A bet is not good because it won. It is good when the price you took was bigger than the fair one. That is what expected value measures.
The formula
EV per 1 unit = probability × odds − 1. If you think an event has a 50% chance and the odds are 2.10, EV is 0.50 × 2.10 − 1 = +0.05. For every 100 units staked on the same bet you expect, on average, 5 units of profit.
All the weight sits on "on average". A single bet either wins or loses. EV only shows up over many.
Value bets
A value beti is a bet with positive EV based on a good probability estimate. The estimate is the hard part. If your probability is worse than the market's, the "value" you see is just your own mistakes.
That is why the market is a useful check. When your model disagrees strongly with a sharpi market, the usual explanation is that you missed information (an injury, a manager change). Our policy is not to treat a model-only gap as value.
Hit rate is not enough
A 70% hit ratei sounds impressive. At odds of 1.40 the break-even rate is 1 / 1.40 = 71.4%, so 70% loses money. At odds of 1.60 the break-even is 62.5% and the same 70% is profitable.
Anyone who shows you a hit rate without the odds shows you half the picture. See also how to spot a reliable tipster.