Programmers using the Betfair API SOAP technology are able to code their projects in VB, C, C++, dot net, java, literally anything. Low level languages such as these have one great advantage, they work from the client workstation and they are extremely FAST.

"What exactly is and what can be achieved with a 'Bot'?"

There are a few basic tenets that need to be clarified and explained with respect to the whole arena of Betfair API technology, interfaces and Bots.

Bots simply represent

Therefore if the “system” you are working to manually cannot provide success, then no matter the level of automation you implement, the automated “Bot” will also fail.

Bots are simply automation of systems and procedures that you would have performed manually via a standard or enhanced interface and manual input operations.

As discussed at the outset of this article, the real truth here is that there are NO paper based racing systems that can work all the time without human intervention at some juncture in the procedure, should certain conditions arise.

Therefore by projection I can state with 100% certainty that

However this by NO means represents the end of the automation story. It is in fact only the beginning.

There are two basic formats for Bot functionality.

1. The first is “Technical Analysis” (trading price movements and fluctuations in the market)

2. The second is “Form Study” (fundamentals of the horse and the race itself)

The general feeling in the industry is that a potentially successful bot would be one that could target “value” in a race or sporting event.

It could hypothetically be possible to develop a computer system based on massive datasets that generates a predicted SP for a given horse in a given race based on fundamental form data analysis of past performances.

By comparing your predicted SP with that on display in the shops, or on Betfair or in the press, it would not be a huge leap to develop a simple automated system that watches the horse SP on Betfair throughout pre race, and place a bet on it should the horses value drift out to what your dataset analysis had determined as being its correct SP, this would represent a “value” bet which over a period of time should result in profit to level stakes.

Part of the reason that no one has been able to successfully do this is that the concept of “value” cannot be accurately determined itself.

Value is a highly subjective aspect of racing.

In support of my theory comes the following nugget of information.

“In 2005 Betfair datasets were analyzed after allegations of race fixing or non-trying to be precise. The theory being that if you know which horse is definitely going to lose then it’s particularly easy to profit from lay betting on Betfair.

The allegations were disproven but other results that were generated from the dataset analysis were particularly surprising.

Fewer “drifters” won races compared to “steamers” as one would expect, but in terms of profiting from the results it was determined that

Why is this data SO important?

But where does one find “value”?

What I or my computer program sees as value in a race, another bettor, or shop, or computer system which has been implemented with a different set of rules will most likely NOT see the same value.

Similarly the “value” can change at any point during the day pre race, as value is linked not just with fundamental form, but also incoming news.

Incoming news factors include.

Any single one of these incoming news factors can influence the “value” that I saw based purely on fundamentals and SP.

Part of the big problem with bots is that they are incapable of responding to news and events pre race and other activity in Tattersalls and when going down to the post.

A bot cannot analyze and make a deduction from a piece of news blurted out on C4 racing by the insufferable but brilliant JM.

All a bot can do is respond to changes in a horses SP. That is to say its activity can only be purely mathematical in its analysis.

A novice punter will already have discovered that racing results are multi factorial in nature with a good dose of luck thrown into the mix.

Pre race issues that will arise dozens of times per day include:

Any or indeed all of these are factors that affect whether a professional punter in the ring would actually physically place a wager.

The fully automated value betting computer system would simply be watching the SP and bet according the rules you have pre determined.

So we can see that attempts at full automation of racing systems have so far ended up in the pile labelled GIGO (Garbage In Garbage Out).

There are fully automated systems on the market which use “Martingale” strategies.

Martingale is a staking plan which places bets on favourites in the win or place market and increases the stakes after each loss in order to achieve a pre determined profit on each race.

The theory is based on the fact that at some point one of your horses will inevitably win, which is of course true.

These fully automated systems are popular with mug punters because, in exactly the same way as “Lay Betting Systems”, they can give the impression that they are working for quite some time, especially if working with each way markets.

The problem is that although win frequency is high, when the inevitable losing run of 5 or 6 races comes it is utterly devastating and will wipe out your entire pot.

To cover these losses you would have to remortgage the house and the bet would not be accepted by any shop or exchange anyway.

By that time the makers of the system will have already taken their money off you and are ready for the next mug punter.

So it can be seen that the requirements and functional spec of my systems had a lot to live up to and high expectations to achieve.

In Part 3 i provide a more in depth analysis of how I moved from design into implementation and testing.  

Technical Analysis Of Semi Automated Betting Systems Using Bet Angel Professional Trading Interface, Bookmaking And Dutching. Part 3.