The BTC Horse Racing Thread
-
@frank-jones might well be. Bit weird if I’m honest. Although no trobs either which I get from elsewhere.
-
@john-folan Is that because the races aren't appearing on the software for today? I'm only getting Ffos Las and Catterick, and only 4 races showing in total - 1 Catterick & 3 Ffos Las!
-
Absolutely nothing today!
-
Sadly back in the land of the living. Will try and catch up properly over the next few days. In the meantime:
Todays Selections
Flat Pace: Back any between 2.0-25 near the off.
Slow Horses: Dob any over 6.0
-
-
@jonathan-jones updating the filter wouldn’t work as it would still depend on the time of day you looked at it. I’ve mentioned to Adam on Asana about taking going out of the qualifiers criteria as I think people are just going to have to check the going near race time as I do.
-
I can't find the previous discussion on here about the going now but yes, we use the official going which can change.
The going value you see is the final going value at the time of the race. If the going changes throughout the life of the market then you might have horses that didn't qualify earlier in the day but did by the time the race went off.
We are planning to store the whole going history instead just one static value. Same goes for any value that has the potential to change during the life of the market. Maybe what we could then do is send you an alert if your strategy uses the going and the going has changed since the last time your qualifiers were run.
The reason for logging these things in the support form is so that we can identify these needs, not lose track of them and figure out solutions for them. I have added the above to our roadmap.
-
@john-folan
I guessed that was the problem. To be honest i cant really see any way round it other than something like you have done or roll my own automation. Not sure its going to be worth logging it in asana as i doubt there is much Adam can do. Maybe update the filter criteria more frequently for the qualifiers but still needs to be discussed here as its a pretty big limitation that people need to be aware of. -
@jonathan-jones you are probably best off raising this in asana on the software. Going probably needs to be removed from qualifiers as a variable as it can and will change on the day. Until then I run a separate qualifier version of my flat pace with going removed. I load them all in and I’ve then got it in Bfbot to exclude them if the going is one of the excluded ones. However, bfbot itself only uses predicted going as it gets that info from the racing post and does not update as the day goes on(they are looking to add this in a later update when I chatted with them about it) so I just check the going of the course on Timeform after the first race and go by that. Then just delete any where the course has changed. Writing this message actually has taken longer than it does to do that everyday lol. I never bothered raising it as this just seemed like a common sense workaround to deal with a changing variable. Since then my results match the software for every selection.
-
Question on the upcoming qualifiers.
When is this list compiled and is it refreshed throughout the day based on the filters?Reason i am asking is the backtest is totally different to the selections downloaded. There is an entire course full of races the backtest included which were not advised. My suspicion is that the going might of changed or been notified late. In yesterdays example there are also a few races the backtest seems to of lost which were included in the selections. These were in the middle of the card at a course which did include others so going not a factor in that one and i dont have a clue why they got missed.
#Edit#
pretty sure its the going that caused this. I have another filter that filters on same going values and that also has exactly the same issue yesterday. A rare event or something we can improve?
-
@jonathan-jones said in The BTC Horse Racing Thread:
@john-folan said in The BTC Horse Racing Thread:
@jonathan-jones I don’t think the forecast price changes but I’m guessing rank will based on non runners?
That is possible and may be what happened here. There was a non runner and that was one of the ranked selections so the backtester traded an alternative horse.
Thats a problem then if any of the 'rank' things are not fixed. You cant use something in a filter if that thing can change after you would of placed your bet. It makes the entire thing unreliable. If you cant look at the backtest and say thats what would of happened if i traded that filter then thats a problem.
We've discussed this. If the price can change, then the price rank could also change at different times. For these "rank" stats, we're going to capture the full history of rank changes so that you can see them at different times. At the moment, you're probably seeing the final forecast price rank for that race after any changes were made.
It would definitely be incorrect to NOT update these ranks as they change. But what we don't currently do (in v1.0 of this software, which already does about 1000 times what any other statistical software does) is store a complete history of all the rank changes. That is what we are working on right now.
-
@chris-osborne said in The BTC Horse Racing Thread:
dealing with over people’s money is a whole different ball game.
Especially when there's a point of failure in either the Exchange going down or your own interntet connection going down briefly.
Had a scare just before a 6f race yesterday where I was trading and just about to get out before the off and either betfair was unresponsive for about 45 secs or my internet was playing up.
Luckily got back in for end of race a was well up in green......but could have been painful! -
@jonathan-jones not sure what the bet Angel filters are tbh. Probably changed a lot since I last used it. They might have it on there.
-
@john-folan
Damn. That compromise with the different automation things has reared its head again. This strategy uses an open window for bet placement. About 2/3ds of the bets are placed inplay. I had issues with fairbot using the old api that i dont have now i swapped it to use the streaming api. If i went to bfbotmanager i would get more accurate representation regarding the selections but would likely get a lot more missed entries.I will keep comparing the day to day actual vs backtest for the time being (though its expensive software if its not actually working to make you money so dont want to do that too long). In theory i guess the races it goes in my favour should match the ones it doesnt but the point of me using this is that i dont have the 'might', 'should', 'i guess' etc.
-
@jonathan-jones I can see that. I had a lot of this with Proform. In a sense a lot of the ranks won’t be known until just before or just after the off. I also had an issue with the number of qualifiers rule. Same thing as you really. Except I was doing an inrunning thing and it didn’t know the max number of qualifiers in a race until after it had finished would be 1. Some things unfortunately will happen like that. It’s a bit like predicted going and actual going sometimes being different. I just keep an eye on that during the day. I might suggest Adam adds a separate filter for each of those to solve the problem there.
-
@john-folan
That sort of thing would be ideal. Might need to look at that. In yesterdays races it worked to my benefit but the size of the difference (80% near enough) has made me worry. -
@jonathan-jones that’s an Adam question I think. Although if you are running Bfbotmanager you can filter by forecast rank in the filter by detailed information at the time of the bet being placed as per the screenshots below. So in theory you wouldn’t need to load your selections in as your filters in Bfbotmanager should pick the same ones as the software does.(with the odd exception caused by very late non runners)
-
@john-folan said in The BTC Horse Racing Thread:
@jonathan-jones I don’t think the forecast price changes but I’m guessing rank will based on non runners?
That is possible and may be what happened here. There was a non runner and that was one of the ranked selections so the backtester traded an alternative horse.
Thats a problem then if any of the 'rank' things are not fixed. You cant use something in a filter if that thing can change after you would of placed your bet. It makes the entire thing unreliable. If you cant look at the backtest and say thats what would of happened if i traded that filter then thats a problem.
As a possible mitigation when does the software list the runners on its filters? does it update throughout the day or just get the list the day before?
