READ-ONLY

The forum is now a read-only archive.

For bug reports & platform questions → [email protected]

Our community lives on Discord and YouTube — come join us!

Longevity of a strategy

52 replies

mikeyc

Customer, bbp_participant, community, 856 replies.

Visit profile

10 years ago #114641

Since there’s no view on how long a strategy might remain profitable, and the goal of everyone here is to make money, would it not be better to push a profitable strategy to high risk, rather than run it at low risk, only to find it loses it edge over months and years?

So if you have faith in the strategy and it look good on a live account after a couple of months, ramp the risk up very high. Withdraw the original capital as soon as you have doubled your money, and then there’s nothing to lose…

Contrast running at low risk (say 2%), it would take years to make any meaningful money without a very large initial deposit, with risk of failure due to the long timescale.

0

clonex / Ivan Hudec

Customer, bbp_participant, community, sq-ultimate, contributor, author, editor, 268 replies.

Visit profile

10 years ago #134956

After the building of robust profitable strategies become routine, the above is the most important key skill in algo’ trading which is often underestimated.

Agree.

Odoslané z SM-G900F pomocou Tapatalku

0

CMKCMK

Customer, bbp_participant, community, 44 replies.

Visit profile

10 years ago #134961

if you have one account one- strategy, you cannot use your capital effectively, can you? because for portfolio of strategies you need less capital than trading each strategy separately…

Yes, that’s right, but to me the reduced capital is insignificant or rather a trade off in order to spread my risk.

 

To be precise, one currency pair per account (several strategies of the same currency pair in one account). Another currency pair, will be on another account….

 

Call me paranoid but the Swiss Frank unpeg on Jan 2015, wiped out many accounts and even brought down some brokers. By having one currency pair in one account will contain the “wipe out” to one account. Other currency pairs are protected. 

0

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134968

That´s the exact idea to assign them the same risk, otherwise the portfolio would make no sense. You can´t know what is the good one tomorrow, hence assign different risks to different strategies is not a good idea for ideal risk spread. Of course you do NOT want to add “bad” strategies to the portfolio at all. I posted some of mine here: 

A very smooth single IS optimization. This is how most IS should look.
But whats their WFM look like? OOS is all that matters.

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

10 years ago #134972

I´ve long stopped using WFM nor any OOS (actually I never used it in my 8 years of trading since I found the concept faulty from the very start), it´s not useful on historic data, there is no real OOS on historic data. The only real OOS is trading on unseen data and not doing anything about it. I´ve concluded for me to always use all data and guess what, that makes money live now.

 

What do you do if your OOS looks bad? Right, you keep on optimizing until it looks good! That is effectively like using all data IS for the optimization since you will never start trading a strategy with a bad OOS. I personally never understood the concept of OOS on historical data, because if any OOS looks bads, you just go on optimizing until the OOS looks good though, which effectively is like using IS for all data from the very start. An interesting read in that relation is, which explains it in more detail:

 

Reality Check: Why there is no such thing as a true ‘out-of-sample test when using historical data

 

His conclusion also is:

 

“Using a system generator you can create and test different strategies until you find some system that is able to work under WFO conditions. Given any symbol and system there is a complicated enough set of rules that will enable a successful WFO to happen (even more so if you also take into account that the WFO process has rules that can be optimized as well, such as window sizes, anchoring, etc). This does not mean that your system is better able to survive or adapt to future market conditions successfully, it just means that it was able to fit a more elaborate testing scheme on past historical data. There is actually not a single piece of actual real quantitative evidence I could find that shows that a system developed to give a profitable historical WFO is actually better able to generate a profitable live trading result than a system that was developed to give profits through the whole back-testing period above data-mining bias. If you do know this evidence, please post a comment with it so that I can include and comment on it. I can tell you that I’â„¢ve personally seen systems with 20 year WFO successful tests fail bluntly under live trading conditions.”


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134976

What do you do if your OOS looks bad? Right, you keep on optimizing until it looks good!

I’ve discovered your problem.
If OOS is bad on the WFM matrix, you throw it away. OOS represents real time. You cann’t go back from real time. If your optimization process isn’t working on OOS its not going to work in real time.

Quants have used this process for decades.
Kevin Davey has been trading with his own money for 10+ years now using this process. Hes pretty successful (understatement). I’ve had the chance to speak with him privately recently. He actually details how to use data without “abusing it”. One of the key abuses and destruction of data is testing a strategy and building it on the full data.
 

0

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134977

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134978

And don’t say your profitable until a couple years go by. You said the same last year then ditched those strategies. A couple months for a 25 years IS 2k trades isn’t much for a sample yet.

0

Patrick

Customer, bbp_participant, community, 422 replies.

Visit profile

10 years ago #134983

geektrader

you mean there is no OOS, only live trading is OOS. at the end of 2014, live trading was 01.01.2015-31.12.2015 (because it was future, unknown)

When you build a strategy (on 2014) e.g. IS 2003-2010 and OOS 2011-2014 than next year was THE real OOS.

If you keep the same workflow, now, one year later, you can backtest instead of live trading and year 2015 is THE OOS. 

Because its the same if you wait one year or if you trade the year. In both cases if OOS performance is bad you dont trade the strategy any more.

 

i know traders using OOS at indexes and they are success.

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

10 years ago #134984

I hear you, but I don´t care really what “quants have been doing for decades”. You know, wars have been there for 10000 years, yet it doesn´t mean they are right. Have you read the article I posted? That´s also from a quant who is doing it for a decade. Main thing for me though is: I am making money, that´s really everything I care about, and it´s more than a few weeks already;)


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134985

A whole few weeks!
don’t get too excited now 😉

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

10 years ago #134986

geektrader

you mean there is no OOS, only live trading is OOS. at the end of 2014, live trading was 01.01.2015-31.12.2015 (because it was future, unknown)

When you build a strategy (on 2014) e.g. IS 2003-2010 and OOS 2011-2014 than next year was THE real OOS.

If you keep the same workflow, now, one year later, you can backtest instead of live trading and year 2015 is THE OOS. 

Because its the same if you wait one year or if you trade the year. In both cases if OOS performance is bad you dont trade the strategy any more.

 

i know traders using OOS at indexes and they are success.

 

If it makes money for them, more power to them! If using historical data it´s a fact there is no OOS, any OOS you perform is already performed on EXISTING data and you throw away strategies that don´t do good on OOS right? You keep on changing the strategy or optimizing it until the “OOS” looks good. So what you´ve in fact done then, is just like optimizing everything IS. Just read the article, it explains it very well.


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

10 years ago #134987

A whole few weeks!
don’t get too excited now 😉

 

Are you trolling or what are you trying to do? One more stupid post like that and you go to my ignore list, no time for non-facts based discussions like you seem to like them, sorry. I am here to make money and just share how I´ve done it so that others could make use of that knowledge possibly. I gain nothing from doing that. If others have another way that makes them money, that´s all fine by me, more power to them! I though give a damn what you think about that nor if you like it or not how I do it, I am here to make money, not to waste my time with stupid comments. Stick to the facts of a discussion and a friendly tone or be ignored by me, easy. I absolutely have no time nor energy to spend on time-wasting troll comments.


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

Threshold

Customer, bbp_participant, community, 722 replies.

Visit profile

10 years ago #134989

Sorry I’m not trolling. Just be honest and dont say “you are profitable” and leave out the highly important fact that its only for a few weeks. Leave up a MyFxbook link if you are truly confident this time. You’re using that statement to undermine my points but you really have only been running it for a few weeks now. I’m defending what I’ve said against non-facts. Thats the last I’ll say on the matter.

 

0

Patrick

Customer, bbp_participant, community, 422 replies.

Visit profile

10 years ago #134990

geektrader, i read it. i am talkin about keeping a year of unseen data. make your generating, tests etc. and then check the OOS-if strategy is bad, dont trade. if good, trade. 

 

like OOS 1 in generating and OOS 2 after. OOS 2 is unseen data.

 

BUT i believe that generating on 29 yrs is good way, more data -less DMB

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

10 years ago #134993

You basically mean: generating a strategy and not touching it for a year, then run another backtest on that year and if it was profitable you trade it? Yes, that´s a real OOS then. Just that you had to wait a year for that to happen and got 1 year older.

 

Yes, less DMB for sure, especially if statistical enough trades (2000+, better 4000+) and lower timeframes (more data points, lowering chance to come from DMB even more). Still no guarantee for profit of course, but as close as you can get.


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

Viewing 15 replies - 16 through 30 (of 52 total)

← 1 2 3 4 →