### FAQ for active users

#### What is ThoroughBet?

ThoroughBet is the proprietary up-to-date scientific system which analyzes data of every horse, jockey, weather, state of ground, punter’s expectations and much more in order to predict the outcome of the every horse race in UK. In other words, we’ve created a tool that allows you to find the most advantageous bet by weighting the available odds against the horse’s actual chances of winning the race. We do not claim that our system can predict the winner of every race - this is simply impossible - or that you will win every time after following our recommendations - even though the percentage of the correct predictions of our system is higher than any other competitor's system. However, we do claim that using our tool correctly will make your profits exceed your losses by far.
In order to get an idea and to assess the quality of our products, the simulation section allows to play around with the data, develop your own strategies and to run simulations of your profits if you were to use our services in that past. And of course, all our predictions are **genuine and verifiable**. Just jump to the **simulation section** and find it out for yourself!

#### Why it works?

Trying to pick a winner is like fishing in muddy water. If you’ve tried it, you know it isn’t easy. In essence, the present betting markets are so advanced that a high level of proficiency in statistics, probability theory and machine learning is necessary for success. We possess the necessary background – and thanks to this background, we’ve developed a sophisticated mathematical model that analyzes huge amounts of data using complicated statistical and machine learning algorithms. Over the past years we’ve collected a substantial data on horse races, both by buying them from services and by scraping them from various open sources. As a result, we have a fairly comprehensive picture on the races. At the end of this process, the system computes the winning probability for every participating horse in every race and compares it with the bookmaker’s odds in order to find the most advantageous bet with the highest expected profit.

If you still have doubts – and you should, as there’s an incredible number of internet scammers who promise huge winnings and non-losing strategies – be sure to check the **archive section**. There, we hash and timestamp our predictions every morning, as soon as they’re available. The hash is stored by different timestamping authorities and even incorporated into the bitcoin blockchain. That way, post-hoc manipulation of our predictions is provably impossible. Any post-hoc manipulation of the files by us or third parties would produce a different hash, and any manipulation would be immediately visible.

#### What is the Thoroughbet Estimate (TBE)?

It is our best guess of the odds that a horse will win the race. It is the result of years of hard work and research all put into a single number. If the TBE equals 8, then we think that the horse will win 1 out of 8 times.

#### What is the Market Price (MP)?

Apart from our own estimate, all other participants in the betting markets also produce their own guesses that are collectively represented by the publicly offered odds. We use the MP that is available a few moments before the race actually starts. For example, if the MP equals 10, then all market participants seem to think that, on average, the horse will win 1 out of 10 times.

The overall idea is simple: if our estimate is lower than the publicly offered odds, that means that we think that the horse has a higher probability of winning the race - thus we should back it. If it is higher, we should lay it (bet against it).

#### What are model and market coefficients?

The above idea has a little complication: how confident should we be about our own estimate? After all, millions of punters have their own ideas and sources of information about the races which are all expressed in the public odds. To let yourself guided only by your own opinion would therefore be too overconfident and foolish. What is needed is a superior estimate that combines your own opinion and everybody else's and weighs them appropriately. Those weights are the model and market coefficients.

#### What is the mixed estimate (MXE)?

It stands for mixed estimate which is the combination of your estimate (TBE) and the public market odds (MP). Of course, if you use public odds other than final exchange market odds, for example odds offered by bookies (Industry Starting Price), the mixed estimate will vary. However, the TBE is available in the morning and you can mix them with any other opinions that you get your hands at. However, if you use odds other than the MP, you may want to adjust the model and market coefficients. After all, a fool's prediction should be weighted less that a smart guy's one.

#### What is the expected return (ER)?

In order to know how much to place on a horse, we have to compute first, how much money we can expect to make with that horse for each pound placed on it. This is called the expected return (ER) and is computed by ER = MP / mix_est - 1 For example, if our final guess, the mixed estimate, is 8 and the MP is 10, then ER = 10 / 8 - 1 = 0.25. That means that for every pound bet on that horse we can expect to make a profit of 25 cents, on average.

#### What is the Kelly criterion?

Given our expected return and our bankroll, how much money should be put on a horse? Obviously, we shouldn't put too much money on one horse, since chances are that you will lose it very quickly as you bet repeatedly. On the other hand, we shouldn't put too few of it, since then we don't make much money. And further, the amount we bet should depend on the amount of overall money we have - the bankroll. Thus, we bet a fraction f of our bankroll computed as f = ER / (MP - 1) If the above example, if the MP=10 and ER=0.25, we get f= 0.25/(10-1) = 2.77%, which means that the Kelly criterion recommends to place 2.77 percent of our bankroll on that horse.

#### How to choose the final wager?

It is often reasonable to wager even less of what is recommended, which is scaled down further by the Scaling parameter, for example 0.5. If our bankroll is 1000 pounds, the wager on the horse is then 0.5 * 2.77% * 1000 = 13.88 pounds. If the Kelly fraction f turns out to be negative, the horse should be laid (bet against).