Test a system in four steps before you stake real money: write the rules down, back-test them using only information you had at the time, paper trade them live, then run a small-stakes trial. Judge the result on profit on turnover (POT), not strike rate. Expect to need well over 1,000 bets before the numbers mean much.
Write the rules down before you look at results
A system you cannot write down cannot be tested. Spell out the races, odds range, how a runner qualifies, the bet type, the stake and the price you will take. If you decide on the day whether a race “looks right”, you are tipping, not testing. Do not change them after seeing results.
Back-test without hindsight, then paper trade
Use only data that existed before the jump. That means a price you could have taken, such as morning odds, not the starting price you only see afterwards. Include every qualifier, losers too, and check the rules on races you did not build them on. Elite Racing archives its race-morning ratings and bet-lists online so they cannot be altered after the race, which is the standard to copy. Then paper trade: log each selection and price before the race, without betting.
How many bets before a strike rate means anything?
More than a few weekends. The table assumes $4.00 average odds and a 25% strike rate. The strike rate margin falls from 12.0 points at 50 bets to 1.9 at 2,000. The POT margin is far wider, because one $4.00 winner swings the result.
| Number of bets | Strike rate margin | POT margin | Chance a true minus 10% POT system shows a profit |
|---|---|---|---|
| 50 | ±12.0 pts | ±48.0 pts | 32.7% |
| 100 | ±8.5 pts | ±33.9 pts | 23.3% |
| 250 | ±5.4 pts | ±21.5 pts | 17.2% |
| 500 | ±3.8 pts | ±15.2 pts | 8.3% |
| 1,000 | ±2.7 pts | ±10.7 pts | 2.8% |
| 2,000 | ±1.9 pts | ±7.6 pts | 0.4% |
Source: Calculated for flat $1 stakes at a 25% strike rate and $4.00 average odds, using the normal approximation for a proportion (1.96 x standard error), as described by Penn State STAT 500. The last column is an exact binomial calculation at a 22.5% strike rate. Hypothetical figures, not a real system.
On these assumptions, a 10% POT needs about 1,150 bets to sit clear of zero. A system that truly loses 10% shows a profit after 100 bets about 23% of the time, while a true plus 10% system shows one only 67% of the time.
Judge it on POT, then trial it small
POT is profit divided by total staked. Strike rate ignores price: 60% winners at $1.50 return $0.90 per $1 staked, a minus 10% POT, while 20% winners at $6.00 return $1.20, a plus 20% POT. See how professionals calculate ROI. For the trial, use a small level unit (see what percentage of the bank to use). Wagering providers must offer deposit limit tools (Victorian Gambling and Casino Control Commission, updated 12 June 2026), so cap the trial at an amount you accept losing.
Frequently Asked Questions
How many bets do I need before I can trust a system?
There is no fixed number. On the table’s assumptions, a 10% POT needs about 1,150 bets and a 5% POT about 4,600.
Does a high strike rate mean the system is good?
Not on its own. A high strike rate at short prices can still lose money, so check POT over a large sample.
What if the back-test looks great but the live trial does not?
Trust the live trial. Common causes are hindsight prices, filters tuned to past results and too few bets.
Horse racing betting involves financial risk. Bet only what you can afford to lose. For support, visit Gambling Help Online or call 1800 858 858.