Profitable automation suddenly loss making

[showtrader]

Hi all

After a number of iterations, came up with a profitable set of rules for trading on all US/UK/IRE racing.

Over a period of over 2 months, only had 5 losing days (small losses at that). Overall ROI on stakes just over 3%.

Then suddenly over the last two weeks, getting filled in all over the place and had 12 losing days and 2 winning.

Initially just put it down to variance, but feels way past that now.

Anyone had something similar happen to them? I guess if someone figures out more or less what your bot is doing, then they can spoof you and take advantage of you. Is that the most likely explanation for this drastic change in performance?

Relatively new to this, so curious what people's experiences are. Is this just part of the battle, hoping folks can't spot what your automation is looking for?

Thanks

[ShaunWhite]

It can happen. I'd suggest dropping down to micro stakes and just keeping an eye on it.

"I guess if someone figures out more or less what your bot is doing, " ... very unlikely unless you're a massive player. There's 000s of people and bots coming and going every day and things do just vary from time to time.

Hopefully whoever is getting your money is also losing even more to someone else and they'll stop in a week or two. Hence suggesting you just drop down to monitoring it on small stakes rather than stop completely. Maybe even a very small change might be enough.

[Archery1969]

Sometimes its very obvious what someone is doing, espeically on low liquidity markets. On greyhounds over sveral months someone was reversing the price on the favourite using £300 - £500 stakes which kinda of stood out like a sore thumb. Then came along someone with deeper pockets and forced their positions to close with a loss no doubt.

I dont know but i imagine their are big players out there using self learning AI / pattern finding software and jumping in and out of the market making thousands of ticks per day.

[ANGELS15]

Hi all

After a number of iterations, came up with a profitable set of rules for trading on all US/UK/IRE racing.

Over a period of over 2 months, only had 5 losing days (small losses at that). Overall ROI on stakes just over 3%.

Then suddenly over the last two weeks, getting filled in all over the place and had 12 losing days and 2 winning.

Initially just put it down to variance, but feels way past that now.

Anyone had something similar happen to them? I guess if someone figures out more or less what your bot is doing, then they can spoof you and take advantage of you. Is that the most likely explanation for this drastic change in performance?

Relatively new to this, so curious what people's experiences are. Is this just part of the battle, hoping folks can't spot what your automation is looking for?

Thanks

It also sounds like you've not really been doing this for long enough to really get a true picture of how the strategy plays out in the long-term.

In answer to your question whether other people had experienced similar, I do a lot of back to lay trading on racing. When I first devised the methodology for it I did quite well for a few months and then when I reached the September of that year bam! a strategy that was working well with 60% success rate was now operating at 40%.

I remember posting about it on the forum and the general opinion was that it could be down to seasonality. Since that time I've been very cautious in September as I've noticed that the selections definately don't perform as well and that seems to be borne out in each year's records.

You mention US racing. Those markets seem quite thin with little liquidity so may be harder to trade/get matched. As for the UK and Irish, again some Irish markets may be thinner than their UK counterparts so again you may need to go back over what you've been doing and see whether that may have been a factor.

[doodaado]

Hi showtrader,

I experienced exactly the same situation about a month ago. I had a strategy for UK/IRE horse racing that was consistently profitable for 3 months, then all of sudden it made big losses on 6 days out of a week. The 6 days wiped out all the profit of the previous 3 months plus 60% more. I really struggled to understand how it happened. It's just not logical. I was doing small stakes of £20-30. The weather, race type, etc. of that week was pretty normal. On day 8, I pulled that strategy completely, couldn't afford to lose more. A few days later, I spoke to my dad (he's a stock trader) and he gave me exactly the same advice as ShaunWhite did above - drop down to micro stakes instead of stopping completely. So I restarted running the same strategy on micro stakes and I've been monitoring it since then. It's not been as good as before, but no massive losses either, ended up becoming a mediocre strategy.

BTW, this strategy has a "sibling" that was derived from the same base principles a few months ago. I used different approaches to risk control, and they evolved in different ways, so they ended up becoming 2 different strategies. The "sibling" is still doing well, hasn't experienced any issue at all.

[doodaado]

I also tried making a small change to the strategy and running on micro stake. I thought if it's AI it wouldn't react quick enough to the slightly varied version. But the reality was that it didn't go well either...

[the4droogs]

If you have a decent idea what markets you may be incurring a loss on, you could set up a screen recording of your ladder interface while the bot is running and see if you can see anything unusual occurring.

Are you working in low liquidity conditions, or is the market fairly well formed?

[Anbell]

I would def check which part of the US/UK/IRE spread was causing the problem

[Euler]

After a number of iterations, came up with a profitable set of rules for trading on all US/UK/IRE racing.

There is an element of seasonality to UK and Irish racing. Not sure about US racing.

But the market shifts in August as we reach the tail end of the UK flat turf season. This may have an impact.

[showtrader]

Thanks all - really useful feedback.

First thing is I made a typo in original post - I'm actually trading AUS/UK/IRE, not US! US is just not enough liquidity at the off (plus race times vs off vary so much) to green up, so waste of time I find.

So I've managed to pinpoint the times of day that I'm getting filled in, I'm pretty certain by someone/AI that's worked out what I'm doing and is spoofing during low liquidity periods.

Now running same thing but not during those periods and back in profit today.

What will be interesting is whether I can dip back into those periods in future (they were profitable in the past), after a few weeks, if that someone/AI has moved onto something else.

If not, I also have some ideas for conditions that would give some indication that a spoofer had moved the market, but not gone down that route yet.

It's good fun anyway, I never really thought it would be as easy as finding something profitable and then sitting back while it just kept on working!

Thanks

[Euler]

I tend to lurk in higher liquidity periods as it allows me to hide in the noise. If you are outside of that period I would imagine any activity will be easier to see.

[ShaunWhite]

I never really thought it would be as easy as finding something profitable and then sitting back while it just kept on working!

That's entirely possible too if it's based on sound principles and I've got one approaching the end of year 3 completely untouched. I've spoken with a lot of people who specialise in automation and they tend to have one core method they evolve and adjust rather than various strategies that come and go. The key to longevity is simplicity.

[firlandsfarm]

Hi all

After a number of iterations, came up with a profitable set of rules for trading on all US/UK/IRE racing.

Over a period of over 2 months, only had 5 losing days (small losses at that). Overall ROI on stakes just over 3%.

Then suddenly over the last two weeks, getting filled in all over the place and had 12 losing days and 2 winning.

Initially just put it down to variance, but feels way past that now.

Anyone had something similar happen to them? I guess if someone figures out more or less what your bot is doing, then they can spoof you and take advantage of you. Is that the most likely explanation for this drastic change in performance?

Relatively new to this, so curious what people's experiences are. Is this just part of the battle, hoping folks can't spot what your automation is looking for?

Thanks

Hi showtrader. You don't say what your stakes are and how they compare to the money flow on your selections. The smaller your stakes and the greater the money flow the more invisible you are. Also don't take it personally! The market changes with every new entrant coming and old entrant leaving. It may just be that someone has found a similar (maybe opposite) strategy that is spoiling yours. Here's what happened to one of mine a while ago ...

AussieDogs.JPG

After a slowish start it behaved admirably for 1,000 bets but then plateaued, thankfully it didn't give up any of it's earlier gains but it had to go.

[deestar]

I never really thought it would be as easy as finding something profitable and then sitting back while it just kept on working!

That's entirely possible too if it's based on sound principles and I've got one approaching the end of year 3 completely untouched. I've spoken with a lot of people who specialise in automation and they tend to have one core method they evolve and adjust rather than various strategies that come and go. The key to longevity is simplicity.

Exactly this.

[bigpapaperry]

everytime i raise stakes , start thinking of holidays and a 2nd income i seem to go on a long losing run.