need to stake 400,000 MODE to access the TG Giga staker tier.

Pick your pair. Start trading. Climb the leaderboard.

Start trading on Mode Trade here.

About Mode

Mode is building Full Stack DeFAI, the L2 and Perpetuals Exchange scaling DeFi to billions of users through AI agents and AI-driven financial applications.

Website: mode.network

Community TG Groups: Telegram groups

Forum: Mode Forum

Mode Terminal: AI Terminal

Mode Trade: Trade

Synth: Homepage

Synth Subnet: Miners Live Scores

• OP incentives will continue in Season 5 for veMODE holders to direct to ecosystem apps, however Streamers and Agents will now be eligible for guages

• Stake Mode to earn OP at approx 20-30% APY

• Voting power carries over from Season 4 to Season 5

• Agents, Streamers and apps can redistribute OP to their users

• Insights must be provided into the distribution of incentives to remain gauge-eligible

• Growing Mode and the Superchain: OP rewards aim to fuel tangible growth for Mode and the Superchain

Introduction

Mode’s governance experiment is about to become a lot more fun and engaging, with the introduction of Streamers and Agents. Building on Season 4 which saw experiments in bribe markets, Season 5 will focus on incentivising education and engagement with AI apps and agents building across the ecosystem.

Key Dates

Start May 8th: Epoch 1 voting begins. Streamers and Agents can apply to be added to guages on a rolling basis.

Two-Week Epochs: Voting and incentive distribution follows a bi-weekly cadence throughout Season 5

Key Benefits

1. Introducing Streamers and Agents to veMODE

Streamers who live stream or record videos on Mode Trade, Mode AI Terminal and Mode Apps will be eligible to apply for guages and receive OP incentives.

Agents that deploy to Mode will also be eligible to receive ongoing incentives.

2. Focus on Educational Content for Growth

This move to bring in Streamers and Agents will help align the ecosystem around educational content designed to onboard users to Mode’s ecosystem of innovative AI powered applications.

3. AI Driven Onchain Experiences

Mode is building at the cutting edge of AI and all new initiatives will focus on enhancing AI driven onchain experiences.

New in Season 5

1. Streamers and Agents added to veMODE

Protocols can now rely on Hidden Hand’s established bribe infrastructure, removing confusion around eligibility or claim processes. This translates to frictionless user experiences. Streamers who livestream content about Mode Trade can now apply for gauges.

2. Transparency

All eligible protocols for guages must provide transparency in how they distribute OP incentives in order to remain gauge-eligible, helping the community stay informed and confident about the deployment of rewards.

3. Tokenomics and Sustainability

Season 5 will use OP as the primary incentive mechanism. Apps, Agents and Streamers that secure votes get a slice of OP each epoch. This model not only rewards active participants but also ensures the broader Mode ecosystem remains vibrant.

4. Agentic Staking and Voting

Thanks to Mode’s AI Terminal, MODE holders can now use AI to stake their Mode to veMODE.

As the AI Terminal comes out of Beta expect to see the barriers for entry to be reduced significantly for all users… watch this space!

Conclusion

Season 5 will give Mode the firepower to grow awareness and education of the cutting edge AI apps agents across the ecosystem. By incentivising Streamers and Agents there will be a constant stream of content for crypto users to engage with and new Agentic experiences for them to participate in.

Whether you’re a Streamer aiming to get votes, an Agent builder or a veToken holder, Season 5 welcomes you to a new era of onchain experiences. Join us in shaping the evolution of onchain governance at Mode.

About Mode

Mode is building Full Stack DeFAI, the L2 and Perpetuals Exchange scaling DeFi to billions of users through AI agents and AI-driven financial applications.

Website: mode.network

Community TG Groups: Telegram groups

Forum: Mode Forum

Mode Terminal: AI Terminal

Mode Trade: Trade

TLDR: AI-powered trading is here.

To learn more about the Mode Trade platform and the vision behind it, read our full introduction here.

Mode Trade Perp Competition

To celebrate the launch of Mode Trade, we’re launching a weekly trading competition that will reward our most active early traders with $2,000 in OP prizes every week.

Whether you’re an experienced perp degen or just getting started, this is your chance to compete, climb the leaderboard, and earn real rewards.

Format

Every week, from

Monday at 00:00 CET to Sunday at 23:59 CET, the trading competition will be live.

Trade on Mode Trade during the week, and if you rank among the top 5 traders by volume, you’ll win.

Main Leaderboard Prizes

The top 5 traders each week will receive:

1. $1,000

2. $500

3. $250

4. $150

5. $100

Minimum requirement: You must hit at least $50,000 in trading volume to qualify for prizes.

Prizes will be sent directly to the winning wallets on Tuesday each week.

veMODE Community Kickoff Bonus

We’re also offering a special one-time kick-off bonus just for the veMODE community.

If you're in any of the veMODE Telegram groups:

1. Trade $50,000 or more on Mode Trade

2. Be among the first 20 participants to do this and receive $100 in OP each

3. $100 in OP bonuses will be sent out directly to the winners wallets once the first 20 people have hit $50k volume

That’s it. Fast, simple, and exclusive to the veMODE community.

Track Your Progress

Check the live leaderboard to monitor your ranking in real time and stay competitive throughout the week:

Mode Trade Leaderboard here.

How to Join

Mode Trade is free to use, but if you want to unlock advanced AI-powered features like:

• The AI Terminal for command-based trading

• Synth predictive market data

…you’ll need to stake 400,000 MODE to access the TG Giga staker tier.

Ready to Trade?

Pick your pair. Start trading. Climb the leaderboard.

Start trading on Mode Trade here.

About Mode

Website: mode.network

Community TG Groups: Telegram groups

Mode Terminal: AI Terminal

Mode Trade: Trade

Synth: Homepage

Synth Subnet: Miners Live Scores

Why Mode Trade?

Retail traders often lack access to advanced forecasting tools typically used by institutional traders, like predictive analytics and sophisticated simulations. Mode Trade closes the gap by being the first exchange to include predictive AI directly on its platform, giving everyday traders the tools to compete at a higher level.

We built Mode Trade for two reasons:

1. To give traders a competitive edge using predictive AI and smarter insights

2. To empower retail users with accessible education and guidance, helping them explore and understand advanced trading tools - navigating with simple text-based commands.

Mode Trade: Trading Perpetuals with AI

At the core of Mode Trade is our cutting-edge perpetual trading platform powered by Orderly. Traders gain access to over 100+ trading pairs, including today's hottest tokens, with many more coming soon.

The platform's highlights include:

• 100+ available pairs

• Up to 50x leverage

• Deep liquidity thanks to CLOB (central limit order book), developed by Orderly

• Tokenized Stocks (coming soon)

Explore the platform here.

AI Terminal: Command Your Trades

The AI Terminal transforms complex trading activities into simple, text-based commands. This intuitive terminal allows you to:

• Execute seamless onchain actions like bridging, swapping, and staking.

• Access predictive insights through integrated Synth forecasts.

• Effortlessly open, adjust, monitor, and close positions via clear, straightforward commands.

With AI Terminal, managing your trading strategies becomes intuitive, precise, and highly efficient, significantly enhancing your decision-making and reaction speed in dynamic markets.

Synth: Decentralized Synthetic Asset Data

Synth provides decentralized, synthetic price data generated on Bittensor, delivering precise probabilistic market forecasts. With Synth, you can:

• Improve risk management through accurate market predictions

• Enhance trading strategies across Mode Trade

• Benefit from transparent, competitive model evaluation and continuous data refinement

How to get access to Mode Terminal and Synth?

1. Believe in $MODE

We deeply appreciate our dedicated community. All Giga stakers and above - holding at least 400,000 veMODE - can exclusively enjoy the advanced features of the AI Terminal and Synth data forecasts.

2. Be a trader

We will offer limited-time whitelisting spots for selected traders to access AI features. Exact details and requirements will be announced on X in the coming weeks.

Mode Trade unifies powerful perpetual trading capabilities, advanced AI-driven tools, and groundbreaking synthetic data solutions, offering traders unmatched precision, efficiency, and market insight.

Join the future of trading today.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Community TG Groups: Telegram groups

Mode Terminal: AI Terminal

Mode Trade: Trade

Synth: Homepage

Synth Subnet: Miners Live Scores

Bitcoin Price Action

During the week of March 3rd to March 10th, Bitcoin experienced significant volatility, with fluctuations in volatility regimes and an overall downward trend. The price started the week at approximately $94,300 and ended at around $80,700, marking both the weekly high and low.

These price movements were largely driven by macroeconomic and geopolitical uncertainties, affecting not only the cryptocurrency market but also traditional financial markets. The most notable volatility spike occurred around midnight (UTC) on March 7th, coinciding with the U.S. president's executive order on the BTC strategic reserve.

Miners Performance

For this analysis, we examined the performance of six miners: the top three according to leaderboard scores at the beginning of the period (midnight, March 3rd) and the top three at the end of the period (11:59 PM, March 9th). The selected miners were:

• Best miners at midnight on March 3rd: Miners 27, 197, and 240.

• Best miners at 11:59 PM on March 9th: Miners 90, 92, and 170.

This commentary provides an overview of these miners' performance throughout the week, particularly in relation to the volatility trends discussed earlier. Our analysis focuses on key metrics such as the Continuous Ranked Probability Score (CRPS), Softmax scores, and Leaderboard values. Miners who were top performers at the beginning of the period are represented in shades of red, while those who emerged as top performers at the end are shown in shades of green. Scores were recorded at four equidistant time points per day.

Continuous Ranked Probability Score (CRPS)

Softmax scores provide a clearer comparison by applying a transformation to the CRPS values relative to all miners. Using a new β coefficient of -0.002, we amplified small differences in CRPS scores.

As a result, Miners 90, 92, and 170 consistently outperformed their peers in most prompts throughout the week. Miners 170, 90, and 92 exhibited the most stability in achieving high scores, except during the second prompt on Monday, March 3rd, and the second prompt on Thursday, March 6th, where Miner 27 outperformed the others. Notably, all analyzed miners performed poorly on Sunday, March 9th, with scores close to zero.

Softmax Scores

Softmax scores offer a clearer view of how miners performed relative to one another, as they consist of a softmax transformation of the CRPS values of the single miners vs. all the other miners. Using the new β coefficient equal to -0.002 for the calculation of the scores contributes to magnifying relatively small differences in CRP scores between miners. Unlike CRPS values, higher Softmax scores denote better performance.

Because of what just outlined, we can observe how Miners 90, 92, and 170 consistently received higher scores than other miners for almost all of the week’s prompts. Miners 170, 90, and 92, on the other hand, seemed to have a bit more consistency in getting the best scores (with the exception of the second observed prompt on Monday 3rd and the second prompt of Thursday 6th, where Miner 27 was assigned higher scores), most of the time as a “lagged effect” of significant volatility spikes. On the last day of the week, that is, Sunday 9th, all miners under analysis performed poorly, receiving scores very close to 0.

Leaderboard Scores

Consistent with the CRPS and Softmax trends, Miners 27, 197, and 240 experienced a significant decline in their leaderboard scores, dropping from above 0.03 at the start of the week to near zero by the end. In contrast, Miners 90, 92, and 170 showed the opposite trend, benefiting from the evolving volatility dynamics.

Takeaways

The past few weeks have been marked by substantial volatility spikes and a persistent downtrend in Bitcoin, signaling weak market sentiment. This environment has exposed certain flaws in the current Synth subnet scoring system. Specifically, miners who incorporate drift in their forecasts or make extreme predictions have, at times, achieved high CRPS scores purely by chance rather than skill.

Due to the Softmax scores being capped at zero on the lower end, some miners have been incentivized to take outsized risks, treating the prediction system like a lottery. To address this, the Synth team is developing a revised scoring framework designed to penalize extreme predictions more effectively. This updated system will leverage CRPS values for calculating smoothed moving average scores, delaying the Softmax transformation to a subsequent step.

As the team finalizes tuning and backtesting this new system, miners are encouraged to refine their forecasting models, focusing on accuracy and robustness rather than relying on sporadic high-risk predictions.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Forum: forum.mode.network

Community TG Groups: Join our Telegram groups

Week From 2025-02-24 To 2025-03-03

Bitcoin Price Action

During the week from February 24th to March 2nd, Bitcoin registered significant volatility. The week opened with a price of $96,277 at midnight on February 24th and closed at $94,274 at 11:59 PM on March 2nd. Bitcoin reached a peak of $96,490 shortly after the week's opening (on February 24th at 12:51 AM) and dropped to a low of $78,268 on February 28th at 8:46 AM. Following this, the price began to recover, returning to levels closer to those at the start of the week.

These sharp price movements caused noticeable volatility, with sharp spikes observed in the first half of the week that drove the price down significantly. However, during the weekend, the price surged again. Notably, an announcement by the U.S. President on March 2nd, around 3:30 PM (UTC) and again at 4:30 PM, in which Bitcoin (along with other cryptocurrencies) was included in a future U.S. strategic reserve, triggered a price surge of approximately 11.6% (from $85,130 to $94,982) within about three hours. The corresponding price movements, along with the log-returns and rolling-volatility during the week, are shown in the figure below. The times of the U.S. strategic reserve announcement are also highlighted in the figure.

Miners Performance

As usual, six miners were selected for the purpose of this commentary: the top three based on the leaderboard scores at the beginning of the period (midnight, February 24th), and the top three at the end of the period (11:59 PM, March 2nd). Therefore, the miners under examination were as follows:

• Top miners at midnight on February 24th: Miners 4, 5, and 196;

• Top miners at 11:59 PM on March 2nd: Miners 27, 197, and 78.

Below, we will discuss the Continuous Ranked Probability Score (CRPS), Softmax scores, and Leaderboard values obtained by these miners during the week under examination. Miners who were top performers at the beginning of the period will be visualized in shades of red, while those who were top performers at the end will be visualized in shades of green. Four equidistant time points per day were selected for plotting the scores.

Continuous Ranked Probability Score (CRPS)

Figure 2 below illustrates the CRPS values scored by the miners during the week. Lower scores indicate better model performance, while higher scores signify poorer performance.

As we observed, Miners 4, 5, and 196 performed relatively well at the beginning of the week, maintaining low CRPS values compared to the other miners. However, as volatility spikes occurred, particularly between February 28th and March 1st, the models of Miners 197 and 78 seemed to respond better to volatility regime shifts. Miners 4, 5, and 196 regained their strong performance once the Bitcoin price stabilized, which happened between March 1st and the beginning of March 2nd.

Softmax Scores

Softmax scores also provide a valuable indicator of miners' performance during periods of high and low volatility. These scores are calculated by normalizing the CRPS values via softmax transformations, offering in this way a clearer view of how miners performed relative to one another. Unlike CRPS values, higher Softmax scores indicate better performance.

Based on the CRPS trends mentioned earlier, we can observe that for most of the prompts, Miners 4, 5, and 196 consistently obtained larger Softmax scores than the other miners, though their results were not exceptional (their Softmax scores never exceeded 0.1 during the week). This is likely because other miners, not considered in this commentary, were assigned better scores. On the other hand, Miners 27, 197, and 78 received lower or similar scores for most of the week’s prompts. However, after favorable prompts, their scores were significantly higher than those of the other miners: Miner 27 peaked above 0.4, Miner 78 around 0.4, and Miner 197 surpassed 0.15. It should be noted that the performance in Softmax scores was not uniform throughout the week; different miners appeared to dominate for specific prompts during short periods.

Leaderboard Scores

As a reminder, Leaderboard scoring is the step to include an exponential moving average (EMA) of the normalized scores to balance short-term and long-term miner performance.

All miners under examination experienced an initial period of downward trending leaderboard scores, likely in contrast to miners not included in this analysis. However, as the week progressed, the Softmax scores assigned to Miners 27, 78, and 197 in the second half of the week significantly boosted their leaderboard scores. Miners 4 and 5, on the other hand, saw their scores continue to deteriorate during the weekend, and, more slowly, Miner 196's score also declined, although it stabilized around 0.01. Miners 197 and 78 showed similar trends, with both experiencing an upward trajectory in their scores, peaking around February 28th (with some lag between them), but their scores decreased again during the weekend. Miner 27, however, saw a significant increase in their scores starting on March 1st and became the leader by the end of the week, with a score of 0.04.

Takeaways

The continuous shifts in volatility regimes of Bitcoin prices during the week caused significant changes in the leadership of the subnet. This was especially visible in the prompt-specific Softmax scores, where the best-performing miners seemed to dominate for isolated episodes rather than maintaining consistent performance across the entire week. As a result, leaderboard scores also shifted frequently, with miners who led during certain periods seeing their scores decline in favor of others.

We also observed that miners are quickly adapting to the new incentive regime following the CRPS scoring update. The fact that some miners have managed to catch up to the established leaders is a testament to how the more competitive environment is driving miners to enhance their performance.

Several factors contributed to these fluctuations:

• Changes in scoring parameters (for example, β=-0.005 for Softmax scores), which stresses the performance of models with lower CRPS scores.

• Week-specific volatility that made the Bitcoin price trend particularly challenging to forecast.

• Miners’ models that struggled to adapt quickly to the volatility swings and changing volatility regimes.

The Synth subnet team is continuously working to improve the scoring mechanism, ensuring that it more accurately reflects the models' ability to predict Bitcoin's price movements. Additionally, we highly value the suggestions and feedback from our miners and are committed to making ongoing improvements to provide a more consistent and rewarding experience for all participants. We encourage miners to keep refining their models and look forward to collaborating with the community to drive innovation and enhance performance over time.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Forum: forum.mode.network

Community TG Groups: Join our Telegram groups

Bitcoin Price Action

From February 17th to February 23rd, the price of Bitcoin opened at approximately $96,073 and closed at $96,263. During this period, the price reached a minimum of $93,381 on February 18th and a maximum of $99,480 before retracing. While the overall volatility seemed more contained compared to previous weeks, there were still significant price fluctuations. Notably, on Friday, February 21st at 2:30 PM UTC, a sharp price drop occurred, with Bitcoin falling from over $99,000 to around $95,000. This price decline was primarily attributed to two key factors:

1. The ByBit exchange hack announcement, where the platform reported a loss of approximately $1.4 billion in Ethereum;

2. A substantial sell-off on Binance, involving around 1,200 BTC, which triggered leverage position liquidations and further contributed to the price decline.

This sell-off moment has been highlighted in the figures below, showing the Bitcoin price action, log-returns, and 5-minute rolling volatility.

Miners Performance

For this analysis, we selected six miners: the top three according to the leaderboard scores at the beginning of the period (midnight, February 17th), and the top three at the end of the period (11:59 PM, February 23rd). This resulted in the following selection:

• Best miners at midnight on February 17th: Miners 4, 58, and 140.

• Best miners at 11:59 PM on February 23rd: Miners 4, 5, and 10.

Note that Miner 4 appeared in the top 3 both at the start and at the end of the period, so this commentary will focus on a total of five miners. What follows is a brief discussion of their performance during the selected period, particularly in relation to the volatility regimes described above. The analysis will include metrics such as Continuous Ranked Probability Score (CRPS), Softmax scores, and Leaderboard values. Miners who were the top performers at the beginning of the period will be visualized in shades of red, while those who were the top performers at the end will be visualized in shades of green. Four equidistant time points per day were selected for plotting the scores.

Continuous Ranked Probability Score (CRPS). Figure 2 below illustrates the CRPS values scored by the miners throughout the week. In this case, lower scores represent better model performance, while higher scores indicate poorer performance.

From the figure, it's evident that unexpected jumps in price volatility (or log-returns) led to an increase in CRPS scores, before stabilizing once the price settled. Models of the miners that led at the start of the period (Miners 58 and 140, marked in red) were consistently outperformed by the miners who were on top of the leaderboard at the end (Miners 4, 5, and 10, marked in green), except for a brief period between February 20th and February 21st. However, the significant Bitcoin sell-off described earlier triggered a sharp change in volatility, and Miners 4, 5, and 10 seemed to adapt better to the new volatility regime than Miners 140 and 58.

Softmax Scores

Softmax scores also provide a useful indicator of miners' performance during periods of high and low volatility. These scores offer a clearer view of how miners performed relative to one another. Unlike CRPS values, higher Softmax scores indicate better performance.

Additionally, it is important to note that at the beginning of the week, the β parameter that governs the calculation of Softmax scores was decreased from -0.001 to -0.002, which amplified the relative differences between miners.

Throughout most of the week, Softmax scores were fairly consistent, remaining between approximately 0 and 0.01. However, there were moments when Miners 4, 5, and 10 outperformed Miners 58 and 140. The first such instance occurred between February 18th and February 19th, when the scores of Miners 58 and 140 stayed close to 0, while the scores of the other miners rose to around 0.02. The second instance occurred between February 21st and February 22nd, when the scores of Miners 5 and 10 surged to a remarkable 0.12, with Miner 4 also performing well with scores near 0.08.

Both of these performance spikes occurred right after moments of increased volatility in log-returns, notably on February 18th and February 21st, suggesting that Miners 4, 5, and 10 had a more responsive and adaptive model that could adjust to sudden volatility shifts.

Leaderboard Scores

Miners 58 and 140 saw their leaderboard scores drop significantly early in the week, falling from values between 0.15 and 0.2 to near 0 within just 48 hours. In contrast, Miners 4, 5, and 10, after experiencing a simultaneous but less dramatic drop in leaderboard scores at the start of the week, quickly overtook 58 and 140 on February 18th. They maintained this lead throughout the remainder of the week.

This shift reflects the ability of these models to adapt better to the changing volatility forecasts, as evidenced by the rise in their leaderboard scores, which climbed from between 0.005 and 0.01 on February 19th to above 0.02 on February 22nd.

Takeaways

Bitcoin's volatility during this week was more subdued compared to previous weeks. However, key moments of volatility change, such as the BTC sell-off on February 21st, proved crucial for miners' performance. Miners whose models could adapt quickly to even subtle shifts in volatility regimes stood out in the final Softmax and Leaderboard scores. This can be attributed to two factors:

(a) Changes in the scoring parameters, such as the decrease of β for Softmax scores and the extension of the leaderboard scores' half-life to 2 days;

(b) A more competitive environment where top-performing miners are continuously refining their models to remain adaptable in the face of unexpected news and volatility spikes, in line with the spirit of the Synth subnet competition.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Forum: forum.mode.network

Community TG Groups: Join our Telegram groups

Bitcoin Price Action

During the two weeks from February 3rd to February 16th, Bitcoin's price action was mostly characterized by moments of low volatility, with prices ranging between $96,000 and $98,000 for most of the time considered.

However, two key moments of higher volatility can be identified even during this apparent calm in the market. The first occurred on the very first day under examination (Feb 3rd), when the price was most likely still influenced by President Donald Trump's U.S. tariff announcement just before the previous weekend. During this day (February 3rd), the price ranged between approximately $91,000 and $102,000.

The second significant moment of price volatility occurred at around 1:30 PM (UTC) on February 12th, when the U.S. CPI index data was disclosed. The results led to a spike in volatility, as the price dropped from approximately $96,000 to $94,000 within ten minutes. It is worth noting that the release time of the CPI data was known in advance, meaning that miners incorporating this information into their models—anticipating a step jump in volatility—could have scored particularly well.

The plots below depict price action (1-minute timeframe), log returns (1-minute timeframe), and rolling volatility (5-minute window) for the period under consideration. In particular, the log returns and rolling volatility plots allow us to observe the difference in volatility between the moments described above and the rest of the market days. Furthermore, the exact minute when the CPI index data was released has been marked to highlight the observed spike in volatility.

Miners Performance

For this commentary, we have selected six miners: the three best miners according to leaderboard scores at the beginning of the period (midnight, February 3rd) and the three best miners by leaderboard scores at the end of the period (11:59 PM, February 16th). This resulted in the selection of the following miners:

• best miners at midnight on February 3rd: miners 5, 10, and 11;

• best miners at 11:59PM on February 16th: miners 4, 58, and 140.

What follows is a short commentary on the miners' performance during the selected period, concerning the volatility regimes just described. In particular, there will be a discussion of Continuous Ranked Probability Score (CRPS), Softmax, and Leaderboard values. Miners who were the best performers at the beginning of the period will be visualized in shades of red in the plots, while miners who were the best performers at the end of the period will be visualized in shades of green.

To facilitate visualization, we selected four periods per day that are roughly equidistant from each other and plotted the scores at those times. Additionally, for visualization purposes, some of the plotted CRPS values have been clipped at the 95th percentile of the observed scores, as some miners were given huge scores that would have rendered the rest of the plot meaningless.

Continuous Ranked Probability Score (CRPS)

In general, it is possible to observe how higher volatility periods are associated with larger CRPS values. It is important to highlight that lower CRPS values indicate better performance for a given prompt. However, higher CRPS values across all miners during certain periods do not necessarily imply that all miners were performing worse. Instead, these were just periods of higher volatility, which naturally results in larger CRPS values.

On the very first day, when BTC price volatility was higher than in the following days, the miners' CRPS values were all between the 60K and 70K range. As the price volatility returned to more regular levels, the miners' values also tended to decrease, going as low as below 20K points. When the U.S. CPI data was revealed, the increase in volatility caused the CRPS values to increase again, this time reaching the 40K-50K range.

While at this point Miners 5, 10, and 11 still seemed to be performing better than the other miners, it was during the subsequent period, when volatility decreased again, that the latter miners outperformed the former ones. This could potentially be due to models used by Miners 4, 58, and 140 that were better adapted to the regime switch from higher to lower volatility or that were better fine-tuned during the latter days of the competition. It is worth noting, however, that all miners obtained lower CRPS values during the last days, ranging between 10K and 15K. This might be a sign of further model improvement during low-volatility periods.

Softmax Scores

Being a simple mathematical transformation of the CRPS values, the Softmax scores also provide a good reflection of the miners' varying performances across periods of high and low volatility. Unlike with the CRPS values, in this case, higher scores indicate better-performing miners.

In particular, we can observe how the Softmax scores tend to be close to—or lower than—0.01. However, for some miners, these scores rose above 0.1 as volatility decreased between February 5th and February 9th.

Another noteworthy aspect was that the forecasts provided by Miner 11 appeared to adapt better than those of other miners to the change in volatility regime that occurred on February 12th due to the release of the CPI index data. This miner achieved a Softmax score close to 0.06 in the second half of the day, while all other miners received scores between 0 and 0.02. Finally, as observed with the CRPS values, a return to a lower-volatility state, combined with possible model improvements, led miners 140 and 58 to outperform their competitors, while obtaining relatively low scores compared to the higher scores observed in previous days.

Leaderboard Scores

Miners 5, 10, 11, and 4 were among the best-performing miners for most of the period under consideration. In fact, their EWMA scores remained between 0.03 and 0.07 during the first week under review. However, due to increased competition and possible improvements in other miners' models, their scores began to decline during the second week (February 10th to 17th). During this period, their scores dropped to values between 0.01 and 0.02, while new miners who joined the competition later, such as Miners 140 and 158, quickly caught up within a matter of days and even managed to take the lead (along with Miner 4) around the last day of the competition (Sunday, February 16th).

Takeaways

The price of Bitcoin has remained largely stable throughout the two weeks under analysis, though notable spikes in volatility were observed on February 3rd and February 12th. During these periods of stability, miners generally performed better, thriving in the low-volatility environment. However, their performance tended to decline when volatility increased—not necessarily because they were performing worse, but simply due to the nature of higher volatility causing increased CRPS values. Additionally, since the competition is still in its early stages, miners are likely still refining and optimizing their models.

This highlights a key challenge for Synth miners: the need to develop and fine-tune models capable of anticipating or rapidly adapting to sudden changes in volatility, rather than relying solely on straightforward price forecasting. Moving forward, the ability to navigate volatility with agility will be a critical factor for success in this competitive space.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Forum: forum.mode.network

Community TG Groups: Join our Telegram groups

In the whitepaper linked below, we demonstrate that it is possible to generate useful synthetic data in a decentralized manner, using the Bittensor ecosystem and land the use case in DeFi Agents. We argue that a synthetic data layer will become a crucial step for DeFi to reach mass adoption, through autonomous and evolving AI Agents.

Synth is a pioneering subnet on the Bittensor network, designed to provide high-fidelity synthetic price data for the emerging agentic economy. By addressing the limitations of real-world datasets, Synth creates a robust foundation for AI-driven decision-making, transforming DeFi workflows and expanding into new industries.

This whitepaper explores Synth’s architecture, capabilities, and vision to establish a comprehensive data layer for predictive intelligence in DeFi and beyond.

Read the full whitepaper here: Synth Subnet Whitepaper

Interested in contributing?

Check out our complete guide on how to setup your first model.

In the guide linked below, we break things down into three key areas:

1. Explaining the Scoring System: What’s the methodology behind how submissions are evaluated?

2. Tutorial for Miners: How to set up and submit Bitcoin price predictions using a simple Geometric Brownian Motion (GBM) model.

3. Sandboxed Competition Results: A look at how miners using similar models, like GBM, perform against each other.

Complete Guide here: How To Build A Simple Model and Start Earning TAO

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Website: mode.network

Forum: forum.mode.network

Community TG Groups: Join our Telegram groups

• Up to 2M in OP incentives are available in Season 4 for veMODE holders to direct to ecosystem protocols.

• Stake Mode to earn OP and ecosystem bribes

• Voting power carries over from Season 3 to Season 4.

• Bribe Marketplace: Protocols can tap Hidden Hand to offer bribes, streamlining distribution and boosting user engagement.

• Greater Protocol Transparency: Protocols must provide more insight into their distribution of incentives to remain gauge-eligible.

• Onchain Activity: OP rewards aim to fuel tangible growth; liquidity, user adoption, and deeper DeFi integration.

• Agentic Staking and Voting: the infrastructure for Agents to both Stake and Vote in governance is now in place via the Great Onchain Agentic Toolkit (GOAT), allowing builders to deploy true Governance Agents on Mode.

Season 4 brings bribes to center stage, giving protocols and voters a powerful new way to align interests and accelerate Mode’s on-chain expansion.

Key Dates

January 9-15: Protocol documentation submissions, ensuring compliance and gauge eligibility prior to the first epoch’s voting.

January 15 (23:00 UTC): Staking deadline to be eligible to vote in S4 Epoch 1.

January 16 (00:00 UTC): Season 4 Epoch 1 Voting Window Begins

Two-Week Epochs: Voting and incentive distribution follow a bi-weekly cadence throughout Season 4

Key Benefits

1. A Unified Bribe Platform

In Season 3, bribes often took the form of isolated airdrops. Season 4 integrates with Hidden Hand, allowing protocols to deposit bribes in one place and voters to track and claim them without juggling multiple channels.

2. Stronger Alignment, Less Hassle

By offering bribes directly, protocols can reward those who support their vision, while voters gain a straightforward avenue to monetize their voting power. This dynamic fosters win-win collaboration.

3. Fostering On-Chain Growth

All OP incentives must be directed into tangible on-chain usage; such as liquidity incentives, yield opportunities, or trading volume - reinforcing Mode’s goal of building an active and scalable L2 DeFi ecosystem.

Introduction

Mode has always seen governance as an evolving experiment; releasing it in seasons ensures the system can adapt quickly. Building on Season 3’s introduction of veMODE and gauge voting, Season 4 shifts attention toward a robust bribe marketplace that makes it easier than ever for protocols to attract voters and for voters to unlock added rewards.

Bribes may sound controversial, but in decentralized finance, they serve as a transparent mechanism for protocols to compete for votes. Rather than distributing incentives arbitrarily, protocols reward voters who support them, encouraging genuine alignment with ecosystems that demonstrate real utility.

Background

From Curve’s veCRV to Convex Finance to Solidly, bribes have proven a mainstay in DeFi governance; a credible way for projects to build strategic relationships with active voters. Season 3 experimented with a mix of distribution approaches and discovered that a centralized bribe marketplace can significantly streamline interactions.

In Season 4, Hidden Hand plays that central role:

• Protocols deposit bribe tokens and set conditions for payout.

• Voters see an aggregated list of bribe offers, choose which protocols to support, and claim bribes when the voting window ends.

No more navigating multiple chats, communication channels and unofficial spreadsheets. Voters have a one-stop shop to compare bribe offers, thereby concentrating on the strategies that benefit them most.

New in Season 4

1. Easy Bribe Discovery and Claiming

Protocols can now rely on Hidden Hand’s established bribe infrastructure, removing confusion around eligibility or claim processes. This translates to frictionless user experiences.

2. Greater Protocol Transparency

Protocols must provide greater transparency in how they distribute OP incentives in order to remain gauge-eligible, helping the community stay informed and confident about the deployment of rewards.

Tokenomics and Sustainability

Season 4 continues to use OP as the primary incentive mechanism. Protocols that secure votes get a slice of OP each epoch. This model not only rewards active participants but also ensures the broader Mode ecosystem remains vibrant:

• Protocols can direct OP into liquidity mining, yield farming, or unique DeFi strategies to encourage ecosystem growth.

• Bribes let them refine that outreach further, offering direct payouts to the voters who shape governance outcomes.

Over the long term, Mode plans to transition to a more self-sustaining treasury approach, but Season 4’s primary focus is maximizing immediate on-chain engagement through transparent incentives.

Agentic Staking and Voting

Thanks to the work completed in Mode’s recent AI Agent Hackathon, agent applications can now stake and vote through the G.O.A.T framework.

This unlocks even more opportunities for interesting onchain games and simplified, abstracted user experiences.

Conclusion

Season 4 sets the stage for bribes as a central growth engine within the Mode ecosystem. By centralizing bribe offers through Hidden Hand, protocols can laser-focus on capturing voter support, while voters gain real financial upside from aligning with the best protocol initiatives. This renewed emphasis on bribes and transparent governance looks to make Season 4 Mode’s most dynamic governance phase yet.

Whether you’re a protocol aiming to sway votes or a veToken holder excited to capitalize on bribes, Season 4 invites you to embrace this new synergy. Join us in shaping the evolution of on-chain governance at Mode.

About Mode

Mode is building Full Stack DeFAI, the L2 scaling DeFi to billions of users through AI agents and AI-driven financial applications. Bolstered by a $6 million grant from Optimism, Mode continues to push for an open, efficient, and inclusive financial future on the Superchain.

Forum: forum.mode.network

Website: mode.network

Community TG Groups: Join our Telegram groups

The initiative introduces the Synth Subnet Testnet, which leverages Bittensor’s decentralized intelligence network to create the world's most powerful synthetic data, starting with BTC price.

This unlocks a powerful tool for AI Agents deployed on Mode. Up until now, the forecasting capabilities of AI Agents have been limited due to a lack of useful data to train on. Synth changes this by incentivising miners to continuously submit synthetic models which capture full distribution of possible price movements and their associated probabilities, creating the world’s most accurate synthetic data.

The project’s goal is to become the go-to location for asking questions on future price probability distributions.

What is a subnet?

Bittensor operates through a network consisting of smaller, specialized sub-networks (subnets), each dedicated to different areas of AI. Synth will join this network on testnet and leverage  Bittensor’s sophisticated TAO economic model that incentivizes the production of high-value AI subnets. In the future, $MODE will be incorporated into a wider model for our AI Agents.

Initially, Synth will see miners generate 100 simulated price paths for BTC every five minutes. Miners will be evaluated on how accurately they model the probability distribution of future prices and their ability to encapsulate realistic price dynamics, such as volatility clustering and skewed fat-tailed price change distributions. Validators will score miners by comparing their outputs to Pyth’s historical price data using a detailed scoring methodology.

Why is this important for Mode?

Mode is building this Subnet to improve the forecasting and probabilistic reasoning of Agents and LLMs across decentralized finance.

Initially it is expected that Synth’s models will be used for training AI Agents, options pricing models and portfolio management. When the subnet is launched on Mainnet applications and APIs will be built for builders to integrate with.

Synth is designed to push the boundaries of what AI Agents can achieve, through continuously improving synthetic data. For more information, visit the GitHub: github.com/mode-network/synth-subnet.

A dedicated discord channel has been created in the Bittensor discord for discussion.

About Mode

Mode is the Layer 2 (L2) building at the intersection of DeFi and AI. Mode’s mission is to scale DeFi to billions of users through onchain agents and AI-powered financial applications. Backed by a $6 million grant from Optimism, our mission is to build a more open, efficient, and inclusive financial future.

To learn more and get involved with the community and core team, check out our community telegram groups