System Builder
 
Notifications
Clear all

System Builder


Chris Higgins
(@that-higgins)
New Member
Joined: 9 months ago
Posts: 4
Topic starter  

Hi

Has anyone found a successful system in system builder

I've always thought that if a system really works based on past ratings then it already exists and therefore logically would be know to those providing a system builder

This is not meant to  be a criticism of Race Advisor and I hope it wont be seen that way but I am struggling to find a reliable system

If anyone can suggest settings that can be created in the new system builder apart from the Strategies perhaps they can be shared?

 

Thanks

Chris

This topic was modified 6 months ago by Chris Higgins

Quote
Avatar
(@mikeywilding)
Member Admin
Joined: 12 years ago
Posts: 2710
 

What are your parameters for considering successful and reliable?

Follow the Race Advisor on social media on Facebook, YouTube, Twitter and Instagram. You can also find my personal accounts on Instagram and Facebook
.


ReplyQuote
Chris Higgins
(@that-higgins)
New Member
Joined: 9 months ago
Posts: 4
Topic starter  

I was hoping you could tell me


ReplyQuote
Avatar
(@mikeywilding)
Member Admin
Joined: 12 years ago
Posts: 2710
 

No problem, I'm sure others will chip in when they next visit.

It's very similar to form reading, and should be thought of in that way, the difference is you're just doing it in advance to find contenders easily.

This means you should focus on specific race conditions. If you take all race conditions it's too broad. Even if you just take a race type, such as All Weather, this is too broad. Horses that perform well in sprints will be different to those who perform well over middle or long distance. Horses that like Standard To Slow will be different to those that like Standard To Fast. Horses that like small groups of runners will be different to those that like big groups.

The most common mistake is trying to build a system to find contenders which covers a lot of conditions. It can be done, but it's much harder. It's far better to start with micro-focuses and build lots of them. For example, All Weather sprint handicaps on standard ground at a specific course is an example of some of what I have.

There is a balance between going to narrow and not having enough data and too broad, for this you want to have at least 100 winning bets, and if you've not got enough data to do that a few nodes down, then you maybe need to broaden the conditions slightly, but only do this slightly to bring in enough.

When setting up I would recommend you use Out Of Sample at 20%. The number of races in here, if you're following the above advice, will be small, but it's enough to make sure the figures are in a similar ball park. If they're not, then I wouldn't consider it stable.

Next I would start by looking at a rating group that you think is most specific to the race conditions, e.g. sprint race use Speed.

When you start try to only use a maximum of one rating from each rating group. You probably won't even need that many. A lot of my systems to find contenders only have a couple of ratings in.

Get the A/E to at least 1.03 as a minimum, but 1.05 or higher is ideal.

System Reliability should be 80% or more, but you can go a bit lower if you don't have enough data, just be cautious until you get more data and stability in.

As you're building something, if it's not working well, flip the bet type in the settings. You are building a Back To Win system, and the results are horrible, flip it to Lay To Win or Lay To Place and see how it performs, if it's horrible enough backing it may perform well with one of the lay bets. Lay To Place is a largely under-used market.

In the nodes, play with the settings for a single factor. Try putting in a number that the rating should be higher or less than. Change this around and see how the stats change, then remove the number and use a top or bottom percentage, then use a top or bottom number of horses, it gives a feel for how it performs under the set race conditions. If you can't get it work but think it should, change the rating out for the Rnk, DiffTp or Raw version depending on what you're using and repeat. This whole process for a single node is likely to take around five or ten minutes. Of course it could be a bit longer, or a bit quicker if you find something that works quickly.

Always keep an eye on your strike rate and biggest expected downswing, this gives an idea of bankroll requirement. The ROI should be 5% or higher, however for lay strategies I would personally drop this down to 2%.

I am holding a live training session for members tomorrow at midday, and it's going to be on the system builder, I will be sending links out in the morning.

Follow the Race Advisor on social media on Facebook, YouTube, Twitter and Instagram. You can also find my personal accounts on Instagram and Facebook
.


ReplyQuote
Chris Higgins
(@that-higgins)
New Member
Joined: 9 months ago
Posts: 4
Topic starter  

OK Thanks Michael...I look forward to the session...I guess I'm asking if any experienced users in the group have managed to create any successful systems that we could all use?


ReplyQuote
Avatar
(@karlinator)
Active Member
Joined: 1 year ago
Posts: 11
 

Ive just received an email for a live session on the system builder.

Please make a recording 

Im currently at work and i wont be able to attend the session.

I would love to watch it when i'm available to do so, tonight.

Thanks


ReplyQuote
Avatar
(@mikeywilding)
Member Admin
Joined: 12 years ago
Posts: 2710
 

@karlinator we have made a recording, please contact our support team and we'll send you a link to it as soon as it's available.

Follow the Race Advisor on social media on Facebook, YouTube, Twitter and Instagram. You can also find my personal accounts on Instagram and Facebook
.


ReplyQuote
Avatar
(@carlnewbury72)
Honorable Member
Joined: 3 years ago
Posts: 738
 

Also missed this today 


ReplyQuote
Avatar
(@mikeywilding)
Member Admin
Joined: 12 years ago
Posts: 2710
 

@carlnewbury72 I've sent an email out which you should have received with the link to the replay now 🙂

Follow the Race Advisor on social media on Facebook, YouTube, Twitter and Instagram. You can also find my personal accounts on Instagram and Facebook
.


ReplyQuote
Avatar
(@ladymansean)
Trusted Member
Joined: 1 year ago
Posts: 63
 
Posted by: @that-higgins

OK Thanks Michael...I look forward to the session...I guess I'm asking if any experienced users in the group have managed to create any successful systems that we could all use?

I'm certainly no expert and have been 'playing' with different settings in the S.B. just to see what is thrown back, the following may be of interest as a starting point.

I selected all uk races, good and soft ground, 1100-1200 yards distance, out of sample 20%, date range 1st Jan 19 to 1st Jan 21

Node 1: SPDFIGLrAdj

Selected Yes

Results

34races, 46selections, 9wins, 19.57%win sr, 29.67 profit bsp, 64.5roi, 1.64a/e, 90% system rel, avg odds 15.15

8.56avg win odds, 16.8avg losing odds, 4 races per mth, 5 selections per mth, -9.26largest drawdown, 18largest 

Expected losing run, 14.66profit at sp, 31.87roi at sp.

I agree that it's great to support each other but would not expect someone to share a proven system that they have probably worked some time on in proving the reliability.

Maybe though, we could invite criticism on systems that we create by posting such as I have done so here?


ReplyQuote
Avatar
(@jon-roberts-2)
New Member Customer
Joined: 6 years ago
Posts: 3
 

@ladymansean

Hi Sean

Just curious what your parameter was for SPDFIGLrAdj, as you don't indicate. It's a filter I use, though I tend to use the Rnk version of it

 

cheers


ReplyQuote
Avatar
(@ladymansean)
Trusted Member
Joined: 1 year ago
Posts: 63
 
Posted by: @jon-roberts-2

@ladymansean

Hi Sean

Just curious what your parameter was for SPDFIGLrAdj, as you don't indicate. It's a filter I use, though I tend to use the Rnk version of it

 

cheers

>=20

 

Forgot to add that detail in, sorry


ReplyQuote
Chris Higgins
(@that-higgins)
New Member
Joined: 9 months ago
Posts: 4
Topic starter  

Hi Sean

Thanks for responding I appreciate that some people would not want to share systems and if you add the human element of FMFR to those computed historical results who is to say we wouldn't end up with different selections anyway therefore affecting the ‘system’

Cheers

 


ReplyQuote
Share:
Back to top button
Close