Horse - In Running Experts

[mak]

Hi there
need some help opinion
During December i find a way to make some profitable trades in running.Not something big, but i was happy with that...

Last week i had to change vps provider because I had problem with the timing of the windows clock.It was loosing almost 5-6 minutes per hour so my triggers where "confused" many times...

I am now testing another provider where I can see that it's internet connection is not as fast as the previous one...and my bots don't perform so well - small losses

What I would like to ask just because I have no experience, are your bots performing ok? I have read regarding liquid issues and cross matching etr
and I don't know right now if bots aren't performing well because of slower internet connection, or there is market related issues or what

Happy New Year btw
mak

[Captain Sensible]

Always hard to pin point why your bots not working without knowing what it does but one week is too short time a time to write things off. At the moment there's so little racing plus it's the xmas period you can't expect things to perform as normal.

If your bots waiting for prices to appear before firing a slow connection is always going to be a problem as the 'value' prices will probably have long disappeared before your bet hits the markets and you'll be only matched with what others see as poor value. Personally I haven't come across any real problems due to liquidity or x matching in the last few weeks.

[doris_day]

mak, it could be both liquidity and connection speed. That will depend on your algorithm.

My algos became rubbish a few months ago because they depend on good liquidity and over the last year, in-play liquidity has gradually deteriorated. That and the fact that over time markets become more efficient as new traders try stuff out. Nothing lasts forever but I suspect my algos will improve once liquidity does. If it ever improves that is. In my opinion in-play horse racing liquidity won't improve until better, 'liver' pictures are more readily available. That for me is the key.

Betfair showed their hand when they introduced the PC. They really don't like 'sharp minds' at all because they remove money from the markets too quickly and so Betfair don't benefit from the extra commission they make because of 'churn'. So now their philosophy is to do everything in their power to make it as difficult as possible for the 'sharp minds' to operate.

[ddaann]

I agree what previous posters said, and to add I've seen similar problem with liquidity last winter as well during "big freeze" and few weeks after. IR punters are vary at the moment as there wasn't hardly any racing for the past month. Horses are unraced and they couldn't train on frozen ground. If weather doesn't get worse in January, I think it should be back to normal (pre December liquidity) soon.

[mak]

Thanks guys for the replies
Today I had my vps updated from the provider and it was clear that the main reason was the connection speed...Today everything worked fine and gain yesterday loses with some extra profit..

Essentially I am happy that is sorted out and saw the bots "working" today

If only I had a UK account..But I am fine with it anyway

Much appreciated the help you provide
Thanks again
mak

[doris_day]

Good to hear mak.
Today was good for me too and I generally find the liquidity in January to be quite good, so here's hoping.....

[doris_day]

After a good bank holiday, the markets have returned to their appalling worst, so I've decided to pack things in temporarily until liquidity improves.
I've also noticed that almost everything to do with Betfair has gradually slowed down over the past few months. Eventhough I have fast connections, markets are slow to load and generally every action related to Betfair is significantly slower than it was six months ago. Could be due to cross market matching, could be due to their servers not being able to cope. Frankly I don't really care that much anymore because I simply can't get out of the markets what I deem to be a reasonable return for my effort.
I have other irons in the fire, so I'll be off and wish you guys all the best.
I'll pop in every now and again to see how things are progressing but meanwhile a coder and I are developing a new interface for LMAX, which we both think has a great future, if only they had an interface which took advantage of what they're offering. In typical Betfair fashion (well, they own a majority share), the LMAX interface is crap and bearing in mind its effectively offering the same concept as Betfair but for financial instruments, it simply amazes me that they haven't got an interface where you can easily insert your trades at the price you want.
Yet again, third parties are going to have to do it for them. You really do wonder who is running these companies sometimes.

Best of luck guys....Its been a great few years but I do have to face the reality of things....Hopefully liquidity will improve but what with the PC too, it just aint worth the effort.

[mak]

Yesterday wasn't a good day for me too, and the reason it was for sure liquidity. I haven't got much experience from the markets to know if that is what happening during very cold days or near the New Years Eve but I think I have to start a diary of how well or not each day the bots performed..

Good Luck to you Doris, even though I believe we will see you soon

[Spike]

Mak,

I've been using bots for in-running horses for about two years now- ever since I discovered Gruss and I've done very well at it My advice is simply that you need to give yourself and your bots much more time. It's innevitable that you will have good and bad days- sometines you will have terrible days. When these come along it's tempting to think that you can see the reason but very often that's an illusion, it's just one of those days. Like throwing 10 heads in a row- sometimes it will happen.

At the moment you only have a very small amount of data (one weather damaged month's worth). That's not really enough for you to know reliably how often or how badly you should be expecting to lose, it may very well be that what you are seeing is simply normal for your bot and you still have some work to do on it.

In running horses used to be easy money but the recent introduction of the cross matcher was a massive change that's ended the party and given all of us a huge headache. I've only just started to get back into decent returns myself after a lot of work over the last 5 weeks. These markets are now very tough to crack so don't be down hearted if it takes a while.

The trick to getting your bots working is patience and analysis. If you're intersetied I'll break down the way I do it for you.

1. The first thing you need is an idea- which you already have, then you just run it and run it for a good long time (a few thousand strikes and at least a few weeks) WITHOUT MAKING CHANGES and gather as much data on it as you can. An easy way to do this is to use the transaction log as it records all the details of all your bets and saves you the bother.

2. Work on that data- go through at and mark it up to see what's what, for instance that might mean working out which bets were opeing bets, which were greenup bets or whatever's relevant to your bots. Make pivot tables and use them to anylise what worked out (if anything worked out) and what didn't.

3. Hopefully your analysis will have shown up some opportunities and some weaknesses - if not, go back to step one. Based on these you now make a second generation of your bots and try it out ALONG SIDE THE FIRST VERSION to give you a control for your experiment. You can do this either using a different instance of Gruss, recording the data separately using vba or putting markers into your bets that will let you easily separate them in the transaction log eg, version one bets £2, version 2 bets £2.03.

4. Now you repeat the analysis - you now have a lot more results for V1 and you can compare them to V2 to see if your conclusions about the first itteration of the process held true.
It may be that at this stage you switch off the worst performer (hopefuly that will be V1) and operate a kind of "winner stays on" system- but before you do that be sure that you're making the judgement based on enough data.

Now repeat steps 3 - 4 forever.

The trick is to be calm, patient and realistic, Don't be too tempted to blame external factors if you lose money, after all that's how the markets are and you can't change them. The only thing you can control is your bots, so if you want to make money that's what you need to concentrate on.

Good luck.

[osknows]

Also, don't fit your system to the data... work out your system FROM the data

[mak]

Guys THANKS
Much appreciated the help over here.That's why this is one of my favorite forums.I know that i have to study a lot about the bet placement of my bots, but i find it difficult for the time being.I have 4 of them on each race and I really don't know how to check them out,since "they" are trading and some times the opposite bets aren't get matched.I will find my way i believe somehow.I also believe that this period is very very difficult for a newbie like me, because there are so many fragile parameters

Today I noticed for example that the bots performed very well without cross matched formula on, and really don't know if they could better with the formula

I am not in hurry though, so testing and trying is the key I guess.

[Spike]

Mak,

Do you mean that they performed well when Gruss wasn't showing cross matched liquidity?

If so- my experiece is also that the "real" bets are a much better guide to what's happeing in the market. For horses the only data I use is the available to back (left hand side) prices.

[mak]

Mak,

Do you mean that they performed well when Gruss wasn't showing cross matched liquidity?

If so- my experiece is also that the "real" bets are a much better guide to what's happeing in the market. For horses the only data I use is the available to back (left hand side) prices.

That's right.At least for yesterday they performed very well.I also use the back prices only. At least for now I have found that bots perform well at small flat races haven't find a way yet for the big ones (above 2m)

[Mistermind]

All these complaints about diving liquidity, can anybody quantify the fall, the percentage of decline for instance?

[mak]

hello everybody...
today and yesterday my bots aren't performing as they used to
and i would like to hear how you are doing if it is not problem
i am talking about uk i/r markets and i can't see any obvious reason why..

the only reason i can think is that the trucks are new for my bots and need adjusting if possible.Don't know just yet

I know that a lot depending of the algos and the strategy of the bots but would like to hear about it