Multiple Greyhound Strategy

[mgreenough]

I have been on this forum for many years now and thought I would put a post detailing my journey through the world of BetAngel and Betfair, and share some advice from my own experience.

First of all I would like to thank Peter for such an awesome tool and Dallas for the many shared files that allowed me to learn how to create my own custom files.

I have been a Betfair user since 2003 and BetAngel since 2008, and have had varying degrees of success and failure (more emphasis on failure). I have tried numerous systems whether it be backing, laying or trading with very limited success over the years on both the dogs and horses. This year I decided to change my methodology and start looking at data, trying to come up with one or more systems.

I have downloaded the last 4 years worth of greyhound logs from the Betfair prices site, and separated the data into two parts, one being the seen data i.e. historical data being filtered on (back testing and looking for trends) and the second part being unseen data to test out the results from the seen data. This gives an element of back fitting in systems into past data, but then testing them on the unseen data to see if still viable. I played around with the data in Excel looking at various statistics and percentages and found at least 3 systems that were looking like they may work with quite a few modifications.

Once I had an idea of something that may work, I then created the rules file in BetAngel and run them for a few months to gather the viability of the system and gather as much data from BetAngel whilst they are running. This gave me insights as to what works and what does not, also any trends seen whilst running in practice mode.

Now from what I found was that if you left all 3 systems running for every qualified race all day, then you have a trend reversal multiple times a day. Think of the profit and loss like a rollercoaster, it goes up and down regularly throughout the day. So what I did next was to identify the optimum exit point for the day for each system. Currently I am growing my bank so the exit point is 10% of my bank each day for all systems combined, this equates to each system stopping after returning anything over 3.33% of the start bank for that day. I don't do trading as all my bets are either straight Back or Lays. I manually watch the screens to ensure I stop all automation should it start to make a loss for the day that may be hard to recover from, as I know from experience now that a losing day is hard to come back from. So far I have only had two days in a negative position. I know that if I leave them running for a longer period then the likelihood of losing more is inevitable. I would rather cut my losses and start with a fresh mindset the next day. Since today though I have increased the exit point to 5% per system - so 15% per day profit (although I ended up 30% in profit today due to a mistake in my stored value)

What I also found is the various different types of staking that can make or break a system. I had multiple instances running with different staking types, as depending on the number of selections you have, one staking type may be better than another. Only by running side by side can you evaluate this.

To run multiple systems with different staking methods was a problem I had to solve early on, which I managed to do via running multiple virtual machines on a couple of old pc's with BetAngel installed in each VM. This gave me the separation for each system to have a really good play around. After 3 months of testing in their own virtual machine I decided to take the plunge and start live with a £30 starting bank. So far it has been pretty successful and my goal is to get to the point where I am getting £300 per day and then stick with those stakes to start building the bank up and take some profit out for myself.

So the overall takeaway from this is not too get too disheartened if your systems that you are testing do not work.

* Try running them for a while collecting as much data as you can from BetAngel and review that data every way possible.
* Don't always run systems the full day (when live) as you may see record profits and then slowly declines throughout the day into a loss.
* Try identify what time periods give the best results and start you automation from there, and stop all automation when your stop loss or stop profit is reached.
* Test staking types for each system you have, as one may be better than another
* Don't get greedy, you may see big profits on some days but you wont always achieve that. Start small and slowly build up.
* Ensure you run long enough in practice mode to gather your data, don't rush to start with real money until you know it is viable. I have been guilty of this in the past and got burnt.
* Test your strategies as back and lay (in separate instances), you may find it illuminating as one of my back systems eventually became profitable lay system.

Please don't misconstrue the info below as blowing my own horn, as it is still very early for me to do that due to only having the systems live for 3 weeks. It is more for transparency and to show it is actually achievable to get systems that work, and also to chart my progress and keep myself focussed.

Current copy of my excel tracker sheet to show how current live systems are performing, including screenshot from Betfair itself. Hopefully that this post is not a curse and cause the whole thing to come crashing down, but with the three differing systems and 3 months of successful testing hopefully that wont be the case.

[Fugazi]

Well done mate.

Hope it is profitable long term and the hard work is rewarded.I know what a bloody pain it is to do this level of testing.

[mgreenough]

Well done mate.

Hope it is profitable long term and the hard work is rewarded.I know what a bloody pain it is to do this level of testing.

Cheers, whether it succeeds or crashes and burns is anyone's guess but it is the fun of it that drives me rather than the money. As long as my bank does not go below the initial £30 then I deem that acceptable LOL.

[elofan0]

Well done its great to hear success

[mgreenough]

Well today went to shit due to raising the targets for each system from 3.3% to 5%, just that small amount totally screwed me over today putting me back one day losing all yesterdays profit. So reduced back down to 3.3% as reviewing the day in full shown that I would of reached 3.3% on all of them. I broke my own rule of not being greedy

[Crazyskier]

Interesting journey, eloquently explained.

Thanks for posting and please keep updated.

CS

[andy28]

Thanks for sharing, I am doing a similar thing but one difference is I let the day go regardless, if I lose a lot so be it. The reason I let it run is I want to look for seasonal swings and also I want it fully automated so I won't be watching if if hits a bad patch it has to be in the plan. Also I let it run because I did in all the modelling so my thinking is if I control what it does then I am varying away from my trial.

Mine has hit a volatile period with big swings but I am thinking it is seasonal as in Autumn in Aussie you have 3yo v 4yo's+ in handicaps and also 2yo's v 3yo's so weight can be key, some may compete with more weight while others won't.

Also I love the stats part of this

[mgreenough]

Thanks for sharing, I am doing a similar thing but one difference is I let the day go regardless, if I lose a lot so be it. The reason I let it run is I want to look for seasonal swings and also I want it fully automated so I won't be watching if if hits a bad patch it has to be in the plan. Also I let it run because I did in all the modelling so my thinking is if I control what it does then I am varying away from my trial.

Mine has hit a volatile period with big swings but I am thinking it is seasonal as in Autumn in Aussie you have 3yo v 4yo's+ in handicaps and also 2yo's v 3yo's so weight can be key, some may compete with more weight while others won't.

Also I love the stats part of this

Thanks for the reply, I can say this has just hit a bad patch as well, as today was abysmal, it is the first two days where I have had over 40% loss of the bank. Beginning to think putting this thread on has jinxed it

Going to go back to practice mode and review the last couple of days of logs to see where/why it has gone wrong. I don't see it as losing money though, rather more like "buying data". The more I know about the losses the better it will be in the future. Bank stands at £62.62 so at least it is double the original £30 starting bank. So that's a win at least.

[andy28]

I read on the forums a while back that you don't lose money you just paid for a lesson.

Mine is a backing or laying strategy and I use a P-Value calculator to tell me if it has legs and I got 0% chance that it is all luck. Still not a given it will not fail in the future but it gives me confidence to let it run its course even when it doesn't perform as I had hoped

[Fugazi]

Thanks for sharing, I am doing a similar thing but one difference is I let the day go regardless, if I lose a lot so be it. The reason I let it run is I want to look for seasonal swings and also I want it fully automated so I won't be watching if if hits a bad patch it has to be in the plan. Also I let it run because I did in all the modelling so my thinking is if I control what it does then I am varying away from my trial.

Mine has hit a volatile period with big swings but I am thinking it is seasonal as in Autumn in Aussie you have 3yo v 4yo's+ in handicaps and also 2yo's v 3yo's so weight can be key, some may compete with more weight while others won't.

Also I love the stats part of this

Thanks for the reply, I can say this has just hit a bad patch as well, as today was abysmal, it is the first two days where I have had over 40% loss of the bank. Beginning to think putting this thread on has jinxed it

Going to go back to practice mode and review the last couple of days of logs to see where/why it has gone wrong. I don't see it as losing money though, rather more like "buying data". The more I know about the losses the better it will be in the future. Bank stands at £62.62 so at least it is double the original £30 starting bank. So that's a win at least.

My main question of your method, is why stop when you hit x profit for the day?

It is just as likely to start losing money as it is to make further profit....

I suspect all it achieves is that it reduces volume and takes longer to find out if your method works or not...

[mgreenough]

Thanks for sharing, I am doing a similar thing but one difference is I let the day go regardless, if I lose a lot so be it. The reason I let it run is I want to look for seasonal swings and also I want it fully automated so I won't be watching if if hits a bad patch it has to be in the plan. Also I let it run because I did in all the modelling so my thinking is if I control what it does then I am varying away from my trial.

Mine has hit a volatile period with big swings but I am thinking it is seasonal as in Autumn in Aussie you have 3yo v 4yo's+ in handicaps and also 2yo's v 3yo's so weight can be key, some may compete with more weight while others won't.

Also I love the stats part of this

Thanks for the reply, I can say this has just hit a bad patch as well, as today was abysmal, it is the first two days where I have had over 40% loss of the bank. Beginning to think putting this thread on has jinxed it

Going to go back to practice mode and review the last couple of days of logs to see where/why it has gone wrong. I don't see it as losing money though, rather more like "buying data". The more I know about the losses the better it will be in the future. Bank stands at £62.62 so at least it is double the original £30 starting bank. So that's a win at least.

My main question of your method, is why stop when you hit x profit for the day?

It is just as likely to start losing money as it is to make further profit....

I suspect all it achieves is that it reduces volume and takes longer to find out if your method works or not...

The reason being is that when you chart the PL for the day it can start of great for example going up to £70-£80 then start slowly declining into negative figures, that was quite a rare event though, but could be the reverse also with a few wins upto £10 then slow reverse into the negative for the rest of the day . Thats why I started using 9am as a start time as by then the greatest decline had occurred and it was a more smoother +/- charting was seen (as it is the last 5 hours of racing in AUS so late evening and liquidity was better). In nearly all days the systems hit their first positive PL which normally allowed me to reach the targets by 10:30am on the AUS races and by at least 5 races on the UK dogs. This allowed for the odd loss to be hit on each also. But the last two days seemed to hit 3-4 losers consecutively which almost allowed no recovery even if left for the whole day, and would of only decline further if more losers were hit.

But still not giving up on what I have so far though, just re-running them as is in practice mode to see if it just a bad run or whether tweaks need to be made.

[Anbell]

My main question of your method, is why stop when you hit x profit for the day?

It is just as likely to start losing money as it is to make further profit....

I suspect all it achieves is that it reduces volume and takes longer to find out if your method works or not...

Nearly every pro on this site will say the same thing.

Another way to look at it is that the upside is capped, but not the downside, so you can lose $57 in a day but can only win a little bit.

[getagrip68]

An interesting read, co-incidentally yesterday's Greyhounds was one of my worst days this year, not sure what was going on, maybe just one of those days.

It's great fun tinkering with data and trying to find something that works but it is far far from easy. Greyhounds are good due to the sheer amount of races making testing a large number of events easier.

I would say though that if you're stopping at certain profit level and playing with staking plans there is a fair chance you're fooling yourself. If a method doesn't make a profit on all selections at level stakes then I'm afraid it's a losing method.

Good luck!

[Wishy]

Great read.....many thanks for posting.

My GH results for yesterday were also appalling!

[Crazyskier]

Great read.....many thanks for posting.

My GH results for yesterday were also appalling!

Perry Barr, Monmore and Crayford all had a fair number of high-priced outsiders win yesterday. Might be worthy of watching as a 'back all at £1 strategy' would have made good return at those tracks.

Though I generally look at the LTP ranking of the winner for my strategies.There are long term trends that my last update thread has identified and I minimise or entirely avoid loss by staking only what I've won on a given strategy / track.

viewtopic.php?t=29056 for more on this and the last year's P&L results, etc...

CS