{"id":2552,"date":"2026-04-01T10:00:00","date_gmt":"2026-04-01T10:00:00","guid":{"rendered":"https:\/\/oddspapi.io\/blog\/?p=2552"},"modified":"2026-03-25T13:27:04","modified_gmt":"2026-03-25T13:27:04","slug":"backtest-betting-model-free-historical-odds","status":"publish","type":"post","link":"https:\/\/oddspapi.io\/blog\/backtest-betting-model-free-historical-odds\/","title":{"rendered":"How to Backtest a Betting Model with Free Historical Odds (Python Tutorial)"},"content":{"rendered":"

Your Betting Model Looks Great on Paper. But Would It Actually Make Money?<\/h2>\n

You’ve built a model. It predicts match outcomes, finds value, maybe even spots arbitrage. But here’s the question you can’t answer without historical data: would it have been profitable over the last 500 games?<\/strong><\/p>\n

That’s backtesting \u2014 running your strategy against real past odds to measure actual performance before you risk real money. The problem? Most APIs charge $79+\/month for historical data. OddsPapi includes it on the free tier<\/strong> \u2014 timestamped odds from 350+ bookmakers including sharps like Pinnacle, Singbet, and SBOBet.<\/p>\n

This tutorial walks you through a complete Python backtesting pipeline: fetch historical odds, get match results, run three strategies, and calculate yield, ROI, and drawdown. Every line of code is tested against live data.<\/p>\n

Why Backtest? (And Why Most Bettors Skip It)<\/h2>\n

Backtesting is standard practice in quant finance. You’d never deploy a trading algorithm without it. But in sports betting, most people skip straight to live betting because historical odds data is either expensive or hard to get.<\/p>\n

A proper backtest tells you three things your model can’t: 1)<\/strong> Whether your edge survives the vig. 2)<\/strong> How bad the drawdowns get. 3)<\/strong> Whether you’re overfitting to noise. Without it, you’re just guessing with extra steps.<\/p>\n

Historical Odds Data: Old Way vs OddsPapi<\/h2>\n
\n\n\n\n\n\n\n\n\n\n\n
Feature<\/th>\nScraping \/ CSV Files<\/th>\nThe Odds API<\/th>\nOddsPapi<\/th>\n<\/tr>\n<\/thead>\n
Historical data<\/strong><\/td>\nDIY \u2014 unreliable<\/td>\n$79\/mo add-on<\/td>\nFree tier included<\/td>\n<\/tr>\n
Bookmakers<\/strong><\/td>\n1-2 you scrape<\/td>\n~40<\/td>\n350+<\/td>\n<\/tr>\n
Sharps included<\/strong><\/td>\nNo<\/td>\nNo<\/td>\nPinnacle, Singbet, SBOBet<\/td>\n<\/tr>\n
Closing line data<\/strong><\/td>\nUnreliable timestamps<\/td>\nYes<\/td>\nYes (timestamped snapshots)<\/td>\n<\/tr>\n
Line movement history<\/strong><\/td>\nNo<\/td>\nLimited<\/td>\nFull opening \u2192 closing timeline<\/td>\n<\/tr>\n
Authentication<\/strong><\/td>\nN\/A<\/td>\nHeader-based<\/td>\nSimple query parameter<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n

The Backtest Pipeline: 6 Steps in Python<\/h2>\n

We’ll backtest three simple strategies on Premier League matches using Pinnacle closing odds as our benchmark. Here’s the full pipeline.<\/p>\n

Step 1: Setup and Authentication<\/h3>\n
import requests\nimport pandas as pd\nfrom datetime import datetime\n\nAPI_KEY = \"YOUR_API_KEY\"\nBASE_URL = \"https:\/\/api.oddspapi.io\/v4\"\n\ndef api_get(endpoint, params=None):\n    \"\"\"Helper to make authenticated API calls.\"\"\"\n    if params is None:\n        params = {}\n    params[\"apiKey\"] = API_KEY\n    response = requests.get(f\"{BASE_URL}\/{endpoint}\", params=params)\n    response.raise_for_status()\n    return response.json()\n\n# Test authentication\nsports = api_get(\"sports\")\nprint(f\"Connected \u2014 {len(sports)} sports available\")<\/code><\/pre>\n
Step 2: Get Completed Fixtures<\/h3>\n
Fetch Premier League fixtures from the current season. We filter to matches that have already been played (statusId=2<\/code> means completed).<\/p>\n
# Fetch Premier League fixtures (tournamentId=17)\nfixtures = api_get(\"fixtures\", {\n    \"sportId\": 10,\n    \"tournamentId\": 17,\n    \"per_page\": 100\n})\n\n# Filter to completed matches\ncompleted = [f for f in fixtures if f.get(\"statusId\") == 2]\nprint(f\"Found {len(completed)} completed Premier League fixtures\")\n\n# Preview\nfor f in completed[:3]:\n    print(f\"  {f['participant1Name']} vs {f['participant2Name']} \u2014 {f['startTime'][:10]}\")<\/code><\/pre>\n
Step 3: Get Match Results<\/h3>\n
For each completed fixture, fetch the final score and determine the outcome (home win, draw, or away win).<\/p>\n
import time\n\ndef get_result(fixture_id):\n    \"\"\"Fetch score and return outcome: 'home', 'draw', or 'away'.\"\"\"\n    try:\n        data = api_get(\"scores\", {\"fixtureId\": fixture_id})\n        result = data[\"scores\"][\"periods\"][\"result\"]\n        home = result[\"participant1Score\"]\n        away = result[\"participant2Score\"]\n        if home > away:\n            return \"home\"\n        elif home < away:\n            return \"away\"\n        else:\n            return \"draw\"\n    except Exception:\n        return None\n\n# Collect results (respect rate limits)\nresults = {}\nfor f in completed:\n    fid = f[\"fixtureId\"]\n    results[fid] = get_result(fid)\n    time.sleep(1)  # Rate limit: ~1 request\/second\n\nprint(f\"Got results for {len([r for r in results.values() if r])} matches\")<\/code><\/pre>\n
Step 4: Get Historical (Closing) Odds from Pinnacle<\/h3>\n
This is the core of the backtest. We fetch timestamped odds snapshots from Pinnacle and extract the pre-match closing line<\/strong> \u2014 the last price before kickoff. Pinnacle's closing line is the industry benchmark for market efficiency.<\/p>\n
def get_closing_odds(fixture_id, start_time):\n    \"\"\"\n    Fetch historical odds and extract pre-match closing prices.\n    Returns dict: {'home': price, 'draw': price, 'away': price}\n    \"\"\"\n    try:\n        data = api_get(\"historical-odds\", {\n            \"fixtureId\": fixture_id,\n            \"bookmakers\": \"pinnacle\",\n            \"marketId\": 101  # Full Time Result (1X2)\n        })\n    except Exception:\n        return None\n\n    markets = data[\"bookmakers\"][\"pinnacle\"][\"markets\"][\"101\"]\n    kickoff = datetime.fromisoformat(start_time.replace(\"Z\", \"+00:00\"))\n\n    closing = {}\n    outcome_map = {\"101\": \"home\", \"102\": \"draw\", \"103\": \"away\"}\n\n    for outcome_id, label in outcome_map.items():\n        snapshots = markets[\"outcomes\"][outcome_id][\"players\"][\"0\"]\n\n        # Filter to pre-match snapshots only\n        pre_match = [\n            s for s in snapshots\n            if datetime.fromisoformat(s[\"createdAt\"]) < kickoff\n        ]\n\n        if not pre_match:\n            return None\n\n        # Closing line = most recent pre-match snapshot\n        # Snapshots are ordered newest-first\n        closing[label] = pre_match[0][\"price\"]\n\n    return closing\n\n# Collect closing odds for all completed fixtures\nodds_data = {}\nfor f in completed:\n    fid = f[\"fixtureId\"]\n    odds = get_closing_odds(fid, f[\"startTime\"])\n    if odds:\n        odds_data[fid] = odds\n    time.sleep(5)  # Historical endpoint has stricter rate limits\n\nprint(f\"Got closing odds for {len(odds_data)} matches\")<\/code><\/pre>\n
Step 5: Build a Backtest Dataset<\/h3>\n
Combine fixtures, results, and closing odds into a single DataFrame \u2014 the foundation for all our strategy tests.<\/p>\n
# Build the dataset\nrows = []\nfor f in completed:\n    fid = f[\"fixtureId\"]\n    if fid in odds_data and results.get(fid):\n        odds = odds_data[fid]\n        rows.append({\n            \"fixture_id\": fid,\n            \"home_team\": f[\"participant1Name\"],\n            \"away_team\": f[\"participant2Name\"],\n            \"date\": f[\"startTime\"][:10],\n            \"result\": results[fid],\n            \"home_odds\": odds[\"home\"],\n            \"draw_odds\": odds[\"draw\"],\n            \"away_odds\": odds[\"away\"],\n            # Implied probabilities (removing vig)\n            \"home_prob\": 1 \/ odds[\"home\"],\n            \"draw_prob\": 1 \/ odds[\"draw\"],\n            \"away_prob\": 1 \/ odds[\"away\"],\n        })\n\ndf = pd.DataFrame(rows)\nprint(f\"Backtest dataset: {len(df)} matches\")\nprint(df[[\"home_team\", \"away_team\", \"result\", \"home_odds\", \"draw_odds\", \"away_odds\"]].head())<\/code><\/pre>\n
Step 6: Run Three Strategies<\/h3>\n
Now the fun part. We'll test three strategies and measure their performance with real metrics.<\/p>\n
Strategy 1: Always Bet the Favourite<\/h3>\n
The simplest possible strategy \u2014 always bet on the outcome with the lowest odds (highest implied probability).<\/p>\n
def backtest_favourite(df):\n    \"\"\"Bet 1 unit on the favourite (lowest odds) every match.\"\"\"\n    bets = []\n    for _, row in df.iterrows():\n        # Find the favourite\n        outcomes = {\"home\": row[\"home_odds\"], \"draw\": row[\"draw_odds\"], \"away\": row[\"away_odds\"]}\n        favourite = min(outcomes, key=outcomes.get)\n        odds = outcomes[favourite]\n\n        # Did it win?\n        won = row[\"result\"] == favourite\n        profit = (odds - 1) if won else -1\n        bets.append({\"match\": f\"{row['home_team']} vs {row['away_team']}\",\n                      \"bet\": favourite, \"odds\": odds, \"won\": won, \"profit\": profit})\n\n    return pd.DataFrame(bets)\n\nfav_results = backtest_favourite(df)\ntotal_staked = len(fav_results)\ntotal_profit = fav_results[\"profit\"].sum()\nyield_pct = (total_profit \/ total_staked) * 100\nhit_rate = fav_results[\"won\"].mean() * 100\n\nprint(f\"=== STRATEGY 1: Always Bet the Favourite ===\")\nprint(f\"Bets placed:  {total_staked}\")\nprint(f\"Hit rate:     {hit_rate:.1f}%\")\nprint(f\"Total profit: {total_profit:.2f} units\")\nprint(f\"Yield:        {yield_pct:.2f}%\")\nprint(f\"Max drawdown: {fav_results['profit'].cumsum().min():.2f} units\")<\/code><\/pre>\n
Strategy 2: Closing Line Value (CLV)<\/h3>\n
A smarter approach: only bet when your model's probability exceeds Pinnacle's implied probability<\/strong>. This is the gold standard for professional bettors. If you consistently beat the closing line, you have an edge.<\/p>\n
def backtest_clv(df, model_edge=0.05):\n    \"\"\"\n    Simulate a model that's slightly better than the market.\n    Bet when your model's implied probability exceeds Pinnacle's by 'model_edge'.\n\n    In practice, replace this with YOUR model's predictions.\n    \"\"\"\n    bets = []\n    for _, row in df.iterrows():\n        outcomes = {\n            \"home\": (row[\"home_odds\"], row[\"home_prob\"]),\n            \"draw\": (row[\"draw_odds\"], row[\"draw_prob\"]),\n            \"away\": (row[\"away_odds\"], row[\"away_prob\"]),\n        }\n\n        for outcome, (odds, implied_prob) in outcomes.items():\n            # Simulate: your model thinks probability is 5% higher than market\n            # Replace this with your actual model predictions\n            model_prob = implied_prob * (1 + model_edge)\n\n            # Only bet if model probability > implied probability (value bet)\n            if model_prob > implied_prob and odds > 1.5:  # Min odds filter\n                won = row[\"result\"] == outcome\n                profit = (odds - 1) if won else -1\n                bets.append({\n                    \"match\": f\"{row['home_team']} vs {row['away_team']}\",\n                    \"bet\": outcome, \"odds\": odds,\n                    \"model_prob\": model_prob, \"market_prob\": implied_prob,\n                    \"won\": won, \"profit\": profit\n                })\n\n    return pd.DataFrame(bets)\n\nclv_results = backtest_clv(df, model_edge=0.05)\nif len(clv_results) > 0:\n    print(f\"=== STRATEGY 2: Closing Line Value (5% Edge) ===\")\n    print(f\"Bets placed:  {len(clv_results)}\")\n    print(f\"Hit rate:     {clv_results['won'].mean() * 100:.1f}%\")\n    print(f\"Total profit: {clv_results['profit'].sum():.2f} units\")\n    print(f\"Yield:        {(clv_results['profit'].sum() \/ len(clv_results)) * 100:.2f}%\")\n    print(f\"Avg odds:     {clv_results['odds'].mean():.2f}\")<\/code><\/pre>\n
Strategy 3: Fade the Public (Sharp vs Soft Line Divergence)<\/h3>\n
Compare Pinnacle (sharp) odds to a soft bookmaker like Bet365. When lines diverge significantly, the sharp book is usually right. This strategy bets with Pinnacle when Bet365 disagrees.<\/p>\n
def get_closing_odds_multi(fixture_id, start_time, bookmakers=\"pinnacle,bet365\"):\n    \"\"\"Fetch closing odds from multiple bookmakers for comparison.\"\"\"\n    try:\n        data = api_get(\"historical-odds\", {\n            \"fixtureId\": fixture_id,\n            \"bookmakers\": bookmakers,\n            \"marketId\": 101\n        })\n    except Exception:\n        return None\n\n    kickoff = datetime.fromisoformat(start_time.replace(\"Z\", \"+00:00\"))\n    result = {}\n    outcome_map = {\"101\": \"home\", \"102\": \"draw\", \"103\": \"away\"}\n\n    for bookie_slug, bookie_data in data[\"bookmakers\"].items():\n        result[bookie_slug] = {}\n        markets = bookie_data[\"markets\"][\"101\"]\n        for outcome_id, label in outcome_map.items():\n            snapshots = markets[\"outcomes\"][outcome_id][\"players\"][\"0\"]\n            pre_match = [s for s in snapshots if datetime.fromisoformat(s[\"createdAt\"]) < kickoff]\n            if pre_match:\n                result[bookie_slug][label] = pre_match[0][\"price\"]\n\n    return result\n\ndef backtest_fade_public(df, completed_fixtures, threshold=0.10):\n    \"\"\"\n    Bet when Bet365 (soft) and Pinnacle (sharp) disagree by > threshold.\n    Bet on the Pinnacle side \u2014 sharps are usually right.\n    \"\"\"\n    bets = []\n    for f in completed_fixtures:\n        fid = f[\"fixtureId\"]\n        row = df[df[\"fixture_id\"] == fid]\n        if row.empty:\n            continue\n        row = row.iloc[0]\n\n        multi_odds = get_closing_odds_multi(fid, f[\"startTime\"])\n        time.sleep(5)\n\n        if not multi_odds or \"pinnacle\" not in multi_odds or \"bet365\" not in multi_odds:\n            continue\n\n        pin = multi_odds[\"pinnacle\"]\n        b365 = multi_odds[\"bet365\"]\n\n        for outcome in [\"home\", \"draw\", \"away\"]:\n            if outcome not in pin or outcome not in b365:\n                continue\n            pin_prob = 1 \/ pin[outcome]\n            b365_prob = 1 \/ b365[outcome]\n\n            # Pinnacle thinks it's MORE likely than Bet365 does\n            if pin_prob - b365_prob > threshold:\n                won = row[\"result\"] == outcome\n                profit = (b365[outcome] - 1) if won else -1  # Bet at Bet365 price\n                bets.append({\n                    \"match\": f\"{row['home_team']} vs {row['away_team']}\",\n                    \"bet\": outcome,\n                    \"pinnacle_odds\": pin[outcome],\n                    \"bet365_odds\": b365[outcome],\n                    \"divergence\": f\"{(pin_prob - b365_prob)*100:.1f}%\",\n                    \"won\": won, \"profit\": profit\n                })\n\n    return pd.DataFrame(bets)\n\n# Note: This strategy makes additional API calls per fixture\n# fade_results = backtest_fade_public(df, completed, threshold=0.10)<\/code><\/pre>\n
Putting It All Together: Performance Summary<\/h3>\n
def print_summary(name, results_df):\n    \"\"\"Print a clean performance summary for any strategy.\"\"\"\n    if results_df.empty:\n        print(f\"{name}: No bets placed\")\n        return\n\n    n = len(results_df)\n    profit = results_df[\"profit\"].sum()\n    cumulative = results_df[\"profit\"].cumsum()\n\n    print(f\"\\n{'='*50}\")\n    print(f\"  {name}\")\n    print(f\"{'='*50}\")\n    print(f\"  Bets placed:    {n}\")\n    print(f\"  Won:            {results_df['won'].sum()} ({results_df['won'].mean()*100:.1f}%)\")\n    print(f\"  Total profit:   {profit:+.2f} units\")\n    print(f\"  Yield:          {(profit\/n)*100:+.2f}%\")\n    print(f\"  Best streak:    {max_streak(results_df, True)} wins\")\n    print(f\"  Worst streak:   {max_streak(results_df, False)} losses\")\n    print(f\"  Max drawdown:   {(cumulative - cumulative.cummax()).min():.2f} units\")\n\ndef max_streak(df, winning=True):\n    \"\"\"Calculate longest consecutive win or loss streak.\"\"\"\n    streak = 0\n    max_s = 0\n    for won in df[\"won\"]:\n        if won == winning:\n            streak += 1\n            max_s = max(max_s, streak)\n        else:\n            streak = 0\n    return max_s\n\nprint_summary(\"Always Bet the Favourite\", fav_results)\nprint_summary(\"Closing Line Value (5% Edge)\", clv_results)<\/code><\/pre>\n
5 Backtesting Mistakes That Will Blow Up Your Results<\/h2>\n
A backtest is only as good as its methodology. Here are the traps that catch even experienced quants:<\/p>\n
\n\n\n\n\n\n\n\n\n\n
Mistake<\/th>\nWhat Happens<\/th>\nHow to Avoid It<\/th>\n<\/tr>\n<\/thead>\n
Look-ahead bias<\/strong><\/td>\nYou use closing odds to decide bets you'd have placed at opening<\/td>\nOnly use data available at decision time. If you bet pre-match, use opening odds for selection and closing for settlement.<\/td>\n<\/tr>\n
Overfitting<\/strong><\/td>\nStrategy has 10 parameters tuned to past data, fails on new data<\/td>\nKeep strategies simple. If your edge disappears when you change one parameter by 5%, it's not real.<\/td>\n<\/tr>\n
Survivorship bias<\/strong><\/td>\nYou only test markets where you found data, ignoring gaps<\/td>\nInclude fixtures where odds were unavailable. A strategy that only works on 30% of matches is fragile.<\/td>\n<\/tr>\n
Ignoring the vig<\/strong><\/td>\nYour model shows +EV but the margin eats the edge<\/td>\nAlways calculate yield after the vig. Pinnacle's 2-3% margin is the best case \u2014 soft books take 5-8%.<\/td>\n<\/tr>\n
Sample size<\/strong><\/td>\n50-game backtest shows +20% yield, you bet the house<\/td>\nMinimum 500 bets for statistical significance. A 55% hit rate needs 1,000+ bets to confirm it's not luck.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n
What Makes OddsPapi Different for Backtesting<\/h2>\n
\n\n\n\n\n\n\n\n\n\n
Feature<\/th>\nWhy It Matters for Backtesting<\/th>\n<\/tr>\n<\/thead>\n
350+ bookmakers<\/strong><\/td>\nCompare sharp vs soft lines across the full market \u2014 not just 40 books<\/td>\n<\/tr>\n
Timestamped snapshots<\/strong><\/td>\nSee exact opening \u2192 closing line movement, not just final price<\/td>\n<\/tr>\n
Sharp bookmakers<\/strong><\/td>\nPinnacle, Singbet, SBOBet closing lines \u2014 the true market benchmark<\/td>\n<\/tr>\n
Free tier<\/strong><\/td>\nHistorical data included at no cost. Competitors charge $79+\/month.<\/td>\n<\/tr>\n
59 sports<\/strong><\/td>\nBacktest across soccer, NBA, MLB, esports, tennis \u2014 same API, same structure<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/figure>\n
Next Steps: Take Your Backtest Further<\/h2>\n
This tutorial gives you the foundation. Here's where to go next:<\/p>\n