> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mrdoge.co/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Explained

> How Mr. Doge's AI identifies value betting opportunities

## Overview

Mr. Doge uses advanced machine learning to analyze sporting events and identify value betting opportunities. Our AI evaluates team performance, match context, and market odds to generate data-driven recommendations.

<Info>
  **Transparency First**: Every recommendation includes clear reasoning you can review. We show our work so you can make informed decisions.
</Info>

## What the AI Analyzes

Our system evaluates multiple factors for each event:

**Team Performance:**

* Recent form and results
* Home vs away performance splits
* Head-to-head history
* Scoring and defensive patterns

**Match Context (Priority Leagues):**

* Team news and lineup information
* Weather conditions
* Tactical matchups

**Market Analysis:**

* Current odds evaluation
* Market efficiency assessment
* Value opportunity detection

<Note>
  **Priority Competitions**: Major leagues (Premier League, La Liga, Champions League, NBA, etc.) receive enhanced analysis with real-time context and deeper research.
</Note>

***

## Understanding Confidence Levels

Every recommendation has a confidence rating based on statistical edge:

### High Confidence (Edge > 5%)

* **What it means**: Strong statistical advantage over market odds
* **Expected accuracy**: 60-65% win rate
* **When it happens**: Clear form advantage, market inefficiency, or strong tactical mismatch

### Medium Confidence (Edge 3-5%)

* **What it means**: Moderate statistical advantage
* **Expected accuracy**: 55-60% win rate
* **When it happens**: Slight market inefficiency or minor form advantage

### Low Confidence (Edge \< 3%)

* **What it means**: Marginal edge, higher risk
* **Expected accuracy**: 50-55% win rate
* **Note**: We rarely recommend Low confidence bets

<Warning>
  **No Guaranteed Winners**: Even High confidence picks can lose. Confidence reflects statistical edge, not certainty. Proper bankroll management is essential.
</Warning>

***

## Kelly Criterion & Stake Sizing

The AI calculates optimal stake sizes using the **Kelly Criterion**, a mathematical formula for maximizing long-term growth.

### How It Works

```
Kelly % = (Edge × Odds - 1) / (Odds - 1)
```

**Example:**

* Odds: 2.10
* Edge: 6.8%
* Kelly = 3.2% of bankroll

### Fractional Kelly (Recommended)

We recommend **10% fractional Kelly** for safer bankroll management:

* Full Kelly: 3.2% of bankroll
* 10% Fractional Kelly: **0.32%** of bankroll

**Why Fractional Kelly?**

* Reduces volatility
* More conservative approach
* Protects against estimation errors
* Still captures most of the edge

### Practical Example

**Your Bankroll: \$1,000**

**Recommendation:**

* Odds: 2.10
* Edge: 6.8%
* Kelly Fraction: 0.032 (3.2%)

**Stake Calculation:**

* Full Kelly: $1,000 × 0.032 = $32
* 10% Fractional Kelly: $32 × 0.10 = **$3.20\*\*

<Tip>
  **Start Small**: If you're new to value betting, consider using 5% fractional Kelly (half our recommendation) until you're comfortable with variance.
</Tip>

***

## Reading Recommendations

Every recommendation includes three key sections:

### Rationale

Why the AI picked this bet:

```json theme={null}
"rationale": [
  "Home team's excellent form (8 wins in last 10)",
  "Away team struggling defensively (10 goals conceded in 3 games)",
  "Strong home advantage (75% win rate at home)"
]
```

**Look for:**

* ✅ Data-backed claims with recent statistics
* ✅ Specific tactical insights
* ✅ Clear logical reasoning

### Risk Factors

Reasons why the pick might not win:

```json theme={null}
"riskFactors": [
  "Away team has key players returning from injury",
  "Weather forecast shows heavy rain"
]
```

**Use this to:**

* Assess if the risk outweighs the edge
* Decide whether to skip the bet
* Adjust stake size if unsure

### Data Support

Statistical basis for the recommendation:

```json theme={null}
"dataSupport": [
  "Simulation gives 52% home win probability vs 47.6% implied by odds",
  "Expected goals: Home 1.8, Away 1.1"
]
```

This shows you the math behind the recommendation.

***

## Market Efficiency Score

The AI calculates a **Market Efficiency Score** (0-100) for each event:

* **90-100**: Extremely efficient, rare value opportunities
* **80-89**: Efficient market, requires deep analysis
* **70-79**: Moderately efficient, value exists
* **Below 70**: Inefficient market, multiple opportunities

**What affects efficiency:**

* League popularity (Premier League more efficient than lower tiers)
* Time until kickoff (closer = more efficient)
* Public attention (high-profile matches = more efficient)

***

## Mr. Doge Picks vs Betting Recommendations

### Betting Recommendations

**What they are:**

* Individual value bets for specific markets
* Available for all analyzed events
* Confidence-rated (Low/Medium/High)
* Single-bet recommendations

**Access:** `GET /v2/ai/betting-recommendations` (2 credits)

### Mr. Doge Picks

**What they are:**

* Curated picks combining multiple recommendations
* Featured selections for maximum value
* Can be single events or parlays
* Pre-calculated combined odds

**Access:** `GET /v2/ai/mrdoge-picks` (10 credits)

<Tip>
  **For Beginners**: Start with individual Betting Recommendations to understand how the AI works. Graduate to Mr. Doge Picks once you're comfortable.
</Tip>

***

## Limitations & Disclaimers

### What the AI Can't Predict

* **Referee decisions**: Red cards, penalties, VAR calls
* **Individual brilliance**: Wonder goals, goalkeeper errors
* **Motivational factors**: Team morale, manager pressure
* **Random variance**: Bad luck, post hits, own goals

### Past Performance ≠ Future Results

The AI analyzes historical data, but:

* Teams change (injuries, form shifts, new players)
* Tactics evolve
* External factors matter (weather, referee, crowd)
* Variance is inherent in sports

<Warning>
  **No Guarantee of Profit**: Value betting requires discipline, bankroll management, and patience. Short-term losses are normal even with positive expected value. Only bet what you can afford to lose.
</Warning>

***

## FAQs

<AccordionGroup>
  <Accordion title="How often are recommendations updated?" icon="clock">
    AI analysis runs continuously:

    * New events analyzed as they become available
    * Recommendations refresh when odds change significantly
    * Enhanced analysis runs 24-48 hours before kickoff for priority matches
  </Accordion>

  <Accordion title="Why some matches are missing recommendations?" icon="question">
    Possible reasons:

    * Data quality doesn't meet standards
    * Event too far in the future (odds not stable)
  </Accordion>

  <Accordion title="Can I trust High confidence picks?" icon="shield-check">
    High confidence picks have a **statistical edge**, meaning over many bets they should be profitable. However:

    * Any single pick can lose (variance is normal)
    * We estimate 60-65% win rate for High confidence
    * Requires proper bankroll management
    * Long-term discipline is essential
  </Accordion>

  <Accordion title="How is this different from tipsters?" icon="users">
    Key differences:

    * **Transparent**: Full reasoning provided for every pick
    * **Data-driven**: Based on statistics, not gut feeling
    * **Systematic**: Same process applied to every event
    * **Accountable**: Track record is verifiable via API
    * **No bias**: AI doesn't have favorite teams
  </Accordion>

  <Accordion title="What sports does the AI cover?" icon="futbol">
    Currently AI recommendations are available for:

    * ⚽ Soccer (all major leagues)

    Other sports receive odds data but not yet AI analysis.
  </Accordion>
</AccordionGroup>

***

## Next Steps

<CardGroup cols={2}>
  <Card title="Data Model" icon="diagram-project" href="/concepts/data-model">
    See BettingRecommendation data structure
  </Card>

  <Card title="Bet Settlement" icon="check-circle" href="/concepts/bet-settlement">
    Learn how bets are settled and tracked
  </Card>

  <Card title="API Reference" icon="code" href="/api-reference/ai/recommendations">
    Access AI recommendations via API
  </Card>

  <Card title="Dashboard" icon="chart-line" href="https://mrdoge.co/dashboard">
    View recommendations in your dashboard
  </Card>
</CardGroup>