• Guest, Forum Rules - Please Read

    We keep things simple so everyone can enjoy our community:

    • Be respectful - Treat all members with courtesy and respect
    • No spam - Quality contributions only, no repetitive or promotional spam
    • Betting site owners welcome - You may advertise your site in the Betting Picks or Personal Threads sections (minimum 3 posts required before posting links)
    • Stay on topic - Keep discussions relevant to the forum section you're in

    Violating these rules may result in warnings or account suspension. Let's keep our community friendly and helpful!

Quantitative approach to betting anyone?

Alex Smith

Alex Smith

Casual Punter
Joined
Mar 31, 2014
Messages
5
Reaction score
0
Points
1
Has anyone considered quantitative/algorithmic approach to betting? I've successfully used it for about 5 years on betfair. Now mostly for Horse Racing. But researching other markets as well.

It is really strange but not many people use this approach despite that it is very straightforward and no guessing involved.

Please share your experience if you have one?..

Also I would be happy to show you what I've got, let me know if you interested ...
 
MadMarket
Well, I am interested to try Horse Racing just for fun. Maybe you can write about "general rules" and your own tactic;)
 
That's quite interesting Alex, I hope you can share it..
I hope it is not that hard to follow... :D
 
Well, General rules are quite simple, there are two different approaches I use:
1. Get specific market or related markets, consider some theoretical dependency, check this dependency on historic data(confirm or disprove theoretical dependency)
2. Get historic data for specific market or markets, find not obvious dependency by running through different options.

Real Example of first approach:
1. I assume there is a dependency between Horse Racing Win and Place market, which means there should be a correlation between two of them (if Win price(Back/Lay) of particular Selection moves then Place price(Back/Lay) of same Selection should move the same direction).
2. Get Historic data for Horse Racing Win and Place Markets
3. Get a number of Selections (preferably with big volumes) and check correlation exists on pre-play and on in-play market.
4. Correlation confirmed for pre-play markets, correlation disproved for in-play markets.

Real Example of second approach:
1. Get Historic data of some market with big volumes (means big opportunities)
2. Observe data under different angles and try to find dependencies (I use tools like SQL DB + Excel with charts and macros)

Second approach is a big less straightforward and probably better suits someone with experience of Quant betting.

Both approaches produce a set of execution instructions which potentially can bring profits, so next stage is always testing of these instructions on real markets.
 
I appreciate the details but I think it'll take some time for me to process this.. whew!!
Thanks again Alex! ;)

Alex Smith said:
Well, General rules are quite simple, there are two different approaches I use:
1. Get specific market or related markets, consider some theoretical dependency, check this dependency on historic data(confirm or disprove theoretical dependency)
2. Get historic data for specific market or markets, find not obvious dependency by running through different options.

Real Example of first approach:
1. I assume there is a dependency between Horse Racing Win and Place market, which means there should be a correlation between two of them (if Win price(Back/Lay) of particular Selection moves then Place price(Back/Lay) of same Selection should move the same direction).
2. Get Historic data for Horse Racing Win and Place Markets
3. Get a number of Selections (preferably with big volumes) and check correlation exists on pre-play and on in-play market.
4. Correlation confirmed for pre-play markets, correlation disproved for in-play markets.

Real Example of second approach:
1. Get Historic data of some market with big volumes (means big opportunities)
2. Observe data under different angles and try to find dependencies (I use tools like SQL DB + Excel with charts and macros)

Second approach is a big less straightforward and probably better suits someone with experience of Quant betting.

Both approaches produce a set of execution instructions which potentially can bring profits, so next stage is always testing of these instructions on real markets.
Click to expand...
 
Thanks everyone for replying. Feel free to contact me if you have more question. Also please follow me on betfairquant.wordpress.com
I'll try to update blog regularly. Next week I'm planning to put a very specific example of quant strategy.
 
Additionally, where can I subscribe for an alert every time you have an update here?
 
Back
Top