Thursday, April 11, 2013

Daikoku trading system re-optimised

I'm pretty exhausted right now. Pretty much spent all day working non-stop and I'm really happy with my results.
I spent most of today re-optimising my Daikoku trading system by extending the backtest to seventeen pairs and 12 years worth of data. Sample size was 600 trades.
The trading system is really simple. Look at the weekly chart at the end of the week and see if the current week's range is less than half of the previous week. Once the new week starts, go long or short if the top or bottom breaks, respectively. We place our stop loss at the opposite high or low.

As part of my (re-)optimisation, I experimented with different profit targets and R:R ratios. This included the new currency pairs that I had backtested. The profit factor for each R:R ratio is below:

A reward-to-risk of 0.15:1 seemed most optimal. It seemed really low, but those were the figures. But all R:R ratios were positive, so you can't really go "wrong" with whatever R:R you use. Slightly changing the reward from 0.15 to 0.2 or 0.1 didn't change my profit factor much, so this value seemed robust.
The equity graph of this system at 2% risk per trade is below.

Now that is a good looking equity curve, with an excellent R^2. I'm always wondering whether this has been over-optimised, but it has been tested over 12 years across 17 pairs. I suppose this is a "scalping" system since we are looking for small, consistent wins. Obviously the good news is that since trades open and close quickly, there's less chance of correlation and overlap with other trades.


  1. Hi Kevin, I enjoy a lot reading your posts due to I began my "career" like trader two years ago too, hence most of your ideas and thinkings are in my mind too.

    This system is very similar than one I´m using now, but, few months ago I changed my point of view about exits. I don´t choose now any particular R:R ratio, I prefer time exit.

    Sometimes it´s better one RR ratio, and other times it´s completely different. My idea of robustness of a system is clearer with time exits.

    I wish you a good trading week.

  2. Hi Dario, thanks for your comment.

    I've always been interested in researching time-based exits but just don't have much time to do it. I'm glad to hear that it works for you. :)

    Good luck and good trading.

  3. Hi Kevin it's Ciaran again. How is your trading going? I was curious to ask what trading costs you assume in your backtests. E.g. For EURUSD, I assume 1.5 pip spread with a 1.5 pip slippage, swaps are negligible, however this is jut my experience and the broker I use is with mt4 platform, so I may be getting a little ripped off... I'm still trading high low breakouts from daily inside bars and moving stop on lower time frame. Seems to be working ok, this year I'm down 4.18% with an initial risk of around 1.5% per trade, so it's not a big drawdown. I'm glad to see you've been making profit, I wish you the best with your trading..


    1. Hey Ciaran, it's good to hear from you. For EURUSD I assume an average spread of 1.2 pips. I use the average from this page as Pepperstone is my broker ( I haven't accounted for slippage in my costs, but I use a profit factor buffer to account for unknowns (i.e. I only use systems where PF is more than 1.25. The extra +0.25 provides a buffer for unknowns).

      Heh, 4% drawdown is nothing. Good luck!

  4. Hi Kevin. Thanks for that. I am also starting to look for diversification opportunities so I can efficiently compound gains without excess risk. Do u have any advice on this topic? Correlations between currencies are quite high, epscially majors, so this prob isn't so good. The ideal scenario is a market that is very uncorrelated but has the high liquidity and also pure tech analysis (usually these two go hand in hand). So, first thought is small cap stocks is a no no. Not sure about indices, since they gap every day from the open and close, but if the gaps are small then it's feasible. My incline is that commodities might provide a suitable diversification here. Do u have any ideas?



    1. That's a bit beyond my expertise, sorry. But to reduce correlation within the forex market, there are three tips I can think of.

      1) I try to use multiple trading systems with their own unique logic that's sufficiently different from each other. Currently I have three systems that I use on the daily chart based on a change in volatility, a one-candle pattern (pinbar) and a two-candle pattern (two-bar reversal). Most of the time, they provide an independent entry signal.

      2) Tightening the size of your trades, pip-wise, will help reduce overlap between two opened trades. Two systems with a 10 pip SL and TP are less likely to overlap than if they had a 200 pip SL and TP. With smaller-sized trades, one trade is more likely to open and close before the other is triggered. This is also true if you trade two highly-correlated pairs like the EURUSD and GPBUSD. Using a tight TP and SL, your trade on the EURUSD may open and close before your second trade is triggered on the GBPUSD.

      3) Having systems based on different time frames will also help reduce correlation. Suppose I have a system on the weekly chart and expect to hold it open for some days before it hits my TP or SL. Having a system on the 4H chart won't pose much of a problem since trades during that timeframe might only last a day. But it'll depend on the pip size of your trades (the smaller the better).

      But yeah, that's my approach, diversifying the systems I trade (or more accurately, the entries and exits) as well as my timeframes.


Note: Only a member of this blog may post a comment.