Horse Racing Strategies Megathread
-
@emily-knowles There was a good explanation on how to have mutiple automations on one race on the BA automation thread
https://forum.betfairtradingcommunity.com/topic/4020/bet-angel-automation-thread/122 -
@jonathan-jones I totally agree, I often have multiple actions to take on a horse, backs and or lays at different odds and I do them all to get as true a result as possible for each strategy.
Which software do you use to automate? I use Bet Angel but I just can't figure out how to have multiple automations on each race or have more than 2 runners per automation rule. I'm open to trying different software if recommended.
I know it's not part of your software but I wondered if @Martin or @Adam could do a video or even get the automation software person to do one for BTC?
-
@tom-2 Thanks Tom
-
@caroline said in Horse Racing Strategies Megathread:
Evening All,
Is there anywhere that shows me what the various horse racing filters mean and roughly how they are derived, in particular things like pace rating and how does that differ from pace letter
Many thanksPace letter is assigned after the race, either L (Led), P (Prominent) or H (Held Up).
Pace rating is compiled from the horse's previous 3 races. A points value is assigned based on the pace letters from those races. L = 4, P = 2, H = 0.
So if a horse's last 3 pace tags are LPP, its pace rating for the upcoming race will be 8.
-
@john861 said in Horse Racing Strategies Megathread:
Hello, I'm a big newbie to Horse racing in general so this could be a stupid question but how have the strategy filters had two picks from the same race and the same horse both backing and laying, also two selections in one strategy in the same race? An example of this is the Haydock at 16:10 with 'no cruise yet' being a selection for both the pace lays and Back the course and distance winner. How can this be the case? how would people advise you to proceed if you're trying to follow?
Also, the Haydock 16:10 also has a fortified bay as a selection, however unlikely to reach the desired odds range but would you just lay both horses in the same race if it did?John, they are different strategies, so are looking for different things in the race. It is upto you to do your own due diligence if you follow both strategies or leave the race alone. You pays your money you takes your chance!
-
@caroline Hi Caroline.
I don't know all the answers as I'm still learning myself, but when you go into the horse racing software, on the right, there is an icon of a camera and if you click it, it'll take you here,...It's got lots of training videos about it.
-
Evening All,
Is there anywhere that shows me what the various horse racing filters mean and roughly how they are derived, in particular things like pace rating and how does that differ from pace letter
Many thanks -
@john861
I guess firstly the selections the software is giving you for each race are horses that match the main filter criteria. Its perfectly possible that the same horse might match both back and lay strategies in terms of the filters. Its also possible that you will get several horses selected per race, per strategy. Not all will end up being bet on. It depends if they hit the odds criteria in the time window defined for those strategies.
Regarding the what to do question, well that is an interesting one. I am currently running several strategies as a portfolio (backs and lays) but doing it 100% automated. I have downloaded all the individual trades for each of the strategies, loaded them into an Access Database then analysed further in excel. There are many, many instances where the same horse is both backed and layed in the historical data. What you want to do about it depends on you. If you leave it then depending on the odds taken on each (usually slightly different as the odds ranges and time windows are different on the different strategies) you will be left with a small residual on either side. If you decide to take one or other then you are basically gambling as far as i see it. You wont see the same results as the backtest. The last option is to simple skip the horse entirely.
The way i am doing it is taking all selections that are given for all the strategies. The automation then bets on all that are within the odds ranges and time windows. If one is a back and one is a lay on the same horse it will place both bets. Thats the closest way of doing to what the backtest would do. Its happened a couple of times so far. Its not caused any problems and my live day totals match pretty closely with the sum of the backtest totals when run the next day. I believe the betfair commission is payed on the market rather than the indivdual bet so you are not losing anything doing it this way.
-
Hello, I'm a big newbie to Horse racing in general so this could be a stupid question but how have the strategy filters had two picks from the same race and the same horse both backing and laying, also two selections in one strategy in the same race? An example of this is the Haydock at 16:10 with 'no cruise yet' being a selection for both the pace lays and Back the course and distance winner. How can this be the case? how would people advise you to proceed if you're trying to follow?
Also, the Haydock 16:10 also has a fortified bay as a selection, however unlikely to reach the desired odds range but would you just lay both horses in the same race if it did? -
@jonathan-jones my biggest concern is the number of trades. Usually, I don't mind multiple horses in one race as it reduces the overall liability. That's not a problem with a bot, but makes me think is it a quantity over quality strategy?
I'm going to sim it myself but wanted to experiment with summer vs winter trading and handicap vs non-handicap. I might look to remove October as well and focus on May - September
-
Interesting one that. Basically all you are doing is laying everything thats in a particular odds range. You wouldnt think it would be so profitable. Makes sense it only works on handicaps though. Removing the fav conditions seem to mke it even more profitable. All my flat only strats also remove October so i have done same for this one. Its one to keep an eye on. Will run it on Sim in May to see how it goes.
-
That option should probably be a default to official start time. I’ll raise it with Adam later. It’s a bit of a pointless stat as you can never ever get that unless the race goes off bang on time.(which they rarely do).
It will stop that eureka moment where you think you have a killer strategy and then the deflation when you realise you never changed it to official start time.
-
@jonathan-jones
Some great points, not least I didn't realise that I had it on actual start time so I'll amend that!
I really like your idea of backing pace horses on flat courses, I didn't think you could do things like that.
Can I put a request in to @Martin for a YouTube video discussing ideas for strategies which combine two or more logical factors. The obvious one is course and distance win for example. I come from a football background so I don't necessarily know what would marry up well or logically for the horses.
-
I've been wanting to dive in and separate the summer vs winter racing for a while. I started having a go here with turf only.
My thought process was this:
- Lay between 3-10 to get a rough idea of the results. Later on I reviewed the entry price and adjusted it to 6-10.
- Remove Fav and 2nd Fav (I know a few people get bothered by this but with Fairbot it's really easy to do. I also wanted to try and filter races that have 1 or 2 decent favourites)
- From there I looked at handicap vs non-handicap
- I reduced the non-runners to fewer than 6 mostly because I didn't want out-of-the-ordinary races. This filter really doesn't make much of a difference overall.
- I then removed the first and last months (March, April and November) because I remember someone saying it's like the first month of a new football season and it takes a while to adjust.
- Finally I went through the courses and removed the ones which are not suited such as Epsom, etc.
I've left it to fixed stake rather than a liability, however, both are profitable. Doesn't seem too bad for some summer trading.
You can remove filters or add more in such as DSLR or whatever suits your automation. I've made this filter with Fairbot in mind.
Let me know your thoughts and any improvements.
-
Thats exactly why its 10 seconds for that strategy. I am not convinced by it either to be honest. It sounds pretty reasonable but its backtest stats nowhere near reflect the stats running live. You could say its just variance but if so its variance miles beyond anything seen over the entire backtest period and the timing of the downturn is incredibly suspicious being almost instantly it runs on out of trained sample data. Back the fav v2 in the presets looks to suffer exactly the same issue. There is a back maiden favourites in this thread that looks good since its creation though so must be something in it.
-
@tony-hastie I find it interesting how much the backtesting results change just by tweaking the time of bet - if I change 10s before official start to 10m or 1h before and the backtesting results are way higher. always wondered why that strategy is set to place the best 10s before official start . I'm a bit uneasy backing favourites anyway, but happy to be proved wrong. I always get concerned watching the build-up to the race and seeing the favourite flip-flopping between 2 or 3 horses with late money coming in, then I'm thinking I may as well be doing eeny-meeny-miny-mo here
-
@emily-knowles
Looks good, though a few things would put me off trading it (and this is 100% personal on my part so no critisism of the strategy at all).
First is it uses 'Actual' time so its impossible to trade according to the rules. You would never get the live results to match the backtest, the backtest knows when 1 min before the actual race start is, you wont. If you change it to 'Official' you will see a pretty big drop off in profitabilty (its still very good but much less). A Lot of people dont mind those things but i like to know i can at least match the rules of the strategy in real life (its one less uncontrolable variable to worry about).
Second its just too high variance for me. Long periods of loss or no growth.
Lastly i like to have simple fundamental resons for the bets. Example: Back the pace horse on flat on courses that flatter fast horses. I cant really see any fundamental reasons why horses that won between 2 and 5 times are actually more likely to win than those that won between 3 and 9 times for example. Overly specific criteria like that make me think its been fitted to the data so i would want to see that live for a while before i bet money on it. And, link that to the high variance anyway and you would need to run it for a very, long time (like a year) to be sure if any poor results you might see were result of back fitting or just expected variance.
One way i mitigate the last issue is to develop the strategy on a subset of the data (so just 2022 for example). Then when i am happy i remove that date filter and see if its performance prior and since is close to the backtested performance. If it is then i am reasonably sure its not back fitted and i will put cash on the line straight away. If it bombs then i just bin it.
-
@emily-knowles nothing wrong with it that I can see, looks good but definitely one to forward test for a while at least due to the course filtering. Concern would be not many selections in past couple of years. Keep us updated!
-
Not sure if this the best place to post this so please delete if appropriate.
Any action no matter how small or in which direction is better than no action at all. So after watching what feels like hundreds of hours of videos I decided to give it a whirl myself. I sat down at my computer and refused to leave until I had something that resembles a horse racing strategy.
My style is to go for high stats, low trades and build lots of strategies rather than have a few high volume/low return ones, does any one else approach it like that? -
This is my first go. I get the feeling it's either really good or I've missed something super important/obvious. I have no idea if I've even got the right end of the stick! (One point to note, I have a high variance tolerance from my poker days)
All constructive feedback welcome and if you like it please feel free to take it and tweak the filters.
