Arbitrage and +EV Betting for Beginners: A 20-Hour Study Plan

The quick answer

Twenty hours is enough to learn the math and the rules behind arbitrage and +EV betting, if you take them in order and use paper money.

Time: 20 hours across seven stages, roughly two to four hours each.
Order: shared foundations first (odds, vig, settlement), then arbitrage, then +EV.
Money: none until you've logged 30 paper arbs and 100 paper +EV decisions.
Output: an odds and stake calculator, a settlement comparison sheet, two journals and a written budget.

Arbitrage covers every outcome across venues for a locked return, and it fails on execution, fees and settlement mismatches. +EV betting takes a price that pays more than your probability estimate says it should; what sinks it is a bad estimate or a long losing run.

The plan at a glance

Stage	Topic	Hours	Exercise
1	Odds and probability	2	Convert four prices by hand
2	Vig and fair prices	2	De-vig five two-way and five three-way markets
3	Settlement rules and order mechanics	3	Build a settlement sheet for your markets
4	Arbitrage math and execution	4	Validate 20 alerts, simulate five failures
5	Estimating probability for +EV	3	Price one narrow market two ways
6	Variance and staking	3	Simulate a season at one edge, then a smaller one
7	Budget and paper practice	3	Write an operating policy, start both journals

Stage 1: odds and probability (2 hours)

Every later calculation runs on decimal odds, so convert everything to decimal first. A positive American price becomes 1 + odds / 100, a negative one becomes 1 + 100 / |odds|, and implied probability is 1 / decimal. Don't mix up profit and payout: payout includes the returned stake.

Exercise: convert -125 and +140. You should get 1.80 (55.56%) and 2.40 (41.67%), and a $50 stake at +140 should show $70 profit and a $120 payout. Then build a small spreadsheet that does it for you and check it against your hand math. How to convert betting odds covers fractional prices too.

Stage 2: vig and fair prices (2 hours)

Add the implied probabilities on both sides of a market and the total runs over 100%. The excess is the margin. Proportional de-vigging divides each side by that total to get an estimate of the fair price. It's an estimate, and other methods give slightly different answers.

Exercise: take -130/+110. Implied probabilities are 56.52% and 47.62%, summing to 104.14%, so the fair split is about 54.27% and 45.73%. Keep those numbers; you'll need them in Stage 5. The walkthrough is in how to de-vig betting odds, and what each betting app costs shows the margin by venue.

Margin is a cost you pay on every bet, which is why the venue matters to both strategies. On Novig, prices are set between players and pre-game straight trades carry no fee for either side; live straight trades charge the taker 0.03 x price x (1 - price) per contract, or 0.06 on live NCAAF, NHL and MLB.

Stage 3: settlement rules and order mechanics (3 hours)

Two bets only offset if they settle on the same event, line, period and conditions. A regulation-time hockey line isn't the same market as a moneyline including overtime, and a three-way soccer line isn't the same as a two-way one. Read how each venue handles pushes, voids, postponements, a player who doesn't play and obvious-error prices.

On a prediction market you also place orders, and an order can sit unmatched or fill partway. Learn what happens to a resting order at game start and what you pay as a maker versus a taker.

Exercise: for the exact markets you'd use, write down what happens if the game is suspended or a player withdraws. Reading list: pushes, voids and postponements, two-way vs three-way moneylines, obvious error rules, back vs lay bets, matched and partially filled orders and limit orders vs taking the price.

Stage 4: arbitrage math and execution (4 hours)

Sum 1 / d across every outcome to get S. If it's below 1, you can cover every outcome at a profit before fees. Stake on each outcome is B / (d x S), and the return on total stake is 1 / S - 1, which runs slightly above 1 - S.

Worked example: prices of 2.15 and 2.00 on the two sides give S = 0.9651. On $300, the stakes are $144.58 and $155.42, and the profit is $10.85 or $10.84 depending on the winner, a 3.61% return against a 1 - S of 3.49%. The formula is unpacked in how to calculate arbitrage stakes.

Then spend most of the four hours on what breaks an arb. Pair a 2.15 regulation-time hockey price with a 2.00 moneyline that includes overtime and a game tied after 60 minutes can lose both legs, and that's a false arb. A second leg can also be rejected or cut to a smaller stake after the first one is already placed.

Exercise: review 20 scanner alerts and record why you'd accept or reject each, then simulate five failures, including a rejected second leg and a partial fill, and work out the loss before choosing a response. Read false arbs, arbitrage vs middling, seven ways an arb loses money, what arbitrage costs and live betting latency.

Stage 5: estimating probability for +EV (3 hours)

EV per dollar is p x d - 1, and the bet is +EV when p beats 1 / d. You'll get p from a sharp reference market with the margin removed or from your own model, and several books moving on the same news don't count as separate evidence.

Worked example: Stage 2's fair price put the underdog at 45.73%. If another book offers +125 (2.25, break-even 44.44%), EV is 0.4573 x 2.25 - 1 = +2.88%. If your estimate is more than 1.29 points too high, there's no edge.

Exercise: pick one narrow market, price it from a reference line and from a simple model of your own, and explain why the two differ. Start with what positive EV betting is.

Stage 6: variance and staking (3 hours)

A winning edge still loses for long stretches, and a lucky run can look like skill, which is the subject of variance and sample size. Learn calibration (your 60% calls should land about 60% of the time over a large sample) and use closing line value as one diagnostic among several.

For sizing, compare flat stakes, a fixed percentage and the Kelly criterion, f* = (b x p - (1 - p)) / b with b = d - 1. The +125 bet above gives f* = 2.31% of bankroll, and a quarter of that is 0.58%. Kelly takes p as known, which it isn't, so treat it as a ceiling and cap total exposure when several bets ride on one game (correlated bets covers that case).

Exercise: simulate a few hundred bets at a 3% edge, record the worst drawdown, then repeat at 1%.

Stage 7: budget and paper practice (3 hours)

Write a one-page operating policy before any account is funded: a separate amount you can afford to lose, a weekly time limit, the records you'll keep and the conditions that make you stop. Leave borrowed money and rent money out of it. Learn each venue's age and location rules, ID checks and withdrawal terms, and set deposit limits where they're offered. Winnings may be taxable; check with a tax professional.

Exercise: start two journals. The arbitrage one logs market identity, timestamps, displayed and accepted prices, fees, exposure under each outcome and final settlement. The +EV one logs the price, reference market, your estimate and method, EV, stake and closing price. How much money you need to start and withdrawal times compared cover the cash side.

After the 20 hours

Paper practice doesn't stop when the study plan does. Aim for at least 30 fully documented arbs, including one valid, one false and one failed execution you can explain, and at least 100 timestamped +EV decisions before reviewing results. Those counts build habits and say nothing yet about whether an edge will last. Back-to-lay arbitrage, player-prop models and automation come later; back-to-lay arbitrage explained is the next step on the arb side.

Study plan FAQ

Should I learn arbitrage or +EV betting first?

Arbitrage. Its math is fixed once the prices are known, and the main skill is checking settlement and execution. +EV adds a probability estimate that can be wrong, which is harder to test.

Can I learn arbitrage betting in a weekend?

The formulas, yes. Validating alerts and handling failed legs takes the practice sessions, which is why the plan ends with 30 paper arbs.

Where should a beginner practice the cost side?

Wherever the cost on the bet type you'd place is lowest, because every cost comes off a thin edge. On Novig, a straight trade placed before the game has no fee, and a resting order that someone fills costs nothing at any time, so the quoted price is the full cost on those bets. It's open to users 21 and older nationwide except Nevada, Arizona and Michigan.

Do I need software to start?

No. A spreadsheet that converts odds, splits stakes and computes EV covers everything in this plan. Scanners are useful later, as long as you're still checking each lead by hand.