Horse Racing Strategy Clinic
-
Struggling to create your favourite strategy in the software?
Worried your strategy is too good to be true, based on the backtesting results?
Not sure whether your strategy rules are meaningful or if you might be backfitting / p-hacking them?
Looking for ideas on how to make your strategy better?
This is the place for you!
Click the export button in the Horse Selection Rules and upload the resulting JSON file here and we can all help out!
-
@Adam said in Horse Racing Strategy Clinic:
@John-Folan There have been no updates to the software for a while. I can look into it if you log a support ticket with some more detail; it sounds more like a tech support request than a strategy clinic thing tbh.
Weirdly it’s righted itself. Finally back trading this weekend.
-
@John-Folan There have been no updates to the software for a while. I can look into it if you log a support ticket with some more detail; it sounds more like a tech support request than a strategy clinic thing tbh.
-
@Adam Not sure what button you have pressed , but my Dob and Trob strategies in my bunker are now showing overall losses. I've not touched them before I went on my nice extended holiday.
-
@Tony-Hastie It's saying that stalls 4 and 6 are the ones you would have made money on in this strategy, but there's not really enough of a sample size to base any decisions off here, with only about 20 trades on each stall.
This also doesn't mean that stalls 4 and 6 are the ones with the draw bias. The lower stalls are definitely the ones with a bias at Chester over 5F (stalls 1-4 all have a positive draw bias). But people know this already, so that's why if you're purely basing your betting decisions on this fact won't make you money, because the prices already reflect the bias. It needs to be combined with other factors.
You can start your strategy off by filtering on the positive draw bias at Chester over 5F, assume that everyone already knows about this bias, and then start thinking "what can I find in the software that the average punter isn't going to know about, or can't easily figure out?", so you can find some value / an edge that others don't have.
For example, some people are probably not going to realise that stall 4 has such a positive bias, which might be why it shows a profit before adding much else in. You could also look at different goings, or try adding pace into the mix as some courses also have a pace bias.
-
@Tony-Hastie if you are really stuck with it you could always just choose the stalls that work best for the strategy and forward test them
-
@Tony-Hastie taken from what @Adam said on the horse racing thread yesterday.
You're usually going to be backing a stall with a draw bias greater than 1 and laying a stall with a draw bias less than 1.
What this means is every 1.01+ is saying it is a potentially value back
Everything 0.99 or less is a value lay -
@Adam Thanks for your draw bias explanation it was very helpful. But having been tinkering all night Im still having trouble joining the dots. Attached is my tinker effort, using a well known draw bias track and distance. Chester, 5f. Am I right in understanding that this strategy is telling me that stalls 4 and 6 are the ideal stalls to run from? And the ideal draw bias is 1.1-1.59? Or am I way off? Let me down gently...Draw bias tinker.json
-
@Andrew-Gray The main issue is that you're using the "Price Range" bet type instead of "Single Price".
This is like saying you're sitting there waiting for the price to hit that exact value, at which point you will somehow time it perfectly and manage to get the bet matched at that price. It also means the horse has to trade at EXACTLY that price, because there's no range, so if it traded at 42 at one snapshot and then the next price traded was 46, it will be missed.
The "Single Price" bet type does exactly what you're expecting; it's like placing a bet at a specific price and letting it sit there waiting to get matched. If the price goes past your specified price, your bet will still be matched at the price you requested.
The other thing is you're using the official start time, but the BSP won't be known until the actual start time. Probably not a huge issue in this case as it's only going to be a couple of minutes off at most and it's unlikely a horse would trade at 40+ and then have a BSP of 10-20.
-
Here's one for you Adam.
I have selected BSP between 10-20 and stick in backs at a higher odds range e.g. 40, or 50 or whatever really, and it always seems to come back with profit.
Sounds way too good to be true so compared to the BSP file downloads I have, and wrote a small python script to run a similar strategy, and it always comes out with a loss no matter what the odds above BSP I select, so the complete opposite of what the BTC strategy is saying.
Any ideas on if I've set up the parameters wrong or am expecting it to behave differently than the simulation output is?
I would expect it to be simulating for e.g. if back price I select is 40, for a back at 40 to be put into the IP market, and only if price goes above that it would be "matched" as a selection, and then the P&L for each matched bet to be worked out. And any where price does not go above that and horse wins it would obviously not count as a bet.
Cheers
-
@Andy-Donnelly It seems to be the other bit actually, the "never won at a course with specific characteristics" bit.
I'm looking into it, I don't think it's doing what I thought it should.
-
@Adam I dont unfortunately, just more an idea I had on an angle where horses might be overrated by the markets. I’ll try and find one though.
Thanks for looking. Does it work to have a filter on horses having not won at a particular course charateristic? So basically remove the won last time out part.
-
@Andy-Donnelly There is a way of doing this that should work, but when I set it up in the software, it never gives me any results, which seems odd.
Don't suppose you have any examples of horses which would have qualified for this strategy on any date do you? That way I can see why the software's not picking it up and figure out what's going on.
-
@Adam is there a way for me to create a strategy on the below criteria using existing filters:
Lay horses that won their last race, but have never won on a track with the characteristics that there next race has.
So as an example, say horse x won their last race, which happened to be on an undulating course. Their next race is on a sharp course and they have never won on a sharp course before.
-
There's nothing inherently wrong with the way you've set it up, but your main issue will probably be getting the exit bets matched in running.
You're betting at BSP, which means you won't be able to place your exit bet until the race has started and will have to deal with the in-play delay.
You also have quite a few very high odds bets which may also struggle to get their exits matched.
You could track it, and if you find an unacceptable percentage of your actual bets are not getting caught, you could try placing the bets just before the start of the race so you can place your exits at the same time, and maybe try restricting the odds range a bit.
-
Good Morning
I'm trying to create a back to lay strategy around front running horses.
What needs to happen is the horse that I back at the start should be near the front, allowing the odds to drop so I can trade out for a small profit (would be done using bots) or exiting for a stop loss if the price goes the other way
I've put this very simple strategy together - but the results seem a bit too good to be true - what have I missed please?!
Thank you for your help