Running Bots (Automated betting)

[osknows]

What conditions have you 'tested' your system and for how long? It's a good thing you are asking these questions but it seems you may not have tested your system fully under 'live' conditions. 80% to 20% is natural variance. I have a system that is approx 97% accurate and profitable and I have a system that is approx 10% accurate and profitable but over days/weeks these percentages vary significantly; accuracy isn't everything - the important factor is yield/ROI & value.

Your system may be simple but obviously it IS losing! You need to investigate why...

Also as knot points out 7 days is nothing. You really need to test your system out over a bare minimum of AT LEAST 1 season. Obviously you can shorten the time to test things out but if your system doesn't hold out long term then you aren't going to do too well.

I'm pretty certain that anyone here who is successful uses their own systems and have experienced similar problems. As long as you stick at it, adapt and analyse your mistakes you will improve. Even when you have a system, things change - it doesn't last for ever!

You have some very good advice on this thread - Good luck!

[Captain Sensible]

I trade manually most days daily so after a while you get used to certain bets coming in at say 30 seconds before the off.

On some of my bots I just added a cell that'd generate a random number and and use that to delay bet submission so my bets wouldn't go into the market at regular times by the normal auto select market.

[doris_day]

As others have said, you don't have enough data. What, for instance do you mean by 'about 7 days' or '7(ish) days' ?
There will always be variance in any system that uses Betfair markets to trade in. If I were you I'd leave it a little longer, gather more data and then come back in a few months and tell us if your 7 days on/off is still continuing.

[pugs]

Before Betfairs “keep bets” I used to lay every single runner/ horse race at 1.01 and overall with my staking plan I made a profit, but like you, one week I would make money hand over fist then the following week not pick up a single win.

After watching and studying thousands of races and in running markets I soon learned that a beaten 1.01 very rarely had anything to do with the horse itself but it was due to pressure of money, i.e., too many in running backers and not enough layers. So I adapted my strategy to “cancel all bets” when I felt the value/potential for me to win had gone. This led to an even greater strike rate and very tasty profits. Some days I could cancel up to 50% of my bets and those days I wouldn’t pick up a single win, and some days I couldn’t cancel ANY before I was matched and those days usually proved very profitable.

I studied so many things before I learned it was pressure of money (type of horserace, number of runners, distance of race, going, racetracks, an uphill finish) and even though things like “heavy going with an uphill finish” and “Lingfield all weather” were good for getting a 1.01 beaten ultimately if there were more layers than backers, backers would simply “hoover” rather than take risks.

The point is, you need to study/learn what makes it a winning or losing day (week) for your system and adapt your strategy accordingly, then maybe if you lay, back. If you back, lay. Or if that’s not possible then just not play that strategy/system on those days. In my case I couldn’t “back” because I don’t have live pic’s, so I just cancelled. But I hope you get my point and understand the importance of identifying why you win one day and not the next.

p.s. What’s the system

[olepunter]

adrianmfoster ,danjuma I would add to your paranoia if you were International users,
Are you International users of BF or UK users?

[olepunter]

in addition I can say that BF have done everything they can to put us off /BF internationals users/

[danjuma]

UK user.

[Skroober]

If you do get to the bottom of this (which I am sure that you will), and if it doesn't compromise your stratergy would you mind posting how / why it turned sour after 7(ish) days?

[nerobot]

How many races would you say would be a good start for training a bot? I know there is no actual answer, but a ball park figure would be helpful. I've currently got 35, which I'm sure is quite small, but good for initial testing I think.

Steven

[Lewis]

35 may be ok for getting a rough idea of how the market moves however that sounds like one days racing which is far too small to develop a working system. There is too much variation solely on the day of the week never mind many other factors that can change the way the markets operate.

[nerobot]

Lewis:

Yeah, I guess when you put it like that, 30 races if quite small, being only a single day's trading. Well, I've now for a remote server set-up so I can collect more data than I was able to do so before. Hoping to get about 1 months data and see if that produces a better system. Thanks for the info.

Steven

[wojtazzo4]

hi guys,

I have tested my system for 5 days, which is about 225 races, and it shows quite nice profit. only one day had a minus figure but only just under 0. I wish to try my system in 'Training mode' with some software.
do you know any program that placing bets in a 'Training mode'?
do you think my system is ready to go using real money?

I was trying to find a catch but I checked everything few times and everything is correct. So I assume that should work.

the only problem I can see is that my bet might not be matched. but using Ł10 stakes should this be a really big problem? I dont think so.

but as you said about Betfair conspiracy and so on I am a bit afraid now. What if I start to win Ł100, Ł200 or maybe Ł500 a day using my bot??? Can betfair stop me in any way???

what do you think?
maybe you know any program that placing bets in a 'Training mode'?

[sjaak1943]

Try it it with low ( 10 cents) stakes.
That's more reliable then any training mode!

[doris_day]

wojtazzo4, I don't think 'training modes' give reliable feedback to probable expectation because placing a bet theoretically is quite different to live trading.
I'd just go for it and use stakes that you're comfortable with and build from there. Sadly, because of the lack of liquidity, you'll soon find your upper limit. It wouldn't make a jot of difference if I upped my stakes to £10000 with a couple of my bots as I wouldn't get matched for more than a few pounds because of the poor liquidity.

Don't worry about Betfair stopping you because I doubt they will unless you're doing something illegal. However, if you're a longterm winner you'll start to have to factor in the Premium Charge, which I can tell you is a pain in the arse. I do what I can to increase my turnover in other markets so I reduce the PC effect but I've come to hate every Wednesday.....
Best of luck with your algo....but just do it......

[wojtazzo4]

to sjaak 1943
thanks for advice. I'd like to try it for a minimum bet as possible but because this is 'in play' system I thinki that 10 pence wouldnt be good because of double time of submitting bets. So sadly I'd rather use Ł2 stakes which I hope won't ruin my bank.

to doris_day
I just wondered if training mode is realistic but you have convinced me that its not a good way to testing any system. thanks for that.

I dont know much about Premium Charge. If I get to earn such a sum that would be nice even they charge me
Hope my system will work in real.

Wishing you all good profit from every race.