Forum

General discussion about Bf Bot Manager v3 software for Betfair, Betdaq and Matchbook betting exchanges
By Beerio
#25523
Just as a suggestion I would check the field size against what type of race it is ie: flat, hurdles or all weather as you will see there's a bif of a difference in what stats will say for each runner bracket.
By echelon
#25532
Hi DN,

One advantage with a lay strategy is that your SR will go through the roof - to 80%+/90%+ levels.

With so many winners you can use the more aggressive staking plans with good success. Some of the ones you've been trying out recently on your backing strategies, such as D'Alembert and Square Root are worth re-visiting.
By echelon
#25535
dn6789 wrote: Mon Sep 27, 2021 10:12 pm I just worry by filtering down so much that my bet numbers will be lower and profits slowish ..
I would start with a price based filter only - e.g. laying all horses priced below 3. Based on those results you could filter further on number of runners, or select Hcap races only etc.

That way you start off with a stack of bets and filter those gradually based on results. Favourites tend to win a fair bit in US races, so best to avoid those completely (which you're doing already).

You may want to include French races as they tend to perform in a similar way to UK/IE ones. South African races seem to "blow hot and cold", so best to avoid those.
By Sr8822541
#25540
echelon wrote:Another strategy you could look into is "lay the field":

https://www.informracing.com/laying-the-field/

Here you place orders in the market for favoured horses at odds-on (i.e. < 2) and let the bets get picked up when the race is in-play as horses jostle for the lead position.
Hey Echelon,

Have you had any consistent success with this method? I tried it way back and was keeping odds at $1.3 to get two or more matches. But often it was just one match, so i was losing a bit and gave up.
By echelon
#25543
Sr8822541 wrote: Tue Sep 28, 2021 5:55 am Hey Echelon,

Have you had any consistent success with this method? I tried it way back and was keeping odds at $1.3 to get two or more matches. But often it was just one match, so i was losing a bit and gave up.
Hi SR - it's not a method I've used much, though whilst thinking of horse laying strategies it seemed an obvious one to look into.

The key to making these sort of strategies work in BFBM I believe is to minimise the amount of markets loaded, to refresh data as quickly as possible and run on a fast machine. This means that we get our bets in very quickly and are more likely to be matched.

The reason I don't use systems like this on BFBM is because I have a stack of markets loaded in live (multiple football markets for match odds, over/under various goals markets etc.) and processing these particularly at weekends can take a long time and make any scalping or quick-fire strategies unusable.
By Beerio
#25553
dn6789 wrote: Wed Sep 29, 2021 10:19 am also the one thing that turns me off lay betting for serious money is this.

I lay something I lose $30 I have recovery set up lets say at 100%

next thing I lay at 4/1 for example - now I am -120 and then I lay again and I am -480 then I am risking 2k

this sort of thing never happens with my other BACK strategies - sure I have DD but only small compared to that.

anyway like I said I know little about this - so lets get more data and be more informed and go from there.

Running recovery on lay bets is very dangerous no matter what your starting 1pt value is, as there will be times that you will hit a longer string of losses and then your roll might go up in smoke for the bot.

I have been running this bot since June 1st with +90pts over the time on £1 stakes. After 634 runs I have seen some ups and down where the DD would kill the recovery and my max price is set to odds <10.

The graph shows you the DD, and the 2nd image is for the losses that it took on based on £1 lays it took. I would say just find yourself a slow and steady system that brings in pts monthly or takes a small loss. I have a general filter for these selections that I need to do daily and haven't been filtering the results just yet to improve the performance of it.

Columns in 2nd image are:
Date, Time, Stake + or - based on £1 Lays, Running PL
Attachments
Snap 1.PNG
Snap 1.PNG (21.36 KiB) Viewed 27386 times
Snap 2.PNG
Snap 2.PNG (30.75 KiB) Viewed 27386 times
By Drizzy12
#25555
I have a few lay strategies on both GH and horses.
I hade found odds above 3/3.5 seem to work best.

I also use recovery but have a safety to reset at 3 loses. It does of course happen and you stuck up the loss and go again. But there are more winning days than losing.

Also noted USA horse favs win alot so definitely steer clear (as mentioned above)
Ever tried a back fav1 there?


Interesting point about the French horses. Never used them but may consider it if it is similar to uk/ire.

Interested with such low sample of bets how it is possible to deem a strategy successful.
I personally use small stakes until 1,000 bets until I make judgements.
How do others deem their strats successful And ready for bigger stakes?
By Casenova777
#25576
Separate to your filters have you ever tried using the drifter/steamer function to pick lay bets. So set up your strategy to monitor the market 2 mins before the off and lay the runner with least support / highest drifter within your accepted price ranges... I'ts a bit hit and miss I've found but feel like theres a profitable way to approach it if you stick at it longer term..
By Sr8822541
#25589
dn6789 wrote: Sat Oct 02, 2021 9:06 pm it really looks like you need recovery on F1 to win -- I am down like $30 -- all the profit is in F2 and GH

F2 is over $200 and so is GH

will look more at this today and make some changes for tomorrow.
Hi DN,

I have also worked with laying AUS GH, esp Fav 2, but I was hitting 80 races a day, which made profit for a while and quickly went on a bad long streaks which didnt happen in my past data analysis. So always a first in the data.

I can see here you are just hitting just 35 races in 3 days which is being very selective. So are you using Lay after 2nd fav wins X times in a row etc.?
By woweee
#25620
Hi dn6789

Could you have a look at my strategies.

Strategy 0-3
Aus Dogs
Back 3rd Fav (Must be in trap 1 or2)
1st Fav has to be certain odds.
Certain type of races
3rd Fav odds 4-9
No NZL

Strategy 3-6
Aus Dogs
Back 2nd Fav (Must be in trap 1 or2)
1st Fav has to be certain odds.
Certain type of races
2nd Fav odds 4-9
No NZL

Strategy 7-9
UK dogs
Back 3rd Fav (Must be in trap 1 or2)
1st Fav has to be certain odds.
Certain type of races
3rd Fav odds 4-9.

Strategy 10-12
UK dogs
Back 2nd Fav (Must be in trap 1 or2)
1st Fav has to be certain odds.
Certain type of races
2nd Fav odds 4-9.



I have tried adjusting race length. Made no difference.
Having 3rd Fav in trap 1 or 2 made no difference.
Amount of runners made no difference.

It might be best having 1st fav in the outside traps.
May try this next.

.
Attachments
(5.52 KiB) Downloaded 870 times
Last edited by woweee on Sun Oct 10, 2021 3:03 am, edited 1 time in total.
By woweee
#25625
dn6789 wrote: Sun Oct 10, 2021 1:06 am I downloaded now and will take a look

just in general from what you have written above to what I have there are a few pretty major differences - but I will see that in the strategies let me get back to you shortly
Thanks for the comments. I will take another look.
I have rewritten the descriptions.
By echelon
#25629
Hi DN,

Great discussion and advice in these threads as ever - many thanks for your continued posts and advice :)

I think your last post has helped me to understand the "granularity" you go down to in your analysis.

I've attached a view from a spreadsheet of September data which I downloaded from:

https://promo.betfair.com/betfairsp/prices

I'm wondering if the analysis I've done here is about at the right level.

Filtering is as follows:

8 runner races
Distance > 400m
Fave 1
Traps 1,2,3,4,6,7
Odds > 2

This filtering would give 20 bets per day with an average profit to £1 stakes of approx. £2.88 / Day.

The SR is 39%

There are a few things that bug me:

1) I don't have a specific tight odds range - though > 2 seems to work better than any others I've looked at. I find it tricky in Excel to easily analyse different odds ranges without using trial and error - do you have any shortcuts for honing down those odds ranges?

2) My trap selection gives the best profitability, though the selection doesn't make too much sense in missing out trap 5 and trap 8. I'm wondering if I'm just "curve fitting" here and will find that next month the traps which are most profitable are completely different. I know for sure that trap bias is all over the place with UK dogs, and believe that traps are more consistent with AU dogs, though to what degree I'm not sure.

3) My SR is a just under the 40% threshold for a decent low favourite strategy - these bets are already filtered a fair bit giving only 20 bets per day, it seems as though it may be tricky to push the strategy through the 40% barrier. Also, the 40% barrier can only apply to the lower favourites - do you use that level on just F1 and F2? The higher favourites don't need such a high SR because when they win, they win at big odds.

4) Do we need to be really specific on distance? I would have thought we either have long distance or sprints and each will have very specific characteristics.
Attachments
Screenshot 2021-10-10 141454.png
Screenshot 2021-10-10 141454.png (48.44 KiB) Viewed 24825 times
By echelon
#25633
Hi DN.

Many thanks for your detailed response containing tons of valuable insights for those of us who are tuned in to your methodology to a degree.

Thanks for the tip on Excel statistical analysis, it's something I'll check out in more depth.

I've started to adopt some of your ideas on money management by looking at linking strategies - though of course, for us to do this successfully, we first need to create individual strategies (and to do that we need to understand how to "filter the crap" out our strategies lol). Regarding setting targets for individual strategies, I had two running today on UK dogs and could have comfortably taken a profit of £20 for each of them and stopped at that, though leaving them to run gave me £5 less. Interestingly had I stopped the strategies at 8pm (1-2 hours before the end of UK racing) then I'd have made an extra £20. Is market timing something you consider?

Favourites in UK horse races tend to win more in earlier races - is this a bias which is seen in AU horse races? I see that the AU and NZ races helpfully have R1, R2 etc. in the market description, so if we wanted to bet on favourites in early races it's straightforward to filter.

Also, you may have noticed that I have in my spreadsheet a view of all the favourites odds. I did this because you mentioned some time earlier that we need to look at the market structure. I thought that formatting the data like that would have made particular market structures jump out - such as F1 and F2 at very close odds, would suggest a clash of the titans and the chance of any other dog getting a look in is minimal. Again I found I was overwhelmed by the data and couldn't see the wood for the trees (if it was there in the first place!).

I totally agree with you on staking being important - from what I've seen, different staking plans are suited to the distribution of winners. One staking plan might do very well with results such as WWWWLWWLLWW, though a completely different one would be more suited to LLLLLWLLLLLWL (where the winning odds are large). What we need to avoid is applying staking to a poor strategy so that it looks good - at some point that will bite us (speaking from experience lol).

Your observations on long distance GH I remember you mentioning some time ago and it surprised me, because I was thinking of UK horse races where favourites typically win flat short distance races whereas the longer jumps introduce all sorts of uncertainty which allows outsiders to win and made the incorrect assumption that distance with dogs and horses would show similar biases.

Ah - the boxes (traps) issue...I've been caught out in the past more times than I would have liked, laying traps which were showing a low % of winners over the year to date. The issue with this is that, typically for a while I would have loads of profits, then suddenly one day everything changed and I had loads of losses which kept mounting up until I had to ditch the strategy. What appeared to be happening here was a simple "regression to mean". I looked back through previous years' trap performance and realised that the win % of each was very similar - this meant that when a particular trap was underperforming, at some point it would need to catch up with the others so that they all end up with the same win % at the end of the year. Having said that, there do seem to be some more consistent trap biases in AU 8 dog races than we're used to seeing in UK 6 dog races - I'm thinking F1 and F2 normally doing well in Traps 1 and 2.

Another filter I've considered using is the venue - though for both UK and AU dogs I've not found anything consistent. UK horses I need to look at next because we even have a saying "horses for courses" which must suggest some sort of bias! I've noticed that AU dogs venues often have unique tracks, in particular the distances - so maybe when you're looking at tight ranges of distance, you're actually looking at similar types of races at a bunch of venues?

Anyway - thanks again for all your help :) Time for more research for me...perhaps tomorrow though as it's getting late now.
By echelon
#25649
Hi DN,

By "same column of data", I assume that you're talking about odds in an Excel spreadsheet?

It would make sense to use high odds for backing (to maximise profits) and low odds for laying (to minimise liability).

What really blows my mind about the strategies here is the level of filtering. The most showing in a particular strategy is only 9!

I find that by looking in the data I can filter down to this level by looking at certain odds, grades, distances, traps etc. of recent bets, However as soon as I test this filtering out, the market changes and the filters required are different. This is a real stumbling block for me and I think is the key thing which is preventing me from seeing any sort of consistency in my strategies.
By woweee
#25650
Hi Echelon

I'm in the same boat as you. I can play around with different filters using the BFBotmanger Bets tab. I can either achieve a high profit and low SR. Or the opposite. Soon as I run it live using small stakes the market changes and I make a loss.
By Beerio
#25697
I 100% agree with DN as you do need to continue to work on your filtering. What started with 7 bots on the 29th of Sep has turned to 4 bots by Oct 18th...

All 4 bots are separate in the filtering they do, however they do pick up the same horse from time to time and it's either a double winner or a double loser. The graph below is based on £10 stakes and showing +£257.70 after 345 runs.

I am planning on making more filters that will work along side the ones I have but happy to pick up a few pts per day and just keep this ticking over. If you have a dream of making this work don't give up people, just keep on filtering.
Attachments
Snap 1.PNG
Snap 1.PNG (13.53 KiB) Viewed 25011 times
By echelon
#25741
Hi DN,

Thanks for that interesting insight into how your spreadsheets are put together.

I would guess that these tables are for AU horse races with 5 to 10 runners and it looks like the favourite number is matched against the runner number?

I don't follow AU horses, so I'm not sure what the runner number actually means as I know horses aren't like greyhounds where the runner is always the trap (box) number.

I assume that the data you're using is coming from the BFBM results screen, are you using any other data in addition to make your selections?

I did some filtering in my results tab picking up 5 field races and looked at favourites 4 and 5 (attached).

Your table suggests that only F5 and runner 5 is valid for the amount you've plugged in that you would like to earn from a 5 runner race.

From my results this doesn't make much sense to me - can you explain a bit more how this all hangs together?
Attachments
Horses_5.png
Horses_5.png (33.71 KiB) Viewed 24898 times
By echelon
#25752
Thanks for the general hints DN.

One thing puzzles me with analysing horse racing data - take AU races as an example.

The horses are numbered starting 1 based on the heaviest weight being carried, the next heaviest at 2 and so on.

These I believe show as runner numbers in the BFBM Results tab and are also called saddlecloth numbers because that's where they are seen on the horse.

If we look at the csv data from the Betfair site, then we see a number as part of the selection name - which again appears to be the runner number.

Given that stall draw (I believe) is an important part of our data analysis, where can we find historical data which contains the stall draw?
By echelon
#25757
Hi DN,

Thanks for all those links and tips - very handy!

I've not used Data Miner before, though it looks really good. Previously to scrape data from sites required a lot of messing around programming and then when you had a program nicely set up, you'd find that the websites you were scraping were updated and the scraping code needed to be written again :roll:

I guess the sort of things that we could do with this additional data is for example to look at form. Average race speed of dogs is supposedly a good way to weed out potential winners - if we scrape those speeds from previous form, then we can compare speeds with odds/traps and whether dogs win or lose from the data we've already compiled in Excel.

From that analysis, we may end up with patterns of odds/traps which are associated with fast winning dogs. So, in effect we're using what could be called the metadata of speed which is the only thing that BFBM understands - because there's no connection to form for the dogs.
By sandres53
#25786
Hello.
I have been following your conversations for 4 months. I appreciate the feedback from DN6789.
I have been working on betting systems for 10 years with IA (AUS UK NZ FR, GH H) and football. I use BF Bot Manageur. It's a good bot. I worked a lot in vba Excel and automation.I can help in excel.
I would publish my results.

sandres53
The Fisherman
By sandres53
#25791
My Results 8/11/21 GH AU/NZ
Filters : Prices/FAV/Runner/SEQ with BFBM
*********************************************
IA : No
Speed ratings : No
Tissue prices :No


T=+78 $ SR:57% 8/14 STOP

Sandres53
The Fisherman
Attachments
Results GH 081121.JPG
Results GH 081121.JPG (142.27 KiB) Viewed 21835 times
By woweee
#25792
Despite reading all Dn's posts I've totally failed at backing.
I'm now Laying the Aus/Nz dogs. The attached strategies are doing quite well.
If the fav is in Trap 1 it stands about 60% chance of winning.
I've also removed all races over 500 meters.

I usually win 10 to 25 pts per day. One losing day in the last 7 of 18 pts.
This almost guarantees tomorrow will be a losing day.
Only had two consecutive losses once.
No recovery.

I think these strategies can be improved by selecting certain tracks.
At the moment I don't have enough data.
Attachments
(3.73 KiB) Downloaded 1068 times
Last edited by woweee on Fri Nov 12, 2021 12:17 am, edited 3 times in total.