A good strike rate is no use at all if the bets are not made at value odds. That is why I decided to write this short piece.

The scenario I want you to imagine is of three people walking into a room where different bets are on offer.

The first person sees a table where you can bet on the spin of a coin. The odds on offer are 10/11 for a correct choice.

He feels lucky and bets on heads, which comes up. So he has another bet, and this also wins. "I'm feeling lucky", he says, "I'm going to stay at this table all night".

The second person sees a table where you can bet on the correct number thrown with a dice. The odds on offer are 11/2.

He has a bet on '3', which loses, and then on '5', which loses. But he's not deterred; "I'm staying on this table", he says, "see you later".

The third person sees a table where you can bet on predicting a playing card randomly chosen from a pack. The odds on offer are 66/1.

He likes the prospect of a big win, so he has a go. In fact he has 10 goes, and none are correct. Luckily, he has brought plenty of cash with him, so he says "I'm staying here for the night".

The three agree to meet up again in 8 hours time, when they will compare how each has done.

Eight hours passed, and it was time for the three to go home.

The coin man was not happy; "I seemed to be winning regularly", he said, "I didn't have any long losing runs, but I'm out of pocket!"

In fact he'd had 1,000 bets, and 500 of them had been correct. He'd been betting at £20 a time, as he knew he wouldn't have long losing runs. So over the night he'd staked £20,000, and from his 500 winners paying £38.18 a time, he'd received a total of £19,090. He had lost £910 when he seemed to be doing so well. People had been remarking at how many winners he was having, but that didn't matter to him now. He certainly wouldn't play that game again.

The second arrived, with a smile on his face; "I started slowly", he said, "and my strike rate was not good all night - I had many losing runs well into double figures. I didn't feel as if I was winning, but now that I've counted up I realise that I've done well."

In fact, he too had placed 1,000 bets over the night at £20 each. He had picked 166 winners and 834 losers. At his quoted odds of 11/2, each winner had paid him £130 - a total return of £21,580. He had won £1,580 on the night and was happy.

The third returned laughing; "I can't believe it", he said, "I chose those cards terribly - I had massive losing runs which I thought would never end. In fact I had 1,000 bets at £20 each and only won 19 of them - good job my bank was big enough for the evening!" But at 66/1, the 19 winners had returned a total of £25,460, making him £5,460 - enough for the cruise he'd been planning.

Simple enough, and we can see why the coin man was attracted by the prospect of regular winners - it made him feel like he was doing well. The dice man had a strike rate of 16.67%, which was more than enough, given that he was offered the value odds of 11/2 each time. The odds quoted were only half a point too high, but that was enough for him and he took advantage. The card man was fortunate to have a big bank behind him - his losing runs were massive, only 19 winners from 1,000 at a strike rate of 1.9%. But the 66/1 he was given was far too high, and he made a lot of money.

If these people were to return every night for a year, betting on the same games and with the same stakes, the coin man would lose £332,150, the dice man would win £576,700 and the card man would win £1,992,900.

They were all winning as often as they were entitled to do - nobody was luckier or unluckier than the others. It all came down to the value their bets offered.

In that example, the value was easy to see. Mathematically, it could be determined exactly. But how do we know what value is in horse-racing? How do we know we won't end up like the coin man?

The answer is that we need to try to determine the actual odds of our horse winning. We can't do this exactly, of course, but we can get fairly close.

Say for example the race we are studying has 10 runners, and we have decided that 4 of them have no chance whatsoever. We can confidently cross those 4 through, but we must be absolutely sure they can't win. That leaves 6 runners, and we may decide that 3 of those can't be dismissed completely even though we don't think they can win. Maybe they have half a chance. Fine, that's 3 runners with half a chance each, and the other 3 are the serious contenders (of which one is our selection).

Our selection, therefore, has 1 chance in 4.5 (3 serious plus 3 half chances). Our interpretation of its actual odds is 7/2.

If we can get better than 7/2 for our selection we bet. The bigger the price above 7/2, the more we stake. It makes sense - if the dice man went back the following night he would stake much higher at that price. If our horse, however, is quoted at less than 7/2 we cannot bet - not ever.

It may well win, but if we continually take chances like that we will eventually come unstuck. If the odds are 3/1 we have to leave it alone and come back another day.

Again, that's a simple example, and I'm sure most punters think along those lines. To get more of an exact figure we need to break our fractions down further - to percentage chances.

If we say that the total chances of all the runners in a race is 100%, we need to then determine how much of that 100% is made up of our horse's chance. In the above example the 100% was split between 6 runners that had either a chance or half a chance. There were a total of 4.5 chances, each one 22.22% of the overall 100%.

A 22.22% chance is equal to 7/2, so if we are right in our judgement (and we must stick to our judgement, as that is the whole point of betting), anything better than 7/2 offers value.

A friend of mine has the same approach, although he tackles it slightly differently. He allocates 100 points to a race (each point is a 1% chance). From those 100 points, he allocates to each horse a number, which he thinks represents its chance of winning. The important thing here is that all the points must come back to 100 when added at the end.

He may analyse a race like this: His selection 27pts, horse 'B' 20pts, horses 'C', 'D', 'E', and 'F' 12pts each, and horse 'G' 5pts.

He has given his selection a 27% chance, which equals just over 11/4. So if my friend can get better than 11/4 for his bet he will be on. If not, he won't.

To convert the points out of 100 you have awarded to each horse into odds, do the following:

Say you gave 28pts: divide 100 by 28 = 3.57, and then deduct 1 = 2.57.

Those are the odds to one: 2.57/1. Given that 2.5/1 is the same as 5/2, you would need better than 5/2 for a horse to which you had awarded 28pts.

Please make sure you don't totally discount a horse and give it 0pts unless you are sure that it can't win. Make sure your entire points total 100 exactly. You will have to change them many times before they do - and don't bother if the points awarded are strange numbers, like 27 or 41. it doesn't matter as long as you work out the odds correctly afterwards.

You will suddenly find that you are spending less time trying to pick winners, and more time selecting horses that are likely to run to the best of their abilities at prices which are bigger than they really should be.

Value - the single most important thing, no matter what you bet on.