In this exclusive feature, P.P.M's computer expert Neale Yardley provides a fascinating insight into what might happen if  you make a million bets. Would you survive? Would progressions improve your chances, and could you survive the runs of outs?

A perennial talking point among punters is whether it's better to bet level stakes, or on a progression basis. Like others, I could talk about the subject for hours, too!

Readers may recall that last year I had a three-part series in P.P.M. about staking plans. My conclusions were twofold. On the one hand, I suggested that progressive staking plans could improve your chances of ending the day in front, but that they would not actually increase your percentage profit on turnover. What they actually did was increase your chances of profit through increased outlays.

What I decided to do was to use my computer to make what might be called the Ultimate Test-a sequence of one million bets. I wanted to see what would happen if a given sequence of one million bets was approached with a progression staking plan as opposed to level stakes. I also wanted to see how long the longest run of 'cuts' would be, and how large the bets might get if a progression was used without a limit on bet sizes.

Now, the method many pro punters use is to back horses paying over the odds. For example, backing horses they regard as 'true' 3/1 chances at 4/1. Doing this, and achieving the right percentage of winners, will result in a profit in the long run. 'True' 3/1 chances win one in every four races but pay back enough for you to break even if you encounter only one winner in every five bets (achieving 4/1 on the 3/1 true assessment).

The first key to success is actually correctly determining the true odds for each runner. You can use a class/weight ratings method, like the one described by Don Scott in his book, Winning More. An easier way is to use your 'gut feeling', or intuition. Quite often, you can look over a horse's recent form and be convinced it's a virtual certainty and, therefore, excellent value at say 7/2.

On the other hand, you may be adamant that a favourite at 2/1 is a definite risk and not worth its price. Even some professional price assessors, after doing their sums, make intuitive judgements like lengthening a price from 5/2 to 3/1, simply because they think, for one reason or another, that it should be longer.

Let's look at this approach another way: If you only back favourites that you genuinely think are over the odds and only when you can secure 4/1 about them, then there's a good chance you may end up backing 3/1 chances at 4/1. (Remember that the average price of all favourites is about 2/1). Since these horses you will back should win one race in every four races, yet pay back five units for every win, you can expect a return of five units for every four invested, or 125 for every 100, a profit of 25 per cent.

But what about those staking plans? Will one be more suited to this type of betting situation than another? I put my computer to work.

Firstly, I simulated 1000 bets, using the theory of backing 3/1 chances at 4/1. I did this by writing a computer program that repeated itself 1000 times in such a way that a winner was found, on average, 25 times in every 100 bets.

I ran my program 10 times and got returns of between 1070 and 1385, spread around that theoretically expected return of 1250, using level stakes. Okay, now I tested it with progression staking. I used a rather aggressive progression plan of doubling bets until a winner was struck. Bets revert to one unit only after the winner.

I ran the modified program through 100 bets. The first time, the level stake return was only 95 units. Using the progression plan with these same bets the outlay was 8,390,014 for a return of 20,974,765. So, using the progression turned a five per cent loss into an astronomical profit (250 per cent on turnover, and a total of 12,584,751 units profit).

Problem! Look at the outlay. Obviously, a long run of cuts occurred (20, in fact) and bets had to be doubled to an unrealistic level. A run of 20 losing bets would mean you would have been investing $2,097,152 on the 21st bet!

On the next run, the program turned 100 units into 140 at level stakes. With the same sequence, however, the progression turned 2,161 units into 5,155, a profit of 139 per cent, compared to the 40 per cent profit at level stakes.

At this stage, I had to conclude that profits could definitely be improved by using progressions. The only concern was that a punter had to be prepared to keep increasing bets until a winner came along-no matter how long you had to wait. It seems to me that provided the level stakes situation is going to be profitable, there is a good chance the progression would NOT lead to excessively large bets.

If, however, you were facing a loss situation with level stakes, then you would find that the progression plan would lead to excessively large bets of a size that any normal punter would be unable to sustain.

In real-life, your bet size is going to be determined both by your available bank and by the maximum bet a bookie will take at a fixed price, or that the TAB will sustain before dividends become affected to a marked degree.

Because of these aspects, we do need to limit bet sizes with progression methods, even though this may mean you revert at some stage to a single unit bet without having found a winner.

Let's go on further: Most progression plans require you to switch to a one unit bet size and start the progression again after, say, 10 bets. The theory is that you normally should get at least one winner every 10 bets and so never get to the end of the progression without a winner.

So would progressions still improve profitability if bet sizes were limited? I reprogrammed my computer to again simulate backing 3/1 chances at odds of 4/1, using the doubling-up progression, but this time with a limit of six bets before starting the progression again.

This meant actually betting 1, 2, 4, 8, 16, 32-and never betting more than 32 units. I ran this program through 100 bets a few times with the following results:

On the first occasion, 100 units returned 130 units at level stakes (30 per cent profit) but the same sequence showed a 5 per cent loss on the restricted progression plan (571 units outlay returning 540). On the next try, 100 units returned 125 at level stakes (25 per cent profit) and the same bets turned 559 units into 665 with the progression (19 per cent profit).

Finally, 100 returned 105 at level stakes (5 per cent profit) while 690 units returned 830 with the progression (20 per cent profit). As you can see, these results are all very different.

This is when I decided to resort to the ultimate test of ONE MILLION BETS, a figure far in excess, I hope, of what most people will achieve in a lifetime's betting! Surely, I thought, with such a huge test we would start to see some uniform trend in the results?

Well, at level stakes, the 1 million bets returned 1,244,535 (24.5 per cent profit). The restricted progression staking turned an outlay of 6,346,424 into 7,884,655 (24.2 per cent profit). In the end, then, both forms of staking achieved almost identical profits-namely around 24 per cent.

I believe the conclusion to be drawn using this progression plan, anyway-is that progression plans do not, in the long run, improve percentage profit on turnover.' But, of course, the absolute profit was far greater using progressions because you were outlaying more.

So, we have to ask, is there any benefit to be gained from using progression staking? My thinking is that the only scope for making profits from progressions without enduring massive bet sizes, then, is to try to avoid long runs of cuts, so that even if you do have a bet size limit, you never have to use it.

The other trick is to use only mild progression plans in which bet sizes are increased only by the bare minimum required to cover losses.

To help in this regard, the Table accompanying this article details a number of useful progression plans for variously priced selections. All these progressions use only mild bet increases, just enough to ensure getting all your money back when a winner comes along. If you want to build a profit factor into them, I suggest you use the 2/1 progression to back horses at 3/1, the 3/1 progression to back horses at 4/1 and so on.

For backing 611 chances

For backing 511 chances

For backing 411 chances

For backing 311 chances

For backing 211 chances

By Neale Yardley