A few questions from a beginner

• 09/17/2017 at 8:23 PM #46434
  jebus89

  Hi all.

   

  I got a few questions for all of you who have actual systems going on live accounts.

  • How long does 1 system usually stay profitable for you?
  • What timeframe is the majority of your systems? (if you have multiple pls say how many youve got going on live account! if thats too personal please say “between 5-8” or something vague like that so i can at least get an idea 🙂 )
  • Without going to specifics unless you want to, hows the earnings pr year ?(or months?) (are we talking low 1K-10K, low medium 10K-20K, medium 20K-30K, medium high 30-50k or higher? I know it may vary but avg?)
  • How long have you been running live systems?

   

  And finally one stupid question: If you have, say 3 active systems generating profit every month, why would you not just increase the contracts every week/month depending on the frequency and profits from trades? wouldn’t you be a millionaire within a year or two?

  

  jebus89

  Participant

  Topics: 118

  Replies: 586

  Been thanked: 238 times
  09/18/2017 at 12:45 AM #46441
  Wing

  1. Define this question please. What timeframe are you talking about? What is our maximum allowed drawdown? Etc.

  In terms of changing market regimes I like to think that if you optimize and OOS test your strategy based on 6 months of data in total, that data will on average encapsulate 6 months of market regime changes. Thus your system will theoretically be valid for around 6 months.

  2. With timeframe you need to think about what you are referring to. The amount of time per bar is not as important as the amount of time your positions are held on average, or how much lookback in terms of time your entry/exit conditions have. You will find that most systems here are in the 5m to 1 day range. My best system so far was based on 30m timeframe (DAX), but it will depend on volatility, spread and signal/noise ratio of what you trade.

  3. Not many people will want to answer. Some have lost a lot of money, and some have earned a lot. The majority will probably be negative, since the broker earns money. The return per month will vary greatly, but in a year I have returned my initial capital several times over. But remember, this is CFDs — leveraged trading. The risk-adjusted return is therefore nothing to brag about in my case. In the beginning you will likely make costly mistakes, so be careful and fully consider your money managment. I could have blown up my account on at least one occasion.

  To your last question: When you start trading really large volumes, especially in illiquid markets, you will have slippage — price moving as you buy or sell. This is also known as “cost impact”. But yes, increasing your contract size as you earn is part of money managment. If you start out with 750000 and want to get to a million in a year, that is possible. If you are trying to grow 10000 into a million in one year, you will probably fail miserably.

  

  Wing

  Participant

  Topics: 6

  Replies: 159

  Been thanked: 77 times
  09/18/2017 at 7:21 AM #46442
  jebus89

  Thx for some answers Wing!

  So let me try to clear this up a bit! I have read that “all systems” that work today, might not work tomorrow, or 1 week from now, or 1 month from now, or maybe years. Not having a working active live system today myself i wasnt sure how much truth there was to this statement so I was wondering if its “normal” to be using a system for say 6 months then usually its broken/useless because… why exactly?

   

  Regarding timeframe, my question was more “what timeframe is the majority of your successful systems running on?” Ive seen alot of guys on this forum going for 30-1h on the dax. makes sense, you should be able to catch alot of voaltility within 30m – 1h compared to 5 min candles.

  About the money-question: i’d figgure that not many would wanna answer this but if someone actually would that would be awsome! I kind of dont wanna put in the 300+hours like u did Wing, for then to have a rather bad system gaining 50$ a month… I just wanna know how good you guys actually have made it.

   

  And ty for giving some insight at the final one there 🙂 Havnt rly thought about the cost impact all that much.

  

  jebus89

  Participant

  Topics: 118

  Replies: 586

  Been thanked: 238 times
  09/20/2017 at 4:10 PM #46718
  Wing

  The reason systems ‘stop working’ suddenly and enter a bad streak of losses are manyfold: It could be bad ‘luck’ (You have the probabilities in your favour, but something fundamental happens like Brexit or North Korea starting a war — it is hard to plan for this), or the market behavious changes (moves from lots of trading in spring to less volatility in summer — and your systems are not specifically trained for low volatility). The most common ‘regime change’ is from mean reversion to trending, and vice versa. This occurs all the time, on different time-frames. The way to combat this is by having several systems running, each a bit different. You hedge your risk. The combined equity graph of all systems is what really matters, not if one system has been performing badly for a month of two.

  The notion of systems and their running equity graphs is actually very peculiar, since it adds a second layer. You are trading for example the DAX, but if you run a system, you are also trading a programmed system, that takes relies on certain DAX price movements to profits.

  I don’t think you need to spend 300+ hours to make some money. I made my best system like 50 hours into it. Most of my remaining time was spent trying to do overly complicated stuff. In fact, the most important skill is to be able to evaluate strategies. Is it a realiable strategy or not and does it fit into my strategy portfolio? To do this, you barely need to be able to program. Just study the many systems posted on this site. I have gone through many of them, and to me, there’s a percentage of them certainly worth running. So, building your own strategy from scratch — not needed if you just want some extra money.

  Cost impact you can ignore for now.

  

  Wing

  Participant

  Topics: 6

  Replies: 159

  Been thanked: 77 times
  09/20/2017 at 9:10 PM #46742
  jebus89

  Deeper into the forums i go! 🙂 Ty for reply

  

  jebus89

  Participant

  Topics: 118

  Replies: 586

  Been thanked: 238 times
  09/27/2017 at 12:56 PM #47500
  robdav

  Hi Wing

  You answers to jebus89 make a lot of sense so I was wondering if I could ask you a couple more.

  I’ve been looking at a lot of the trading strategies, writing my own etc and I keep coming back to the same question. How far should you back test/walk forward your system?

  Market conditions change from trending to mean reverting etc. So what’s the point back testing a mean reverting system going back three years when we know the market might have been trending a lot during that timeframe.

  Obviously, we can ‘fit’ a system to say the last 30 days and it might pick out 10 trades at 100% wins which might look good but won’t work going forward.

  I’d be interested to hear your thoughts/advice and anyone else’s for that matter.

  Thanks

  Rob

  

  robdav

  Participant

  Topics: 65

  Replies: 116

  Been thanked: 7 times
  09/27/2017 at 4:02 PM #47550
  Despair

  Usually you want to test your strategy just in different market situation since you never know what will come in the future. So even if you are developing a mean reverting strategy it will be nice to see that it doesn’t lose too much in periods when the market is trending.

  If you adapt your strategy just on recent data you might get in trouble when the market changes.

  2 users thanked author for this post.

  Nicolas, robdav

  

  Despair

  Blocked

  Topics: 97

  Replies: 1034

  Been thanked: 218 times
  09/28/2017 at 9:38 AM #47612
  Wing

  What I try to look for when I test is this:

  • Systems does ‘okay’ on some uncorrelated markets, like the DAX and some currency pair.
  • Backtest data includes negative trend, positive trend, low volatility, high volatility, and mean reversion.
  • Most important factor IMO: the amount of trades in your backtest — this is your statistical base and decides the degree of certainty you can have in the results. I trust a backtest with 500 trades exponentially more than one with 25 trades. I avoid backtests with less than 50-100 trades.
  • How far to look back: As far as possible, is usually the best. There are advanced methods in PRT to include adaption in your strategy. This means that you can weight recent data as more significant than old data. This way, you system can sort of adapt to market changes even when you are running it.  Example I have posted here: https://www.prorealcode.com/topic/code-correlation/

  1 user thanked author for this post.

  robdav

  

  Wing

  Participant

  Topics: 6

  Replies: 159

  Been thanked: 77 times
  09/28/2017 at 11:27 AM #47640
  robdav

  Thanks Wing. I will take a look. It looks like I also need to find a system that trades more.

  

  robdav

  Participant

  Topics: 65

  Replies: 116

  Been thanked: 7 times
• Author
  Posts