EdgeStat Academy

0. Why Most Bettors Lose

Sports betting is a market. There are two prices on every bet: yours and the book's. Both have edges baked in. The book builds in a 4–5% vig (sometimes 10%+ on props). You're trying to find spots where your true probability of winning is high enough to clear that vig and still profit.

The average bettor loses because they bet for entertainment, react to recency bias, chase losses, and stake non-systematically. The sharp bettor wins for the opposite reasons. This page teaches you to think like the second group.

1. Reading Odds

Three formats:

• American (US): –150 means risk $150 to win $100. +150 means risk $100 to win $150.
• Decimal (rest of world): 2.50 means a $1 bet pays $2.50 total (profit + stake).
• Fractional (UK): 3/2 means risk $2 to win $3.

Conversion:

American → Decimal:
   if odds > 0:   decimal = 1 + odds/100
   if odds < 0:   decimal = 1 + 100/abs(odds)

Decimal → Implied Probability:
   prob = 1 / decimal

2. Implied Probability & Vig

The "implied probability" of an odds line is what the price says about the win likelihood. –150 implies 60%. +150 implies 40%. Add both sides of a moneyline and you'll get a number greater than 100% — the overage is the book's vig (a.k.a. juice).

To get the vig-free probability, divide each side by the sum:

p_home_raw = 60.0%
p_away_raw = 45.5%
sum         = 105.5%
p_home_fair = 60.0 / 105.5 = 56.9%
p_away_fair = 45.5 / 105.5 = 43.1%

The fair price is the market's true probability estimate. If your model says the home team is 60%+, you have an edge worth measuring.

3. Expected Value (EV)

EV is the average profit (or loss) per dollar bet, over the long run, given your win probability and the price.

EV per $1 = p_win × (decimal - 1) - (1 - p_win)
          = 0.64 × (1.676 - 1) - 0.36
          = +0.072

That's +7.2 cents of expected value for every dollar bet. Over 1,000 plays at $100 each, expectation is +$7,200 — assuming your model probability is correct. If it isn't, EV is negative and you bleed money no matter how lucky you get short-term.

4. The Kelly Criterion

Once you know you have an edge, the next question is: how much should I bet? The math-optimal answer is the Kelly fraction, which maximizes the geometric growth rate of your bankroll:

f* = (b × p - q) / b

where:
  b = decimal odds - 1
  p = your win probability
  q = 1 - p

Example: $5,000 bankroll, you estimate 60% win prob at +120 (decimal 2.20).

b = 1.20
p = 0.60
q = 0.40
f* = (1.20 × 0.60 - 0.40) / 1.20
   = 0.267    → 26.7% of bankroll

That's $1,335 on one bet — way too aggressive for most people. Sharps use fractional Kelly, typically ¼-Kelly, which cuts the recommended stake to ~6.7% of bankroll ($335) and slashes variance roughly 75% while preserving most of the long-run growth.

5. Variance and Sample Size

A 55% sports bettor will go on multi-week losing streaks. A 60% bettor will too. The market is noisy, and the difference between "good" and "lucky" only shows up after 500+ plays.

Quick rule of thumb: at 55% hit rate, the standard error on your hit-rate estimate after n bets is roughly √(0.55 × 0.45 / n). After 100 plays that's ±5%. After 1,000 plays, ±1.6%. Don't make decisions on small samples.

6. Closing Line Value (CLV)

The most important leading indicator of long-term profit. CLV measures the gap between the price you got and the closing price (the last price before the game starts).

If you bet Dodgers ML at –148 and the line closes –165, you beat the close by ~7% in implied probability. That's positive CLV. Bettors who consistently beat the close are profitable bettors — even in months where they get unlucky on results.

EdgeStat's published target: +2.5% average CLV. We log every play and publish the CLV on each.

7. Market Efficiency

Not all sports markets are equally sharp. Loosely ranked from sharpest to softest:

1. NFL sides, NBA sides: brutally sharp, billions of dollars settling them, very hard to beat
2. NFL/NBA totals: still sharp but slightly softer
3. MLB ML / RL: sharp on big games, softer on day games and small-market teams
4. NHL: moderately sharp
5. MLB / NBA player props: very soft. Books can't manually sharpen all 1,000+ daily props.
6. Same-game parlays: trap. Heavy juice. Avoid.
7. Niche sports: WNBA, NCAA-T, MLS — softer but lower limits.

8. How Sharps Actually Think

They don't think "I feel like the Dodgers will win." They think:

• "My model says 64.0%, market says 59.7%. Edge = +4.3%."
• "At –148, EV is +5.4 cents per dollar. Kelly stake at ¼ is 1.7 units."
• "I beat the close last time on this team — model is calibrated."
• "If I lose three in a row, my bankroll drawdown is –4.8u, well within tolerance."

It's emotionless and probabilistic. The fun part is being right on average, not on any individual play.

9. Top 10 Mistakes to Stop Making

1. Parlays. Multi-leg parlays compound the vig. A 4-leg parlay at –110 each leg has ~25% built-in juice. Bet the legs separately.
2. Same-game parlays. Even worse. Books price out the correlations to their benefit.
3. Chasing losses. Bet sizes should be a function of edge, not how much you lost yesterday.
4. Bet without a model. If you can't articulate your win-probability number, you don't actually have a bet — you have a hunch.
5. Single-book betting. Always shop. The price difference between books is often bigger than your edge.
6. Public-side bias. 70% of public tickets on a side ≠ side will win. Often it's the fade.
7. Bet too big. Stake > ¼-Kelly without warrant. Variance kills you before edge can save you.
8. Bet too small. If you actually have +6% edge, betting 0.25u is leaving money on the table.
9. Ignore CLV. Results are noisy; CLV is your true signal.
10. No record keeping. If you don't log every play with price, stake, and outcome, you're guessing about your own performance.

10. MLB-Specific Edges

MLB is our flagship sport because the inefficiencies are exploitable. Specific edges:

• Day games after night games: bullpens are exhausted, totals over.
• Ace-vs.-back-end-starter mismatches: public lines lag, especially on the spread (-1.5 / +1.5).
• Wind at Wrigley: 10+mph out = OVER tilt; 8+mph in = UNDER tilt. Books only partially adjust.
• Marine layer in San Diego / San Francisco: evening games trend UNDER.
• Coors Field: nearly always OVER. But sharps will fade public-loved overs if wind blows in.
• Umpire selection: wide-zone umps lower totals; tight-zone umps push more walks/runs.
• First-5 totals: remove bullpen variance, get cleaner pitcher-vs-pitcher exposure.
• NRFI: bet against teams with weak top-of-order vs. dominant starters.

11. Bankroll Management

Three rules:

1. One unit = 1% of bankroll, max. If you have $5,000, 1u = $50. Most plays should be 0.5–1.5u.
2. Re-baseline monthly. If your bankroll grows 20%, raise your unit size. If it drops 20%, cut it. This is how Kelly self-regulates.
3. Reserve fund. Keep 6 months of personal expenses outside of your bankroll. You should not need this money for rent.

12. Glossary

ATS: against the spread. Win/loss record on point spreads.
Bankroll: total funds set aside for betting.
CLV: closing line value — gap between your price and the closing price.
Cover: winning a spread bet by the spread amount.
EV: expected value.
Edge: your modeled win probability minus the market's implied probability.
F5: first 5 innings. MLB market that excludes the bullpen.
Handle: total dollars wagered on a market.
Hold: book's profit margin on a market.
Implied probability: what the odds price says about win likelihood (before vig adjustment).
Juice / Vig: the book's built-in fee on a bet.
Kelly Criterion: stake-sizing formula that maximizes long-run bankroll growth.
Line shopping: comparing prices across multiple books and taking the best.
NRFI / YRFI: No Run / Yes Run First Inning.
O/U: over/under (total).
Push: tie — stake returned.
Reverse line movement (RLM): line moves opposite to the public ticket %.
Run Line: MLB spread, fixed at ±1.5.
Sharp / Square: professional bettor / recreational bettor.
Steam: sudden coordinated line movement triggered by sharp money.
Unit (u): standard bet size, typically 1% of bankroll.
wOBA / wRC+: weighted on-base average / weighted runs created plus.
xFIP: pitcher's fielding-independent ERA, normalized for HR/FB ratio.
xwOBA: expected wOBA based on exit velocity and launch angle.