Punter's Genie

What's this website all about? [TOP]
As it says on the home page, we want to do the following:
Provide a continuous stream of predictions for as many races as possible with a strike rate and winning odds to guarantee a long term profit, using prediction models that we will constantly strive to improve.
We are tackling the problem of horse racing prediction from a mathmatical/statistical perspective and we have designed our models and methods to provid a long term steady profit. Our first models set out our baseline for returns at 135%. This provides proven profits over our test period (which is currently 4 months worth of races). We can then hone this model to improve the strike rate and average return. Essentialy we start from 135% and only change the model and race selection rules if they improve that return of the long term. This happens quite a lot and returns should be currently much better than our baseline (as I write this I'm just coming out of a week having made a 170% return - nice and steadily, no praying for the long shots. That's across nearly 60 bets so I didn't rely on a couple of nice 10/1 shots coming in).

If you're so good, why don't you just gamble rather than setting up this website? [TOP]
Good question. As I mentioned in the last answer, we do gamble ourselves. However, we just aren't the high rollers. Maybe in the future we could make a living out of our own bets, but at the moment our banks are small and out stakes in the 5 to 10 quid range!
Besides, we are interested in the whole process of the modelling and prediction of horse racing. It's an interesting problem to try to solve. I think we've done a pretty good job so far, and I don't see why everyone else should miss out on benefiting from our work.

Why are you doing this? [TOP]
Because we know the theory, we know the problem, we constructed a solution and it worked :-). Oh, and it keeps the old grey matter in shape - it's good fun for us maths and IT types to tackle an open ended problem like horse racing. Especially if you find a solution.

Why are your methods better than anyone elses? [TOP]
We believe that out methods are not yet being used in the field of horse racing. They are used in high finance, customer relationship modeling and mathematics. Others have suggested that the types of models we are using could be applied to horse racing, but usually in the conclusions of academic papers. We have the experience, skill, knowledge and drive to make these models work for you.

What sort of history do you have? [TOP]
Between us we have a couple of theoretical physics degrees, a masters degree and a year studying some proper-crazy maths at Cambridge. We also have 8 years in the IT industry and 6 in forcasting and mathematical modelling. We've been gambling for about as long as we can remember, and, yes, we have probably fallen, at one time or another, into ALL of usual traps laid out for gamblers. It's possibly because of this that we are so driven to build these models that will allow you to beat the bookies!

How good are you? [TOP]
Pretty good, we think. Without blowing our own trumpet too much, of course. As we mentioned in "What's this web site all about", above, we have a base return of 135% that we work from. Over a year this isn't a bad return - certainly better than a bank account. Say you had a £10,000 bank to bet from, and that is all you bet all year. You'd end up with £3,500 extra in your pocket. Thats OK. However, because we provide so many races to bet on each month (for the non-handicap model alone there could be 100 eligable races per month in the middle of the season), you could easily bet this amount each month. That's £3,500 per month.

How we do what we do

How do you build the prediction models? [TOP]
We can't, of course, tell you exactly what we do - we have spent a great deal of time and effort putting these models todether. Basically, we take as much historical data from racing as we can and try to find the things that are important for predicting the outcome of future races. Once we find those things (in some cases we can be working with tens of thousands of combinations of variables), we then run the models against a test set of races and look at the results. We then look at the races that we predicted correctly and work out what things made that possible for those particular races. This is how we come up with the Bet Index. This information is then fed back into the models to make them better. This is why you will see us update the models fairly regularly.

Which races do you model? [TOP]
We are currently modelling all races except Handicap Maidens. Whatever we do, we just cant find a model for these that is very profitable. This could well change in the future though.

How early can you place the predictions on the site? [TOP]
Usually the predictions wil be up by the end of the day for the next day. Because we need to do some manual work with the data before it gets posted, it is possible that they may not be up until the morning of the race day. Either way, they will be available well before the meetings start.

Why does the Non-Handicap model pick out mostly favourites? [TOP]
Because mostly favourites win the non-handicap races!

How many races will have bet recommendations? [TOP]
We will endeavour to post recommendations for all flat races. However, due to various circumstances, there may be days where we can't post the predictions for some meetings, or where we can't post any predictions at all. We will try to kepp these days to a minimum, as after all, this whole system is based on the fact that we are dealing with as large a number of races as possible.

How to use what we do

What's that table in the Predictions section all about? [TOP]
The table may look pretty complciated, but it contains a wealth of information that shows how we make decisions on which bets to place. The main columns you will be interested in are the "Graph", "Bet Idx" (Bet Index) and "Stk" (Stake). The Graph gives you a quick indication of the win-probabilities of the horses in the race in relation to each other. The Bet Index column shows you how much strength there is in the prediction, with 100 representing an average strenght prediction. The Stake column shows what our recommended stake would be (in points) for this race.
For a fuller description of the table, look [[Link in here]].

What are the numbers in brackets after each course on the home page? [TOP]
These represent the number of races at that particular meeting.

Why aren't all the races available for a date listed on the Predictions page? [TOP]
There are various reasons why any particular race may be missing. For example, we may not have enough data for that particular set of circumstances to produce sensible predictions. Or the race could have been abandoned. These are rare events though, and predictions will normally be missing only because we couldn't get the data processed and on to the site for that day.

How do I get access to the predictions? [TOP]
For the moment access to the predictions is free - you just have to become a member and off you go! Click [[Link in here]] to join.
We will soon be moving to a "paid-for" system for access to the predictions. When we do, members will need to subscribe to the prediction service on a daily, monthly 6-monthly or yearly basis. The fees that we will be charging can be found [[Link in here]], and compared to other similar sites, we think you'll find them pretty favourable!

Why do I have to subscribe even though I'm already a member? [TOP]
We think this is the easiest way to give as many people as possible access to our services. By becoming a member you will gain access to certain parts of the site, but only by taking out a subscription will you gain access to the predictions. However, they are all free at the moment, so make the most of them!

What else do I get access to if I subscribe? [TOP]
We will be publishing all sorts of analyses of our data that will help you make decisions about your racing bets. These will all be available to subscribers.

And if you have any problems...

I can't log in even though I have successfully joined. [TOP]
If you are having any problems with your account, please contact us using the [[Link in Here]] Contact Us page. We will be putting more tools on the Members page to help you manage your account.

How do I cancel my membership? [TOP]
For the moment, if you wish to cancel your account, please contact us through the [[Link in here]]Contact Us page with your full name and username. We will then send you an e-mail for confirmation.

I've forgotten my password! [TOP]
There will soon be a facility to have your password automatically reset by using the details you provided when you joined, but for the moment use the [[Link in here]]Contact Us page to request a password reset.