General Betting
There is currently 1 person viewing this thread.
Yesterday I conducted an experiment IR in the 20:40 at Ripon. The race went off without Figaro who failed to enter the stalls at the last minute, but too late for Betfair to remove the horse from the card.
Figaro’s price drifted out to around 10. I then started to back the horse fairly aggressively, placing repeated back bets of £100 every 2-3 seconds at 1.01 (to a total of several thousands) to see what would happen. My primary aim was to see how easy it would be to try and drive the price down to a point where it would have a material impact on the prices of the remaining horses in the race and provide a can’t lose dutching opportunity.
I was quite surprised by what happened. Each time I placed a bet I would get matched as low as 6, but almost instantly the price would return to around 10. This happened repeatedly. I never managed to get a price lower than 6, and as the race went into its final stages I realised I didn’t have enough available bank muscle to drive the price any lower.
This raised many questions in my mind, not least the behaviour of the bots. Clearly the bots were happy to return the price (by placing lays) back to its original level despite repeated money coming in for the horse. They were doing this in complete ignorance of what was unfolding. This on the face of it would seem to present some opportunities.
I guess my overriding question is: how do these mindless bots decide on the prices they are going to offer? It clearly has nothing to do with what’s happening in the race, and little to do with what money is coming in (well, certainly not a few thousand).
Any thoughts?
Figaro’s price drifted out to around 10. I then started to back the horse fairly aggressively, placing repeated back bets of £100 every 2-3 seconds at 1.01 (to a total of several thousands) to see what would happen. My primary aim was to see how easy it would be to try and drive the price down to a point where it would have a material impact on the prices of the remaining horses in the race and provide a can’t lose dutching opportunity.
I was quite surprised by what happened. Each time I placed a bet I would get matched as low as 6, but almost instantly the price would return to around 10. This happened repeatedly. I never managed to get a price lower than 6, and as the race went into its final stages I realised I didn’t have enough available bank muscle to drive the price any lower.
This raised many questions in my mind, not least the behaviour of the bots. Clearly the bots were happy to return the price (by placing lays) back to its original level despite repeated money coming in for the horse. They were doing this in complete ignorance of what was unfolding. This on the face of it would seem to present some opportunities.
I guess my overriding question is: how do these mindless bots decide on the prices they are going to offer? It clearly has nothing to do with what’s happening in the race, and little to do with what money is coming in (well, certainly not a few thousand).
Any thoughts?
sort by:
Show
per page
Replies: 15
|
By:
I'm not going to give away any big secrets here - but I don't mind telling you in general terms how my bot setup works. I play mainly in in-running horse racing markets, I'm exactly the person you're thinking about.
Basically I have a whole bunch of different bots that are looking at the whole market and constantly looking for tell-tale signs of likely value, they do this in a series of ways: 1: Looking for signs of over-backing- if horses are being backed in more quickly than seems sensible I get stuck in to laying them. In this situation it doesn't much matter what you lay, you just lay, lay, lay. Botting for beginners - easy. 2: Looking for signs high volatility/instability. In this situation my bots price the whole market at 100% and then bracket the prices, so if a selection is priced at 2 then I might simultaneously back it at (say) 2.1 and lay it at 1.9. This is easy to say - harder to do. 3: Looking for chances to strike value back bets, this is the trickiest trick but the best there is if you can do it because there's less competition and it's the easiest way to work with the X-matcher rather than against it. I'm not going to give any clues about how you do it though because it's a bit of a black art. 4: Simple market regulation - keeping the market at close to 100%. I don't really get very involved in this because it's very competitive, it's the strategy most affected by the X-matcher and other strategies offer better returns for smaller risk. Chances are market regulation bots are what matched your bets (if it was bots at all). Potentially they may have lost money as a result because 10% is quite a lot of the market and it may have lead to them pricing other runners poorly, however it's unlikely that they will have lost much because best execution will probably have got them close to true value on the rest of their bets anyway. It probably won't surprise you to know that I've already thought about how to avoid the circumstances the OP describes and I dare say that's the case with most botters. Although I've been caught out once or twice like this it's a situation that I've got pretty well covered. You can pick up the signs of it in the market easily enough and these markets are pretty rare so it's not a huge issue. On the whole if markets look too dodgy I simply stay out of them. In the specific market in question I was active and I made the mighty sum of £6.96 backing and laying Abdicate and Sir Francis Drake. That's more or less a scratch market for me. I matched about 120 bets (not many), I guess my bots just didn't get very excited about that market. My total profit yesterday was £653.05 which is a bit more than my average since they introduced the IR X-matcher. I'd say that most of the time it's more or less impossible for a manual punter to take on the bots in any significant way. There will be a few exceptions when people make programming errors, but as you can imagine I take pretty good care to avoid that and test things extensively with limited liabilities. There are very few players who have the resources to push a normal market around. You'd be risking tens of thousands at a time to make any impact on the market and your chances of getting it back are pretty slim as you'd probably just create value for someone else. If your sole aim was to spoof bots into losing money that can be done, but actually getting your hands on the money that they lose is nigh on impossible. Of course there will always be exceptions in quiet markets or with poorly designed bots but you're on a hiding to nothing if you try to take on the bigger players. There is sometimes a little bit of sparring between rival bots trying to spoof each-other but it's a bit of a silly thing to get involved with as it can be expensive and rarely offers good returns. Doing it for pride is just dumb IMO. |
|
By:
Thanks for that Aye Robot, very interesting. For some time now (several years) I've been running a pattern recognition bot IR driven by a neural network. It doesn't employ the techniques you describe above (well not consciously - the NN is a black box, therefore its rationale is unknown, although much can be surmised). The profitability levels are not dissimilar to yours, perhaps 2-3 times that in recent months, particularly it seems, since the introduction of the cross matcher. But it is now clear to me that there are whole aspects of the market that I haven't looked at that merit further investigation.
|
|
By:
£1500-1800 a day? That's a pretty good score. what more can you need to know about IR bots? Even before the XM came in I wasn't making that much - more like £1k a day.
|
|
By:
That's what I like about this game. There's always something new, some unexploited possibilities.
|
|
By:
aye robot
can i be ure best friend |
|
By:
If you don't mind me asking Artisan - how many IR horse bets do you match each day?
|
|
By:
2000-3000. Yesterday was 2662.
|
|
By:
 artisan |
|
By:
How does your neural network work then? Presumably you have some sort of programming background?
|
|
By:
Something to bear in mind:
If Betfair do not have the time to remove a horse from the betting and it remains there, players will have money "trapped" in that bet. For example, if you have laid the horse and it is holding £100 of your account and you need that money for in-running, you need to trade at any price to loosen up your money. So if you laid at 11.00 for £10 /liability £100, you have £100 locked up. If the race is going in your favor and you don't need the money, fine. But if you need to loosen up the money, then you need to trade (even £20 at 6.0 will loosen it up). Signed Been there when I overexposed my account one day |
|
By:
does a bot back then lay
or lay then back does it ever red up |
|
By:
I don't really have a programming background, and for most of what I do that's not necessary. I've just cobbled together a few proprietary pieces of software.
I use them in a number of ways. Firstly, (and this is how I started)I use one to do pre-race selection. The NN has been trained (using things like distance, type of race, prices before the off, some form info etc.) to select those races which are more likely to be profitable for my strategy. For this I just use Braincel, an Excel Add-in. I then use a couple of other NNs (Braincel/Excel is not fast enough IR) that have been trained to look for profitable patterns IR. The secrets here are around getting/selecting the right sort of data to train them in the first place, and then keeping them up to date (progressively trained) to adjust to changing market dynamics. |
|
By:
Bots do whatever you want.
But right back to my days of basic, cobol, fortran.....the programs are only as good as the programmer |
|
By:
Yep, bots will do whatever you want (within reason). Personally I don't red or green up (I see that as losing value). My software just goes for value and lets those bets run out. 38% of my races are losers.
|
|
By:
Artisan - have just sent you a message [:)]
|