Glossary

Open glossary of value betting terms used by Σ Value Sniper. Each definition includes the formula, context, and why it matters for our methodology.

Quick jump CLVEVKellySharp consensusFair lineVigBeat-ratePaper tradingROISharp spreadEloWalk-forwardBootstrap CILine shoppingMarket efficiencyStop-loss2-way / 3-way

CLV — Closing Line Value

[ leading indicator ]

The percentage gap between the odds you took and the closing line at a sharp bookmaker (Pinnacle, Betfair Exchange, Matchbook). Formula: CLV = k_taken / k_close − 1. Positive CLV is the most predictive metric of long-term profitability — it stabilises 5-10× faster than ROI, making it the leading indicator we track.

EV — Expected Value

[ profit per bet ]

Projected long-term profit per bet given your estimated edge. Formula: EV = p × (k − 1) − (1 − p), where p is your true win probability and k is the bookmaker odds. Positive EV is the entry filter — anything below zero is gambling, not value betting.

Kelly Criterion

[ stake sizing ]

Stake size formula maximising long-run log-growth of bankroll: f = (b × p − q) / b, where b = k − 1, p = win probability, q = 1 − p. Σ Value Sniper uses ¼ Kelly with 1.5% bankroll cap — full Kelly is mathematically optimal but psychologically unbearable in real conditions (50%+ drawdowns destroy decision-making).

Sharp Consensus

[ true probability proxy ]

Median of multiple low-margin bookmakers (Pinnacle, Betfair Exchange, Matchbook, smarkets). Used as the best available estimate of true outcome probability. Median (vs single source) eliminates stale lines and single-book noise — if one source is mispricing, it gets outvoted by the other two.

Fair Line

[ de-vigged price ]

The implied price after removing bookmaker margin from sharp consensus. Calculated by normalising sharp implied probabilities so they sum to 1.0 (vs the inflated 1.02-1.10 they sum to with vig). Fair line is the benchmark against which value is measured.

Vig — Vigorish / Margin

[ bookmaker commission ]

The bookmaker's commission baked into odds. Calculated as vig = (1/k_home + 1/k_away) − 1. Pinnacle ~2-3%, Betfair Exchange ~5%, soft bookmakers 5-10%. Lower vig = sharper book = more reliable as fair-line reference.

CLV Beat-Rate

[ statistical signal ]

Percentage of bets with positive CLV. The most statistically tractable signal of edge — at 70%+ beat-rate over 50 bets, p < 0.01 in a binomial test (statistically significant). Σ Value Sniper currently runs at 77.4% beat-rate on N=93 paper bets.

Paper Trading

[ validation phase ]

Recording bets in a database without placing real money — equivalent to algorithmic trading backtesting on live forward data. Σ Value Sniper requires 200+ paper bets with positive CLV before any real-money deployment. Currently at N=93.

ROI — Return on Investment

[ lagging indicator ]

Net profit / total stake. The lagging indicator of betting profitability. Requires N=500-1000 bets to converge within ±5% accuracy. Standard deviation at avg odds 2.1 is approximately odds / sqrt(N) — at N=24 that's ±43%, meaning a profitable strategy can show -43% ROI by pure variance.

Sharp Spread

[ disagreement filter ]

Disagreement between sharp bookmakers expressed in percentage points of implied probability. Σ Value Sniper rejects bets where sharp_spread > 8 pp — high disagreement signals likely news shock or stale line, not exploitable value.

Elo Rating

[ sanity-check model ]

Skill-based rating system originally developed for chess. Σ Value Sniper uses custom Elo trained on 91k+ matches across multi-sport coverage as a sanity-check — sharp consensus drives bet decisions, Elo confirms direction. Glicko-2 was tested but lost in backtest (ADR-009) — Elo wins on this dataset.

Walk-Forward Backtest

[ honest backtesting ]

Backtesting where the model uses only past data at each prediction time — no future leakage. The honest way to test predictive models. Σ Value Sniper used walk-forward with 2015 warmup on tennis dataset to lock filter parameters (ADR-008).

Bootstrap Confidence Interval

[ uncertainty estimation ]

Statistical resampling method for estimating uncertainty in sample statistics. At N=93 bets with ROI +13.7%, the bootstrap CI95 likely still crosses zero — meaning ROI direction is not yet statistically conclusive even though the point estimate is positive.

Line Shopping

[ price optimisation ]

Comparing odds across multiple bookmakers and taking the highest available price for the same outcome. Single biggest skill in value betting after edge identification — going from average prices to best-of-N can add 1-3% to ROI.

Market Efficiency

[ where value lives ]

How quickly a betting market incorporates new information into prices. Premier League, ATP top-100, NBA — highly efficient (small inefficiencies, hard to beat). Lower-tier soccer, niche leagues, women's sports, smaller MMA cards — less efficient, more value opportunities.

Stop-Loss Criteria

[ risk discipline ]

Conditions that trigger pause or halt of betting. Σ Value Sniper rules: CLV per bet < +3% over rolling 50 bets → halt. ROI < −5% over N=200 → review. Pinnacle margin > 5% for 7+ days → pause (market regime shift). Three independent guard rails.

Two-Way / Three-Way Markets

[ market structure ]

Two-way: only 2 outcomes (tennis, NBA money line — no draw possible). Three-way: 3 outcomes (soccer regulation result — home/draw/away). De-vigging math differs between the two — three-way requires distributing margin across all 3 prices, often with non-uniform proportions.