Trading What I see !?

[goat68]

My wonderful new GH strategy now -£100 over last 2 weeks, i'm not going to look at that negatively though, it's just a drawdown, admittedly larger and longer than any in the 6month backtest (max -£75, longest 7days)

So, if you add the last 2 weeks data to your backtest and run it again what does it say "should" have happened in the last 2 weeks ? Does your backtest match the real world - and what does the "paper" version say should happen ?

Yes I did that a couple of days ago, they do match give or take a few £s, last 2weeks backtest==paper==live, in fact "live" is doing the best!

[ShaunWhite]

My wonderful new GH strategy now -£100 over last 2 weeks, i'm not going to look at that negatively though, it's just a drawdown, admittedly larger and longer than any in the 6month backtest (max -£75, longest 7days)

So, if you add the last 2 weeks data to your backtest and run it again what does it say "should" have happened in the last 2 weeks ? Does your backtest match the real world - and what does the "paper" version say should happen ?

Yes I did that a couple of days ago, they do match give or take a few £s, last 2weeks backtest==paper==live, in fact "live" is doing the best!

Are you measuring PnL or something more useful? And what does 'a few £' mean in terms of accuracy bearing in mind that £2 in £100 is a margin of error in excess of the roi you're trying to find.

Anyway you'll have lots of lovely bet data from your tests and live now to learn exactly how discrepanies happen and how to estimate/accommodate them in your backtests. Those live loses should pay for themselves in knowledge.

Last one... If you suspect the market has changed then try to find out in what way they've changed, that's key to being able to evolve a strategy as the underlying conditions change or to target different variations at specific conditions.

[goat68]

Last one... If you suspect the market has changed then try to find out in what way they've changed, that's key to being able to evolve a strategy as the underlying conditions change or to target different variations at specific conditions.

not got a clue! Maybe due to Xmas or Covid !

[goat68]

So, if you add the last 2 weeks data to your backtest and run it again what does it say "should" have happened in the last 2 weeks ? Does your backtest match the real world - and what does the "paper" version say should happen ?

Yes I did that a couple of days ago, they do match give or take a few £s, last 2weeks backtest==paper==live, in fact "live" is doing the best!

Are you measuring PnL or something more useful? And what does 'a few £' mean in terms of accuracy bearing in mind that £2 in £100 is a margin of error in excess of the roi you're trying to find.

Anyway you'll have lots of lovely bet data from your tests and live now to learn exactly how discrepanies happen and how to estimate/accommodate them in your backtests. Those live loses should pay for themselves in knowledge.

Last one... If you suspect the market has changed then try to find out in what way they've changed, that's key to being able to evolve a strategy as the underlying conditions change or to target different variations at specific conditions.

I do suspect I need to do a whole year round-trip to get to know the seasonal variations of a greyhound market, I can see why it takes years to become profitable now......

[goat68]

Do i stop live til new year...assuming it's an end of year effect?
Examined some of the losing bets, seem to be lowish odds 3-4, that steam rather than drift...
I cant see any fault in strategy, if i had gone live and had a £200 rise followed by a -£100 drawdown i wouldn't feel as bad, just my luck live straight into drawdown!!

This is where luck and psychology come together, those that time it lucky on a rise hang on in there, those who just go into a drawdown chuck a profitable strategy in the bin!

[ShaunWhite]

I can see why it takes years to become profitable now......

O ye of little faith Believing that a year ago would have informed your whole year's plan.

[goat68]

This strategy has to be profitable, it makes sense, backtests show a 0.6% return over 6months, it cant just suddenly switch off...!
It's just randomness trying to fool me to giving up just as it hits the bottom!
No chance, let it rip!
Account is down to £200, going to just let it keep going.

[ShaunWhite]

!!
This is where luck and psychology come together, those that time it lucky on a rise hang on in there, those who just go into a drawdown chuck a profitable strategy in the bin!

Can't agree that luck or psychology come into it unless you don't have a methodology to fall back on. It's science not belief. Your launch which has started with a record loss is a sign you should immediately put it on min stakes to keep accumulating intel and investigate what's actually changed, volatility, time to get filled, time between runs if you're using form, blah blah blah. You say you can't find a reason but you know there is one, so don't stop looking till you find it. There's no such thing as can't.

Tbh being 200 up and then losing shouldn't be seen any different to straight losing. If you'd made 100k would you be happy to lose 50k? 'Drawdown' is a period of returns that have a %likelihood rather than being a PnL figure.

[goat68]

So luck/psychology refers to exactly the video Peter's reposted on twitter, Trading Profitably, it messes with your mind and all his green and red blobs...it's exactly what i am doing, ie letting it mess with my mind. Im happy with the strategy, it's just going through a 2week bad bit of variance.. i shouldn't look at it too often, admittedly Peter says dont look in the middle of the day, probably should be dont look in the middle of the week for me!

[goat68]

Ok in going to do a mental test, im not going to look at my account until Sunday evening. It's currently at £200.
I'll be honest and let you know if i peeked before...!

[ShaunWhite]

This strategy has to be profitable, it makes sense, backtests show a 0.6% return over 6months, it cant just suddenly switch off...!

Why can't it? Why shouldn't it?

I could show you several strategies (live or just tests) that have suddenly died or reversed overnight. This year it seems to have been a theme. Just remember how tiny these markets are and how many people are influencing them.

Again its why setting parameters and getting a total doesn't mean much unless you understand why and how it works so you also understand when/why it doesn't.

[goat68]

This strategy has to be profitable, it makes sense, backtests show a 0.6% return over 6months, it cant just suddenly switch off...!

Why can't it? Why shouldn't it?

I could show you several strategies (live or just tests) that have suddenly died or reversed overnight. This year it seems to have been a theme. Just remember how tiny these markets are and how many people are influencing them.

Again its why setting parameters and getting a total doesn't mean much unless you understand why and how it works so you also understand when/why it doesn't.

My strategy revolves around one particular setup, when a dog does X previously and is now doing Y, then the market initially prices it low, and then drifts to bsp, on average!
Id be surprised if it's changed all of a sudden

[ShaunWhite]

not got a clue! Maybe due to Xmas or Covid !

Those are examples of causes, try looking for symptoms. Volatility, fill rates etc etc.

Incidentally, it's worth setting up a script (or whatever you do in flumine) to rerun all the various strategy configs you've ever tried, overnight on just the previous day's data. Strategies come and go, dead ones can come back or failed ones can start working. I'm not sure how flumine works but I have config.txt files that control every aspect of my strategies, incl references to any bespoke code that was required to derive a particular signal; none of my stuff has anything hard coded. I run the same code but just with various config files, much like Guardian rules files I guess...but it's better

If you don't have the necessary configs or hardcode things then I assume you're using some sort of versioning in your development, so perhaps you can just reference all the versions you've created to date instead?

Now you've seen how strategies can come and go I'm sure you'll have already thought about re-running all your old stuff, doing it daily overnight just keeps the burden to a minimum, mine runs just before my archiving, backups, clock reset and reboot. Looking at them daily is a time sponge though so I tend to glance at them every few weeks to check if anything has risen from the dead. I'm not keen on doing a load of work that just sits collecting dust so failed things get added a line added to my SQL script which is called from my end of day batch, and the worst that happens is my end of day processing takes an extra 3 mins and each day the individual charts are appended with the latest info. No effort required.

Excuse me if I'm teaching granny to suck eggs, your an IT guy so all that won't be hard to think of but maybe you didn't see the value in doing it, until now? It's well worth slotting that somewhere in your trading system seeing as it's easy.

[goat68]

if strategies can easily come and go, it then makes a mockery of backtesting. The future is truely unknown, nothing except the secret sauce the 0.01% of Betfair traders who make ALL the money off us minions, can help you... !

This bot has got today to make some money then it is off!

[Kai]

if strategies can easily come and go, it then makes a mockery of backtesting. The future is truely unknown, nothing except the secret sauce the 0.01% of Betfair traders who make ALL the money off us minions, can help you... !

This bot has got today to make some money then it is off!

why not just have a look at other profitable bots to see how they do it