One of yours...
(i) Assess chances in each race you wish to bet on.
(ii) Give runners a % chance of winning
(iii) Correlate that % chance to a price
(iv) Compare your prices with the prices available for the runner
(v) If the price is 10%+ higher than the price you have calculated, write the runner down, with your price and the price available or use a computer.
(vi)Do this for 1 month.
If after 1 month, you could not back each runner to either win a certain amount or on level stakes and show a profit, then
(a) Refine how you assess chance because you have it wrong
(b) Repeat points (i) to (vi) until your recorded results result in a profit. I would suggest of at least 5% ROI over the month
After you have reached point (b) successfully the first time, repeat the process for another 5 months. As soon as you reach any month that your records do not show a profit. Go back to (a) because you still have the assessment process wrong.
Once you have an assessment approach that has won for 6 straight months, you could trial it with real money. A conservative approach would be my suggestion. If after a long period of success, potentially increase stake (within reasonable limits with consideration to the total won to date).