Forecasting the next no

Forecast - based on what? How are the numbers generated? Is it random (like rolling a dice, where the probability of rolling a 6 is ALWAYS 1 in 6, EVEN IF the last 100 throws have all been a 6), or non-random; in which case what is the source of the non-randomness.

If this was do-able, the bookmakers would be out of business and racing enthusiasts would all be millionaires... Based on history, it looks like the probaibility of a win is just a shade over 4:10. However, that's a real oversimplification as it doesn't take into account a host of other things that a simple placing doesn't measure (improvement/decrease in an animal's performance as the season goes on; the length of the race, the going, the quality of the opposition in each individual race, etc).

Sadly, when you add these things on, I reckon you're leaving the world of Excel and maths and entering the realm of magic... . Good luck, but if you do ever make a million, please remember me!!

I realize this is a couple of days old. If the OP is still interested, here is what I see.

There really is not much to base a prediction on for A84. I am not an expert on this, but it seems that all you really have is a probability distribution, and all you can really do for predicting A84 is to randomly enter numbers into A84 based on that distribution. It is not the kind of programming I do, so I can't think of the formulas immediately off the top of my head, but I'm sure that something based on the built in RAND() random number generator function could generate a random number in A84 based on the distribution in D1:D4. If this is indeed what you need help doing, let us know and I'm sure someone here can help you generate this random number.

If your sole interest really is to generate such a random # in A84, then that is all there is to the problem. However, I suspect that there is more to the problem after picking that random number. When analyzing these kind of "random" events to attempt to make predictions, we often use a strategy called a "Monte Carlo" algorithm. The basic idea behind a Monte Carlo algorithm is:
a) select random input number (based on any assumptions or distributions that you may want to impose on this random number).
b) compute desired output value or values based on that input value
c) store results
d) return to (a) and repeat the process with a new random input number
e) repeat 1000's or 1000000's of times
f) analyze the collection of random inputs and outputs for patterns or probabilities that inform your desired forecast.

Monte Carlo algorithms can be fairly simple or quite complex. Due to the large number of iterations needed to get a meaningful analysis, they are usually computationally intensive. It gives us a chance to analyze these seemingly random processes and try to make decisions.

If you need further help understanding Monte Carlo algorithms, I would suggest starting with Wikipedia: http://en.wikipedia.org/wiki/Monte_Carlo_method
http://en.wikipedia.org/wiki/Monte_Carlo_algorithm
or put "Monte Carlo algorithm tutorial" or similar into your favorite internet search engine.