A betting strategy that holds up takes prices above the fair price for each result, stakes small enough to survive losing runs, and tests itself against the closing price before trusting its profit. Of the three, price decides whether a method can make money at all. With illustrative numbers, a team with a 62.5% chance is worth +$3.75 per $100 staked on average at $1.66 and -$5.00 at $1.52: the same team in the same game.
The betting strategies that work are ways of finding, sizing and checking those prices. Staking systems and chasing losses leave every price where it was, so neither can beat a bookmaker's margin, and a tip sold on hype comes with no evidence that its prices beat it. Even a sound method loses often, bookmakers can restrict accounts that keep winning, and there is no guarantee of profit.
Price is the only edge that lasts
Convert a bookmaker's prices into chances and they total more than 100%, and that surplus is its margin. Bet at the market's own prices and, if the margin is spread evenly, you lose 1 - 1 / that total per dollar on average, about 2.4 cents at 102.5%. The way past it is a price above the fair price, where fair price = 1 / your best estimate of the chance.
Example: Team A plays Team B in a market with no draw. A margin-free market, or your own model, gives Team A a 62.5% chance, so its fair price is 1 / 0.625 = $1.60. Illustrative prices, not a real game.
| Bookmaker price | Edge: price / $1.60 - 1 | Expected value per $100: 0.625 x price x $100 - $100 | Break-even strike rate: 1 / price |
|---|---|---|---|
| $1.52 | -5.0% | -$5.00 | 65.8% |
| $1.60 | 0.0% | $0.00 | 62.5% |
| $1.66 | +3.75% | +$3.75 | 60.2% |
Team A wins 62.5% of the time whichever price you take: short of the 65.8% that $1.52 needs, clear of the 60.2% that $1.66 needs. Picking Team A takes the same judgement in all three rows: only the price decides whether the bet makes or loses money on average.
That gap is what value betting is about, and the expected value calculator runs the sum for any price and chance. Your betting edge is an average over many bets, not a forecast: on these numbers the $1.66 bet still loses three times in eight.
A price ends up above fair in three ways:
- One bookmaker is out of line with the market, which comparing prices across bookmakers finds.
- Your fair price beats the market's own, which takes the models and ratings covered below.
- A promotion already on your account adds value, which valuing a betting promotion measures. A bonus bet that loses pays nothing.
Arbitrage betting backs every outcome when the best prices' implied chances total under 100%; the arbitrage calculator splits the stakes. Matched betting backs a selection with a bonus bet already on your account and lays it at an exchange. Bookmakers can limit accounts that use either.
Risk: Betting involves risk. A +EV bet can still lose, and often does; every fair price is an estimate that can be wrong; and arbitrage and matched betting only work if every bet is accepted or matched at the price shown. There is no guarantee of profit; see responsible gambling for limits and support.
Testing a method against the closing price
Closing line value (CLV) compares the price you took with the last price before the start: CLV = your price / closing price - 1. By the jump or the start of a game, the market usually holds the most money and news it will ever have, so the close is its best estimate of the chance. Beating it again and again is the earliest good evidence that your prices sit above fair.
For sport, take the margin out of the close first. Say the market closes at $1.60 for Team A and $2.50 for Team B (illustrative):
- closing book = 1 / 1.60 + 1 / 2.50 = 62.5% + 40.0% = 102.5%
- margin-free close for Team A = 1.60 x 1.025 = $1.64
- CLV = 1.66 / 1.64 - 1 = +1.2%
Measured against the raw $1.60, the same bet shows 1.66 / 1.60 - 1 = +3.75%, which overstates it because the raw price still carries the margin.
The close also corrects your fair price. It rates Team A at 1 / 1.64 = 61.0%, not your 62.5%, so your real edge on this bet was probably nearer 1.2% than 3.75%. Closing line value covers the method in full.
For racing, the usual close is the Betfair Starting Price, and closing line value in horse racing shows how to measure against it. Two limits apply: a close in a thin market, with little money behind it, is a weak benchmark, and CLV is evidence about your prices, not money in your account.
Staking for the losing runs a real edge still has
Your bankroll is money set aside for betting that you can afford to lose, kept apart from everyday money. How much of it rides on each bet decides whether an edge lasts long enough to show, because good bets lose in runs. With illustrative prices, here are 400 bets of $20 ($8,000 staked) at the same 3.75% edge and three different prices:
| Price | Fair price | Win chance | Expected profit | Standard deviation of profit | Chance of being behind after 400 bets |
|---|---|---|---|---|---|
| $1.66 | $1.60 | 62.5% | $300 | $321 | 16.3% |
| $4.15 | $4.00 | 25.0% | $300 | $719 | 34.6% |
| $8.30 | $8.00 | 12.5% | $300 | $1,098 | 41.7% |
The working, on the $1.66 row:
- expected profit = bets x stake x edge = 400 x $20 x 0.0375 = $300
- standard deviation (the typical swing in profit) = stake x price x square root of (chance x (1 - chance)) x square root of bets = 20 x 1.66 x 0.4841 x 20 = $321
- finishing ahead takes 241 winners (241 x $33.20 = $8,001.20 back on $8,000), and at a 62.5% chance per bet, 240 winners or fewer turn up 16.3% of the time (binomial distribution)
Variance in betting and losing streaks in betting show how deep and how long those runs get, and risk of ruin whether a bankroll outlasts them.
A common way to set stakes is as a share of the bankroll, often counted in betting units and compared with other staking plans. The Kelly criterion sizes that share from the edge: stake share = edge / (decimal odds - 1). At the same 3.75% edge it gives 0.0375 / 0.66 = 5.7% of the bankroll at $1.66 but 0.0375 / 7.30 = 0.5% at $8.30, so it stakes less where the swings are larger.
Kelly assumes your edge estimate is exact, and it never is, so many people stake a half or a quarter of it. Fractional Kelly and bankroll management take both ideas further, and the Kelly criterion calculator runs the sum.
Australian online bookmakers must offer deposit limits, and under the National Consumer Protection Framework (measure 6) a decrease applies at once while an increase waits 7 days. That delay makes a deposit limit a cap that a bad afternoon cannot raise.
Where a fair price comes from: models, ratings and framing
There are three ways to get a fair price:
| Route | What it needs | The main trap |
|---|---|---|
| Take the margin out of a strong market | Prices from the exchange or many bookmakers, and a method for removing the margin | It can only match the market, so the edge comes from bookmakers slow to follow it |
| Build a model | Clean historical data, a simple first model and an honest test | Overfitting: a model that explains the past and prices the future badly |
| Frame a market from your own ratings | A rating for every runner, turned into chances that add up to 100% | Ratings that quietly agree with the bets you already wanted |
The first route is the cheapest, and hard to improve on where plenty of money trades; fair odds compares the references. A model can be a spreadsheet of team ratings, a Poisson model for goals or machine learning, and how to build a betting model walks through the steps. Framing is the racing version.
Example: a five-runner race. Your ratings turn into raw chances that add up to 110%, because each runner was judged on its own. Divide each by 1.10 so the frame totals 100%, then compare. Illustrative numbers, not a real race.
| Runner | Your raw chance | Framed chance: raw / 1.10 | Fair price: 1 / framed chance | Bookmaker price | Edge: price / fair - 1 |
|---|---|---|---|---|---|
| Runner 1 | 44% | 40% | $2.50 | $2.20 | -12.0% |
| Runner 2 | 27.5% | 25% | $4.00 | $3.80 | -5.0% |
| Runner 3 | 22% | 20% | $5.00 | $5.90 | +18.0% |
| Runner 4 | 11% | 10% | $10.00 | $7.50 | -25.0% |
| Runner 5 | 5.5% | 5% | $20.00 | $13.00 | -35.0% |
The bookmaker's prices add up to 109.7%, so its market as a whole is not generous, and the +18% gap on Runner 3 is the part to distrust. Suppose the exchange has Runner 3 at about $5.80 (illustrative), a 17.2% chance (1 / 5.80). The bookmaker's $5.90 then beats the exchange by only 5.90 / 5.80 - 1 = 1.7%, and your 20% is the outlier.
When your price disagrees with every bookmaker and the exchange by that much, look for what you missed, and pass if you cannot find it. How to frame a betting market works through a full field.
Test any route the same way: write the price down before the market can sway it, then check it against the close and score it with a Brier score. Backtest a strategy only on prices that were available when you would have bet.
Risk: Betting involves risk. A fair price from a model, ratings or a frame is an estimate, and a bet above it can still lose. See responsible gambling for limits and support.
Records, and how many bets before a result means anything
Record every bet: event, market, selection, bookmaker, price taken, stake, cash or bonus bet, closing price and result. How to track betting results sets out the columns. Here is a hypothetical record of 250 bets of $20, all at $2.40 for simplicity:
| Measure | Working | Result |
|---|---|---|
| Turnover | 250 x $20 | $5,000 |
| Return | 109 winners x $20 x 2.40 | $5,232 |
| Profit | $5,232 - $5,000 | $232 |
| Yield | $232 / $5,000 | 4.6% |
| Strike rate | 109 / 250 | 43.6% |
| Break-even strike rate | 1 / 2.40 | 41.7% |
| Standard error of the yield (luck's typical swing), with no edge | 2.40 x square root of (0.4167 x 0.5833) / square root of 250 | about 7.5 percentage points |
A 4.6% yield looks good, but it sits less than one standard error above zero. A punter with no edge at all lands anywhere from about -15% to +15% on this many bets (two standard errors either side of zero), about 19 times in 20.
On profit alone, a 3.75% edge at $1.66 needs about 4 x (1.66 - 1) / (0.0375 x 0.0375) = 1,877 bets, roughly 1,900, before its expected profit stands about two standard errors clear of luck. That is the average case: at that count a real edge this size clears that bar only about half the time. At double the bets its expected profit sits 2 x square root of 2 = about 2.8 standard errors clear, and it clears the bar about four times in five. Sample size in betting explains that rule, ROI vs yield the measures, and strike rate betting why a high strike rate can still lose money.
CLV gets there sooner. Suppose the same 250 bets averaged +1.8% CLV, and each bet's CLV typically varied by about 5 percentage points (both illustrative). The standard error of that average is 5 / square root of 250 = 0.32 points, so +1.8% sits more than five standard errors above zero while the yield sits less than one.
Your bookmakers keep a second record. Under the national framework (measure 7), each must send active customers a monthly activity statement showing, among other things, the number of bets, the wins and losses and the net result. Betting activity statements shows how to check your figures against them.
Risk: Betting involves risk. Past results are no guarantee of future results: a record that is ahead can still be luck. See responsible gambling for limits and support.
Habits that cost money: chasing, systems and tipster hype
The costs below use the -5% bet at $1.52 from the first example.
| What people try | Why it fails | What it costs |
|---|---|---|
| Chasing losses with bigger stakes | More money goes through the same prices | Each extra $1,000 staked at -5% costs about $50 on average |
| Staking systems such as martingale, Fibonacci or Labouchere | They reorder stakes and leave every price alone | $10,000 staked at -5% loses about $500 on average, in any order |
| Backing favourites because they win more often | Strike rate is not profit | At $1.52 you win five bets in eight and still lose $5 per $100 on average |
| Believing a winner is due | Each result is independent of the last | After three losses, the next 62.5% bet still wins 62.5% of the time |
| Following tips or systems sold on their winners | A record shown without every bet, or without prices, proves nothing | The subscription, plus bets with no measured edge |
Chasing losses, the martingale system, betting systems that don't work and the gambler's fallacy take each habit apart. Judge a tipster by checking a tipster's record: every bet, the price taken and the closing price. Scamwatch, run by the ACCC, calls sports "investment" schemes gambling rather than investing, and often scams, whether sold as prediction software, a syndicate or tips (checked October 2026). Sports investment scheme scams lists the warning signs.
If losing has you raising stakes or betting for longer than you planned, treat that as the signal to stop, not to try a new system. Signs of a gambling problem lists what to watch for, and Gambling Help Online offers free, confidential counselling on 1800 858 858 or online, 24 hours a day, 7 days a week.
Account restrictions: the limit a winning method meets
Expected profit = edge x turnover, so the stake a bookmaker will accept matters as much as the edge. At $20 a bet, the 400 bets at $1.66 were worth $300 on average. Cut to $5 a bet, the same bets are worth 400 x $5 x 0.0375 = $75: a quarter as much, from an unchanged edge. Can you make money betting weighs this limit with the margin and variance.
Bookmaker terms generally let a bookmaker limit stakes, withdraw promotions, void bets in the cases its terms set out, or close an account. A customer who keeps beating the closing price is costing it money, which gives it a reason to act. Many terms also restrict automated betting, third-party access and multiple accounts, and that risk is yours; why bookmakers restrict accounts explains how restrictions work.
A restriction is a limit to accept, not to dodge with a second account or with someone else's account or identity. Bookmaker terms generally require accounts in the holder's own name, and every Australian online account must have its holder's identity verified before the first bet.
Racing has one partial exception. Racing Victoria and Racing Queensland are among the racing bodies that make off-course bookmakers accept fixed-odds bets on eligible races until they stand to lose a set amount, with exclusions (checked October 2026). Minimum bet limits explains where they apply. How to beat the bookies maps the approaches that can clear the margin and the limit each one meets.
Risk: Betting involves risk. An expected profit is an average, not a forecast. See responsible gambling for limits and support.
Sports betting strategy in Australia, and how racing differs
The method is the same for both; only its inputs and rules change:
| Question | Racing | Sport |
|---|---|---|
| Usual fair-price reference | The Betfair exchange near the jump, or your own frame | A margin-free price from many bookmakers and, where it trades, the exchange, or your own model |
| Closing price for CLV | The Betfair Starting Price | The last margin-free price before the game starts |
| Settlement traps | Stewards set deductions on fixed-odds bets taken before a late scratching; dead heats and place terms follow the rules that apply or your bookmaker's terms | Extra time, tennis retirements, abandoned games and players who do not take the field are settled under each bookmaker's own rules, which differ |
| Australian rules that shape it | Tote dividends are declared after the result, so they cannot be compared with a fair price in advance; some races carry minimum bet limits | Under the Interactive Gambling Act, an online sports bet is lawful only through a provider with a licence from an Australian state or territory (s 15AA), and only if it is placed and accepted before the event begins (s 10B and s 8A(3)) |
For sport, that last row makes an Australian strategy a pre-game one: price, stake and closing line are all fixed before the start. AFL betting strategy, NRL betting strategy and cricket betting strategy apply the method code by code.
For racing, the exchange gives a reference with no bookmaker margin in it, since a betting exchange takes its cut as commission on net winnings rather than in its prices. Horse racing betting strategy and greyhound betting strategy apply the method race by race.
How to bet smarter: a checklist before each bet
A smart betting strategy is mostly a set of habits that protect the price you take and the bankroll behind it:
- Write your fair price, or note the margin-free market price, before you look at what bookmakers offer.
- Compare prices across bookmakers and take only a price above fair, by a margin you set in advance.
- Treat a lone price well above every other bookmaker as a likely mistake or old quote, and check it before you bet.
- Size the stake from your bankroll, as a fixed share or a fraction of Kelly, and never above the cap you set.
- Record the bet with the price you took, then add the closing price once the event starts.
- Review yield and average CLV each month, with the number of bets beside them.
- Keep a deposit limit with each bookmaker, never raise stakes to recover a loss, and stop when betting stops feeling like a choice.
Risk: Betting involves risk. A checklist improves the prices you take, not the result of any one bet, and a sound process can still lose money over hundreds of bets. See responsible gambling for limits and support.
Where B337's screeners fit this method
B337's +EV screener searches sports markets for bookmaker prices above a fair price and lists them ranked by edge.
It builds each fair price by removing the margin from prices across the market, exchanges and global market references among them, and combining the results. The arbitrage and middles screeners look for two other kinds of gap, and the positive EV betting page explains the +EV screener's method.
Screener rows come with Terminal Pro and Full Automation, or as a Screeners plan set up with the team. Terminal View does not include them, and a free account sees the live counts with the rows locked. You decide and place every bet in your own bookmaker accounts, on Terminal Pro by clicking a price on the board, and bets placed through B337 use credits.
The Terminal collects prices on a repeating cycle, so a price on screen can lag the bookmaker's own; check it in your bookmaker account before you bet.
B337 is software, not a bookmaker, and it does not give personal financial or betting advice. Bookmaker names are trade marks of their owners. B337 is not affiliated with them.
For free and confidential support call 1800 858 858 or visit gamblinghelponline.org.au.
Risk: Betting involves risk. The fair price behind every screener row is an estimate, and an arbitrage or a middle only works if every bet is accepted at the price shown. Many bookmakers restrict or prohibit automated betting and third-party access in their terms, and using B337 may breach them; a bookmaker can limit stakes, void bets or close an account, and that risk is yours. See responsible gambling for limits and support.