Kaizen – improving automated strategies

I’ve been focussing more on the automated tipping strategies that I created of late, certainly when it comes to horse racing. I’m not as good a judge nor as knowledgeable as a lot of horse racing tipsters and I think it’s best to focus on what I am good at. There are 4 main strategies that we produce the tips for automatically; these being Market Movers, HorseLays, Bet Monkey and Footy Movers. In the past I’d come up with a theory and set some criteria for the automation, let the tips fly for a while and look to change at some point if it obviously wasn’t working. Just the other day I was reminded of the Kaizen methodology and I felt inspired to commit to this kind of approach, not just with the strategies in fact but the BettingTools website as a whole.

The Japanese word Kaizen translated to English simply means “change for better” but it’s generally accepted (or at least is the understanding of most) that it represents continuous small improvements. Quite often the task of making something better overall is all too daunting and can seem impossible but by breaking things down looking for small improvements, it is more manageable and can have a great effect. We’ve seen just how powerful getting half a point extra in odds can have to the bottom line and I’m now looking to implement this kind of approach in most things I do.

I thought it would be a good idea to write about the thinking behind the automated strategies, problems that have been encountered and recent changes that have been made in trying to improve them. Hopefully by showing you what has been considered so far, it might trigger some ideas for further improvement.

Market Movers

We’ve had a few iterations fo the Market Movers. Originally we listed anything that shortened by x percent in probability which was disaster! Nearly every favourite was picked up and the profile was not very profitable at all! So we eventually pressed the reset button on that and filtered in a minimum price. That worked much better particularly over the summer months but not so good during the National Hunt season. There were still a lot of selections and losing streaks were still pretty long at times.

Since then I’ve begun to look at number of runners, how much has been matched and I’ve also made sure that selections aren’t created too close to the off. Movements right before the off can occasionally be a tell-tale sign but in my experience as a whole they come to nothing far more often. You have to remember that right before the odds the horse racing traders (scalpers) are pushing big money around to move the market for their own purposes. So for this reason I’m now focussing on earlier moves and mostly trying to look beyond the favourite.

This has seen fewer selections but even more crucially, they tend to shorten even further and I’m pretty sure they are beating the Betfair Starting Price (BFSP) on average now. I am going to look at adding the betfair start price column to the main tipster results page so people can compare for themselves. Sometimes tipsters can be going through a bad spell but are still picking the right horses and this should be helpful in determining whether this is the case or not.

Potential improvements could consist of looking closer at the numbers of runners and money matched as well as analysing the types of races (Groups, Handicap, Novices. Distance etc) individually. Currently I’m treating them all the same.


Horselays were unbeatable once upon a time and had a lot of followers. This was back when I was looking at the market myself and picking low price losers. Attempts to program something identical to my thought process and analysis of the market graphs on betfair have not been as fruitful unfortunately. Again, looking at the number of runners and looking for a decent sized moves, I also seem to have had more success sticking to handicap races.

The last few months showed tiny profits but nothing significant and the number of selections were small. I tinkered a few days back but it didn’t seem to work and this profile definitely needs some more thought. It’s back to handicaps only for now. With liability usually exceeding any potential profits and not providing the buzz of a big winner, I think the demand for these is lower and so I think they’ll remain the lowest priority out of the four profiles.

Bet Monkey

Bet bloody Monkey! Once, an unbelievably successful football tipping profile that just kept delivering big winners. A pick of the Holy Grail selections which although had early success ultimately fell on its arse when showing it to the world. So what went wrong with these tips I hear you ask. Well, I believe there were 3 main reasons for their eventual failure.

1) We started listing tips with exchange prices – The football betting markets on betfair are a lot less liquid than they used to be and there ar often gaps between prices. In theory our model has still seen value in the selections but in reality we should be getting better prices and it ultimately affects the bottom line.

2) The loss of data from the National League – The National League / Conference or whatever you want to call it was unmistakably our most profitable league for these selections. A query done just yesterday showed that these accounted for over half the overall profit of the Bet Monkey profile!

3) More efficient markets – A bit of a guess this one but when we started on these selections some 7 or 8 years ago (maybe more actually) and were analysing mainly using shots and form data the level of analysis that went into producing the prices by the markets were likely to be much simpler than they are today. With increasingly more people getting into data collection and the introduction of metrics like Expected Goals (xG) it is more difficult to find an edge.

Given the above, it’s no surprise to see Bet Monkey posting pretty dismal (although not toto destructive) figures over the last year or two. Whilst I don’t think it’ll be realistic to see the sort of returns we did when betting on the National League, there are a couple of simple improvements we can make to give ourselves a fighting chance. So I will now be listing these against bookies prices once again and will focus only on Leagues 1 and 2. From yesterdays’ analysis it was quite apparent that the Prem and Championship selections are not really profitable overall. It makes sense that the more popular leagues, where there’s much more money, would have tighter prices. We may have picked up the Huddersfield win against Man Utd on Saturday but it was a fluke and the Prem sections are not performing well overall.

Footy Movers

Similar to the original Market Movers criteria these are currently very simple. The football markets can be ridiculously thin in terms of liquidity leading up to match day though and I was picking up ‘moves’ that weren’t really moves at all; just money taken out of the market and the next available price to back was much lower. What I now do is check the best price available to lay too to see how tight the market is before declaring it a ‘move’.

This has definitely improved selections and many more of them seem to further decrease in price than they were doing before but I think I can go further than this and look closer at amounts matched. Again like the Bet Monkey tips, these may look better if listed against bets bookie prices and I of course advise anyone following to back these with the bookies when the price is better.

I hope this has been insightful and gives some hope to anyone looking to follow any of these selections. If you have any ideas for improvements, please let me know.