Less Is More

[ricardodeano]

Hello folks.

This post is really aimed at those just starting out in the world of automation. I include myself in this group (despite labouring in practise mode for the past 6 months ) so apologies to the old hands if this sounds patronising or condescending.

I do not have a golden ticket to give away, I am very much in the process of being in the hunt for a profitable rule set (or two!) myself but I thought I would share some insight (and a spreadsheet) that may assist those finding their feet.

Like many before me, I set out to discovered a "One Rule to rule them all." Sadly, I'm 99% convinced that such a rule does exist and this where the concept of 'Less Is More' comes into play, as does the spreadsheet.

I've started this week on a new test rule and I am using the attached spreadsheet to monitor it's progress. Obviously it's only in it's infancy with a small amount of data (57 races) but it does illustrate the 'Less Is More' concept.

So far, the Original rule returns a loss of -£25.88 for the two days. Not a huge loss but not a profit which of course makes me sad. However, if we do a bit more digging and exclude certain parameters, in this case race type and/or distance, we can see that there are certain distances that are really dragging this rule down.

To get to the headline:

If I were to rule the against all races, I'd be in a negative position (see Original, col D).
However, if I select fewer races based on certain race distance criteria i.e. excluding 5/6/7f races, I'd be in a profitable position (see Excl 5/6/7f, col M).

And this is exactly what I mean by 'Less Is More'. You may think you have a negative rule set, but take a step back and see if there are common factors, in this case distance, that are causing this negative result. If it's consistent, filter them out and take it from there.

By all means, take the spreadsheet away with you and use it for your own purposes. You may want to record further parameters (e.g. number of selections, going, etc.) and maybe, it'll help you find an edge in your rule set that you previously didn't think was there.

Anyway, apologies for the long post and I hope it helps someone out even if I can't find a profitable rule set myself!!!

NOTE: You may also want to check the spreadsheet's formulae in case I've made a balls up

[jimibt]

Rich,

Thanks for the info and the attached. I've definitely concluded the less is more mindset as well as I often see a perfectly good day start to get trashed either by distance types or courses (Yes, Ballinrobe and Cartmel, I'm talking to you!!). By focussing attention AWAY from the uncertainties, I too have found that my net result can be dramatically enhanced. Of course the trick is to have the luxury of time to discover these anomalies and be able to hone your rules to fit your filtered list. It looks like you're well on your way to doing that!

As an aside, one further observation that I have made is definitely related to the spread of the odds on races between 8 and 13 runners. A tightly packed range of about 5-6 runners here makes for great partying on short term B2L and L2B's (just picking a few ticks and being disciplined).

Now, if only we had a *replay market* function

Keep us posted on progress... and thanks again

[Blackmoor]

Hi Rich,

I'd say you are taking the correct approach, but I'd want a lot more sample data before I started excluding distances or number of runners.

One of my systems initially ran for a short time - 3 weeks or so, and after initial analysis I jumped to the conclusion to exclude certain race profiles. As I was still in the testing phase, I didn't actually implement the exclusion, as I was regarding it as a data collection exercise really. Over the longer term, these race types have proved to be the best performing - it's just that the win rate was different to the other races, but yielded greater profits. Had I implemented the initial exclusion early on, I would have missed out, and would have a worse performing system.

The key thing is to remember that you have to lose some of the time..! - and it's about taking the longer view. Personally I assess my systems on a monthly basis. If you added exclusion criteria after every losing day you wouldn't have a system. I am sure that many good systems have been abandoned too early, because it was decided they were bad/flawed when in fact they where they did actually have potential - had they been given time, and had sufficient data analysed in order to create appropriate filters.

Good luck !

Steve

[ricardodeano]

Hi Steve

Absolutely agree with you and I have learnt that taking a short term view is not a good idea. I just wanted to put down a small amount of data just to illustrate the point.

I'd previously done a similar thing for a much larger sample of data but I didn't want the file to be massive and the two days on this new rule set highlighted the difference quite well.

Onwards and upwards!

[jimibt]

Hi Steve

Absolutely agree with you and I have learnt that taking a short term view is not a good idea.

yup, as the song says, what a difference a day makes... still a mystery to me how a set of seemingly random events can conspire to make one day a cracker and the next one a heartbreaker

[warren0]

Hi Ricardo

Just curious but what were the stake sizes on your positions for these initial 57 races? As in did you keep it uniform or did they vary?

Interesting thoughts on it though and thanks for sharing

W

[ricardodeano]

Hi Ricardo

Just curious but what were the stake sizes on your positions for these initial 57 races? As in did you keep it uniform or did they vary?

Interesting thoughts on it though and thanks for sharing

W

Hi Warren

Stake size was a single £50 bet placed just before off time. No variance in stake, just greening up when certain criteria were met.

I'm on my hols this week but hopefully it's been doing ok in my absence!