Mostly the time delay in waiting for good setups to occur. Also, my personality is more suited to a higher winning percentage in my trades, so I am spending more time on robots and discretionary systems with 75%+ winning percentages.
I won’t forget about donchian and will continue to monitor this thread, but at the moment it isn’t my main focus.
I trade it different than letting it run its course. To many good trades can turn to losers. thats why % I think is so low.
Once Im in trade I use a combo of ATR and ADR. for my stops.
If it lets me ill ride big trend if not I get out and take my profit or cut loss short well before it reverses to new high/low.
As long as you don’t miss the long trend by getting stopped out on a small reversal, this might help to further decrease losses…
Can you explain in more detail how you use ATR and ADR?
Mike
EUR/HUF broke above the MDMA this week so I went long. My stop is below the four week low (I am trading the four week channel with this one.)
-Adrian
Well the results are in and it doesn’t look good for the USDJPY. First the 55 day breakout
And then the 20 day breakout
And finally the 50 day break with 10 day trailing
The 20 day break seems a failure period. And it would have been many a tuff year on the 55 day break. But I suppose what it does demonstrate is that if one stuck to the rules then the system works.
I’ll run a few optimizing runs, see if it can’t be tweak performance. Suspect ATR might be working against us here.
Here’s the new bot and as always with source code. If you had experimented with the earlier ones you would of noticed it didn’t trade XXXJPY pairs. This one can.
Turtle Trader Build 11.zip (18.4 KB)
And if you not up to anything tonight kick back, melt into the soul of you inner universe an explore the outer to the tunes at Raw FM - Dance Floor Radio | Australia’s Most Dynamic Dance Music Radio Station.
Well it’s not pretty but still it’s a tweak. Half your risk if following the turtles on the USDJPY and try a 60 day breakout with 15 day trailing. Finally apply an ATR of 32. Here’s the result
Compared to the 55/20 setting
Got a couple of things to share tonight. First and most exciting. For the sake of running a back test I took the optimized settings I had discovered for the EURUSD pair on the 1 hr chart and simply applied them to the USDJPY. Here’s the walk forward results. Speak for itself.
Potential of $160 000 return on a $10 000 investment. Draw down @ 15% on a 1%risk.
Next I’m going to share my own personal bot that I use on the 4 Hr chart and its methodology. I had hope a senior member would tweak it for me but he either had better things to do or liked it so much he kept it for himself. I hope its the first and he might come back to it and offer his view.
Ok so its a donchian channel break strategy on the 4 hr chart. But instead of using ATR I use Fib levels to define an inside channel. The difference between the donchian channels and the inside channel is then used to determine entry conditions and the variables for money management.
Entry rule. After a break and if no order exist, the price must first retrace back inside the channel. Once the price has retraced a limit order (buystops and sellstops) is then placed at the donchain channel level. The order then trails at the donchain channel level until it gets filled and the variables recalculated every 4 hrs to reflect current market conditions until it gets filled. If an limit order exist in the opposite direction it is immediately canceled.
Money Management. Risk is what you want it to be. I’m trading this in a basket of 10 pairs and have risk set at 0.5%. For the test results I’ve set risk at 1% to puff up my ego :). This system utilizes a fixed SL and TP. You’ll only ever have one trade open per pair. The analysis is either right or wrong. Each trade is allowed to trade out. Why, because that is what history has taught us. Doesn’t mean there ain’t better rules for Money Management. Just means I’m not clever enough to figure them out.
SL and TP are based on “n” - volatility. The turtles and many others for good reason used ATR as a measure of “n”. I use the upper and lower levels of this inside channel which at the end of the day are nothing more than fib levels. So my valve of “n” is the difference between the upper donchain channel and the upper inside channel for long trades and the lower channels for short. SL and TP are then simple multiples of “n” So if “N” = 50 pips, then a SL of multiple 0.5 would be 25 pips, A multiple of 1 is 50 pips and a multiple of 2 represents a SL of 100 pips. Same for TP.
Now varying these multiples is an important factor in this system and bots operations. First and fore most it affects the risk:reward ratio. Risk:Reward is calculated as the take profit multiplier divided by the SL multiplier. So a TP multiplier of 3 and SL of 1 yields a R:R of 1:3. So to a TP multiplier of 1.5 and SL of 0.5. The other important thing to remember is that “n” is a fib level. So an upper inside channel level at the 40 fib mark with a multiplier of 0.5 yields the same SL in pips as the 20 fib level at SL multiplier of 1. The fib level is important because the retracement of price back inside the channel triggers the end of a trend so time for the next trade. Too shallow and too many false entry signals get generated. Too deep and opportunities are missed. It’s a delicate balance but one luckily we don’t have to worry about. The bot does all the hard work. And it doesn’t vary that much between pairs. We just need to understand it.
So about the bot. The bot is designed to trade long and short independently. One of the secrets of using a bot is knowing when to turn it off. By having the variables coded independently it allows for switching between long and short trades.Also experience has demonstrated to me that pairs behave differently in either direction. Independent code allows the bot to be optimized specifically to the pair and direction. Again the reality is the difference in these variable values is quite small. I currently use it across six USD pairs, 3 JPY pairs and the EURGBP. In order to use the bot you will also need my channel indicator and tonight I’m feeling generous. But you can’t get them until the next post.
So the reults. First the EURUSD
And next the USDJPY
I’ll let you guys ask any questions.
Now why do we do all this. Well we want a bit of diversification in our markets. Now we have three bots that can potential trade 10 pairs. Thats 30 strategies running at once. This in turn allows a smoothing of our growing equity curve, reduced risk per strategy, reduce drawdown and maintain yield. To demonstrate this I created a portfolio of all 3 bots used so far on the two reference pair.
Heres the equity chart.
As a portfolio, using just these six setting together greatly reduces our draw down and multiplies our yield. Adding more pairs in would allow a further reducing of risk yet maintain the equity curve.
And now my gift. The three bots with source code, two indicators, presets, back test results and portfolio analysis that I have reference so far I have packaged up into a zip file for your use. All I ask in return is that there are plenty more skilled in programming and talented as traders than me. If this is you and you can tweak these bots please feel free too but report back and share here.
Enjoy the week end all
Bob
Turtle trader.zip (3.76 MB)
I use the ATR for my entries. I basically take the ATR and mulyiply by .20 and I use that for entries. And I add that to the top or bottom of the channel instead of just a pip above channel.
Hopefully this will keep me out of a few trades i shouldnt be in. Dont know if that s gonna be the final way I do it though but thats how I have been using it when I trade with doc.
I still use ichimuko because I have been using that for longer than I want to admit lol
I still havent come up with a plan with the ATR yet, still trial and error there.
A nice return and a very acceptable drawdown…
I am thinking of trading my two favourite DCs (20 for entries and 10 for trailing SL) on 3 different TFs to take advantage of short, medium and long term trends and diversify risk…
Still need to figure out the best combination…
For intraday i would take 5min/15min or 30min…then maybe 1h or 2h for trades that last a couple of days…and for longer trends 4h/6h/8h or daily…
For intraday trades i would put limit orders only once a day for every pair/commodity…on the other TFs trailing limit orders until they get hit…
Hopefully that way the different trends (short, medium, long term) can compensate wins and losses on each pair and on bad days/weeks/months preserve the capital and on good ones grow the account…
Thoughts and comments appreciated…
Mike
Just my view bro (you trade the way you do) but I’m seeing 24 as a re-occuring breakout on the 1 hr chart. 30 / 36 on the 4 hr chart. 50/60 seems to be of value on the daily. As for trailing, 6/9/12 seem a good combo across all timeframes.
Again just my view but excluding the daily charts, there could well be a better practice for managing exits. Doesn’t mean T/S isn’t valid. I’m going to continue development with a T/S for the time being.
I think if one chooses a particular price range, that is curve fitted system trading. You must trade multiple channels to avoid the troubles associated with trading a curve fitted system. System diversification.
-Adrian
isn’t the same settings on different TFs exactly that?
entries on DC 20 on TF 15min is not the same as entries on DC 20 on TF 1h nor on TF 4h…or do i miss something?
mike
thanks for your view…
when you say 6/9/12 seem good trailing SLs for all timeframes, would you use only one of them or would you start trailing at 12 and as the trade goes in your favour close the gap by trailing tighter switching to 9 and 6 at the end…could be a way to protect profits…
mike
At the end of the day it’s up to you. However I do see this step reduction as a very viable exit strategy. From my understanding there are many traders that use time based trailing stops. And if you follow me then time is probably the best filter we have. I like the idea very much. I’ll tweak a bot and run a few tests. Might take a week or two. My programming skills are very basic.
Back testing is a very slow methodical task. To run one test over 10 years takes 10 min in non visual mode. Hours visually. You start in non visual but once you have a set of results you’re happy with the fun starts. That involves running a visual back test, taking lots of screen shots and analyzing.
But this thread has revived my interest in donchian channels, tick chart trading has become boring although its still my staple diet. And as old m8 Adrian says, it’s about system diversification.
i like the idea of a reducing trailing stop…and with a time component it could even be better…
how could we implement time based stops in combination with a reduction of the trailing stop? every new candle? or weekly?
this can be a nice thought/discussion process…
No. We are agreeing here. The 20 hour channel and the 80 hour channel will likely profit at different times and produce different entries and exits.
-Adrian
