The BTC Horse Racing Thread
-
Rubbish day yesterday.
Big differences in backtest to realilty. Backtest says 1.79 profit, reality was 14.6 point loss. One obvious and explianable miss of the back the fav. 2 horses swapped place after the official start but before the off resulting in me backing the loser but the backtester backing the winner. Sod all i can do about that one and i have had a fair few of those work the other way round.Pace Lays was the other. A bet taken on the limits in reality but missed in the backtest resulting in a massive 10 point difference. Put in a support request on that one because obviously it should of traded (wondering if its down to traded volume @ that price), just no idea why its missing from the backtest data (though last week i had a similar issue on the same strategy and an hour later it was showing on the backtest).
-
@nick-allan
At a guess i would say its probably how your automation is set up then. I use Fairbot so cant comment on BetAngel at all. Fairbot automation is very simple. Its easy to craft rules that match the strategies and i have pretty much no missed entries now its set up to use the streaming API. Only thing with it is that it works on a strategy file per race and you can only have one. So, if you want to run multiple strategies on multiple horses in the same race it would be a lot of work to configure those every day (unless you can code and hack a workaround). -
@jonathan-jones Sorry I should have clarified that the short odds lay isn't a favourite strategy you're absolutely right the only filter that I have on that is to take out class one and two horses which when I looked at backtesting were the ones that were winning and therefore unsuccessful trades. But everything else is included and my BOT looks at the number of runners an only executes when there's more than four or less than 15, so the number of runners criteria is covered. Again I don't know if runners are categorised at BSP because obviously you could have 15 runners when the bot fires (5 secs be the race scheduled start) and then two runners might refuse and the software may recall that as 13 runners and therefore a valid trade, if you see what I mean.
Yeah I've realised that having the favourite condition in the lay the beaten fav strat in the software wasn't working because of course favourites can change dramatically in the final five minutes before a race, so I took that out of the filter and my bot basically won't fire if the selection is first or second favourites as per the rules. What I am seeing is selections in the results where the BSP odds are way higher then maximum odds of 15, but of course some horse races go off 10 minutes late and of course in that time a horse can massively drift so its not not a great metric. Id prefer a price at schedule start but again people have raised this before.
The back to favourite is probably the closest in reconciliation from live trades and software results and yeah I agree that once the software makes that change I think that's going to be a lot better. in fact this might be the one and only strategy I take forward and forget the other two if the inconsistencies continue.
I think the hardest thing with betangel is how you actually checked it with the inconsistencies it has a log feature bar obviously if the bot hasn't fired it doesn't log anything anyway so you're none the wiser. I believe that the BFBM is better at discovering bugs but I always had the issue of running multiple strategies across the race being too problematic. I don't know if that's been ironed out now and I can see that there is a BFBM button on the software so this might be a better product to run on the horse racing software but I've still got six months of a subscription left to go on it.
-
@nick-allan
If you are using the classic (preset) versions of these then there are things you need to change. Youn may already of done these but just in case you havent thought about it i will repeat them here:Short odds lays has number of runners condition. The Short odds lays itself DOESNT specify the favourite though i have often seen it implied in selections that it does. This often causes differences as horses drop out. Your best bet with this one is to create a 'filter only' version with that removed to provide your daily selections and handle the number of runners in your automation.
Lay the beaten fav also has the favourite condition and looks for selections from 11 mins (think its 11) out. Thats never going to give consistent matches between backtest and live. Bin the condition. To be honest its performed fine without it.
Back the favourite you are stuck but as its only looking for selections in the last 10 seconds i cant see it being too bad. I notice the odd one but nothing that makes me worry. Plus with the new release imminent giving odds on the 2nd, 3rd favourite etc i think this is one that will end up getting improved and made more consistent.
Back C&D winner. Bin the favourite and bin the number of selections condition. This particular strategy is impossible to execute according to its current rules. I am pretty guilty of coming up with ones like that that lose me money before i realize exactly what i have done. Without those conditions its still very, very profitable but re-evaluating the courses after making the change should smooth out the now bumpy equity curve to something more acceptable.
Beyond that double check your automation. I thought mine was perfect and still found a bug this week.
-
@jonathan-jones Yeah the three strategies that I'm using is the short odds lays, lay the beaten favourite, and back the favourite, and of course all three of those strategies utilise the favourite and we all know the issues that we've had in the past so I won't bring those up again. Yeah I'm checking the rules and making sure that I've got everything set up correctly but I'm still seeing these discrepancies. I don't know if it's a Bet Angel issue…………………. I've spent afternoons where I've literally checked every single trade and it seems to be firing correctly.
The only thing I can do is continue to forward test an ignore the back testing but of course that a big part of the software success, so you can see my frustration. 10 mins out the odds can fly all over the place, fav can become 5th favs so I not sure that’s a good metric, id prefer something at the race start (and of course a lot don’t go off on time either making it even harder!)Ive not looked at C&D but basically I cant make these strats reconcile so little point in trying to add more, it will make me tear my hair out more
-
Depends what the strategy is. There are some obvious issues like that which we have discussed over and over. In terms of the Back the C&D winner there is also the fact it has the number of selections criteria (which i think Adam is going to look at in terms of giving us more options). Its looking for bets from 10 mins out but already knows how many selections there will be. Obviously we cant know that. At the moment i am calling them 'impossible strategies'. I think it is important to remember the software is executing them exaclty and correctly, its just you wont be able to.
I have seen differences with simple ones over the past few days BUT i have been messing around over the past few days too so that could well be down to me. I would say double check the strategies you are using to make sure all the rules aew actually capable of being followed. Remove everything that not either an obvious 'form' thing that the filter will pick up or a simple odds criteria. You should see the consistency pick up then.
-
I think it's the moment the problem is that the software results and my results are so different and I can't figure out if that's an issue with the bot not firing correctly or it's simply the conditions that the software sets? I know there's a big issue at the moment with the favourites in the fact that it's taken at BSP and obviously favourites can move around so much in the last five minutes before any race because of scalpers moving money in the market. The issue with that is the I can't really trust the backtesting so it's all about forward testing for me which means I have to keep the stakes low which means I'm not making any money to cover the costs of the software.
If I'm perfectly honest I'm getting a bit frustrated with it. I've got the software for another five months and hopefully I can find a couple of strategies that will make it work and this issue with the favourite gets resolved so the software results are similar to my results [although I realised there is going to be some variation but not the amount I’m getting] -
@john-folan ...bit late for that, its all dropped out!!!
-
@karl-pick said in The BTC Horse Racing Thread:
After a dodgy 10 days, where I began to doubt everything the last 3 days have turned a massive corner, hopefully it continues!!!
Horse trading will send you grey lol.
-
@karl-pick
I downloaded the individual resaults for all the strategies i trade. Grouped by day to see how drawdown etc would work with a portfolio approach. What is obvious is you can get a period of several days where your balance goes down, then up to a month as it basically flatlines before that drawdown period is done. In that context 10 days is nothing. -
@jonathan-jones I make you right on the flat pace backs. The AW is profitable, but looking at turf only, these seem golden on the turf March through September. I am running a different version looking at the rating for that particular race and it looks good for March through October. Flat season runs March to November but that really is the arse end.
These look a good addition for those months.
-
After a dodgy 10 days, where I began to doubt everything the last 3 days have turned a massive corner, hopefully it continues!!!
-
Very good day yesterday. I am running:
Back C&D winner (slightly modified so its possible to trade)
Back Fave v3
Lay beaten Fave
Low Odds Lays
Pace Lays
Pace Backs (Nothing on this till next month earliest i dont think)Only on sim till the end of the month so i can check the automation but plus 14.78 points yesterday. Looks like a good mix of backs and lays. STill got to go through them to check with the backtest etc but makes a nice change. Feb has been a bit crap so far.
EDIT#
So, just gone throughthe backtest for yesterday and its a pretty close match. Backtest says i should of got 14.6 points so thats close enough for me. If anyone is interested the breakdown was (according to backtest but no reason to doubt it):Back C&D winner: 8.81
Back Fave v3: -0.06
Lay beaten Fave: 0.98
Low Odds Lays: 1.97
Pace Lays: 2.94 -
@lee-colvin I had;
Jered Maddox at 17:30 Dundalk @ 5.8
Ghumama at 19:30 Dundalk @ 6.4
Local Bay at 19:35 Southwell @ 2.2
Bot had a few misses too though so ended up +3.5ish points. No short odds lays for me, although I may have missed them as loaded the tips manually after some of the races had already finished.
-
@lee-colvin said in The BTC Horse Racing Thread:
@john-folan said in The BTC Horse Racing Thread:
2/2 for me today.
Looking forward to Adam’s addition re odds differences between horses ala Little Acorms. Should give us more angles to trade backing and laying.
What was your two today matey? Just the one that fired for me, 14:35 at Kelso which was a lovely little winner
Had one at Dundalk as well. I wouldn’t worry about them not firing there all the time. It’s one of the less profitable courses.
-
1/1 for me on short odds
-
@paul-foster The above at Kelso was a SOL mate, it’s crazy how we all get slightly different results eh? Makes it even more interesting
-
@lee-colvin bot didn't fire any short odd lays today, but the C&D winners was more than enough to make up for it.
-
@john-folan said in The BTC Horse Racing Thread:
2/2 for me today.
Looking forward to Adam’s addition re odds differences between horses ala Little Acorms. Should give us more angles to trade backing and laying.
What was your two today matey? Just the one that fired for me, 14:35 at Kelso which was a lovely little winner
-
I had a few good wins on some of the backing strategies today, haven't got back to the bot to check which strategy they came from!
