Donchian Channel Trading

Commendations on doing unusually well, then - several of the original Turtles (interviewed by Jack Schwager in his Market Wizards series of books) say exactly the opposite, and even explain in some detail why such systems typically no longer work in today’s strikingly different markets.

i think they all try and optimize it which causes problems with the system…trust me …still and will always work.

in other words, you/they think that breakout systems in general don’t work anymore?
because that’s basically the message behind saying that DC doesn’t work in today’s markets…

i don’t think so…breakouts always work as long as price moves…only way that breakout systems won’t work anymore would be if price stays eternally in a tight price range…that’s not the case…not in the past, not now and i doubt that it will in the future…

however, if a system of the past doesn’t work in today’s markets, i think it’s only because one doesn’t adjust the parameters to the current market conditions…let’s say that in times when the turtle system was invented and successful much more and longer trends happened than maybe you just have to change the parameters (go down from the longer DC/timeframe to a shorter) and the same system might work very well in the current market…

mike

I don’t think that and have not seen any of the original Turtles suggest that.

I don’t agree, sorry.

I trade breakouts myself, because all my trading’s based around support and resistance.

But that doesn’t make Donchian channels useful to me, and if you read the Schwager books I mentioned, you’ll see that they’re no longer useful to the original Turtles either, and you’ll see them explaining why much better than I can!

I find donchian channels and the turtle rules very useful cus it gives me a framework to go by…its simple and makes sense to me…but to each their own.

hey guys

i am back to trade DC 24 (1h) breakouts with 2x ATR trailing stop losses…as already backtested by Bob Bill Browne…

maybe i will also have a look at putting the initial trailing stop at the opposite DC (as breakout) and let it trail (with automatic TS) from there…not sure if we already backtested that version

mike

G’day Mike

All the best for you bro. With my improved programming skills (which are still pretty basic) I’ll try and bust out a testing EA and post some results over the next week. Still devoted to my boxes but the love for DC trading has never left. So surprised more haven’t adopted this methodology seeing the initial success we had. Let’s see if we can get things rolling again.

hi bob

thanks for your help…i only manage to backtest in excel (and get strange results sometimes), so your help is crucial…

i think i will stay with the DCs now, nothing else worked long term…so count on me…

lets see if other guys get interested in DCs…

Hi guys , im glad there is renewed interesting with DC …never miss a trend…a nice week for me …short audusd, short audcad and short nzdcad…lots of opportunities last week…i trade same strategy as the turtles using daily charts.lets keep this conversation going …very interesting…patience is key and accept DD

cheers

G’day DC lovers.

Back to get the ball rolling. Over the past week I have rebuild a bot and back tested data for 5 years and the results are in.

First the rules.
These are very simple mechanic rules and should be easy enough to follow

  • Markets Test pairs are EURUSD,EURJPY,GBPUSD,USDJPY,GOLD.

  • Timeframe 1 hour chart.

  • Variables Channel high, channel low, ATR and delta “n”. Delta “n” as introduced by the turtles is the measure of current volatilty and is expressed as a multiple of the ATR indicator. For the turltles on the daily chart that was 2x. Variables are recalculated at the start of each hour.

  • Money Management For purposes of testing I have used a fixed perecentage(2%) of a notional account balance of $100 000. Why is this important. It eliminates the effects of compounding that can rapidly distort test results. Each trade will risk only $2000 each time making it easy to compare results.

  • Entry Rules Very simple. If the Ask price breaks high than the channel high enter market long. If the Bid price breaks the channel low enter market short. Conditions Only one trade is to be active at any time in a specific market. If a current trade closes during the current bar then a new trade can only be entered into after the start of a new bar. If no order exist the next entry signal must be taken.

  • Stop Loss Stop loss on a long order is entry price minus (ATR x Delta “n”). Stop loss on a short order is entry price add (ATR x Delta “n”).

  • Trade Management I tested a few trade management systems and came back to one of two choices. To trail the stop immediately or set a break even then trail. As we are aware different instruments beat to a different drum and no singular rule can constantly be applied across the board. Testing demonstrated on of these two options was the best fit. To trail Immediately. On the start of each new hour recalculate stop loss according to the stop loss rules. If stop loss is greater than the current order stop loss on a long order or less than on a short order modify the current order to the new stop level. To set break even then trail. Here we introduce one more rule. The first adjustment of the stop can only occur once that level is equal to or greater than the entry price on a long order, equal to or less than on a short. Then trail as normal. This choice will be expressed as a true/false Boolean of trail immediately

Optimized values.
I have back tested and optimized for four variables. DC period. ATR period, Delta “n” and trail immediately. Optimization was preformed with the goal of a return 20% a year with draw down kept at a minimal. Our goal after all is account preservation. Here are the final values

Test Enviroment
Testing was done on MT4 with Birts Tick data suite with data sourced from Dukascopy. Commission was set at $7.00 round trade and current swap values applied. Orders where entered into at next available price to represent slippage on each trade. Time was set at +2GMT with daylight savings of +1 applied during US summer.

Results
EURUSD


EURJPY

GBPUSD

USDJPY

Gold

Combined into a single portfolio we get the following

Break it down by day and hour we get this

So a lot to digest there. What people want to do with this information is up to them. I’ll be here to answer any questions. As normal this information is provided for educational purposes only and I have attached the testing EA

Channel Break v1.0.ex4 (11.6 KB)

3 Likes

Thanks very much, Bob. This is certainly a really valuable and interesting contribution to the forum.

To me, it’s also a rather surprising one (I mean its contents, of course, not the fact that you’ve kindly produced it!).

Most surprising of all, perhaps, is the enormous discrepancy between Mondays/Fridays and the midweek days. I’d have expected some, but nothing like this at all.

What’s your own feeling about that? Looks like (in this context, at least) the people who do most of their trading on midweek days are perhaps right?

1 Like

hi travis

great to have you here for discussing DCs…
i also had a nice week, didn’t start to good but thursday and friday were great…
a little reminder, i trade DC 24 on 1h charts and every combination of the 8 major currencies (AUD, CAD, CHF, EUR, GBP, JPY, NZD, USD = 28 pairs).

hi bob,

great work, as usual! thanks a lot…

i will try to find the time to test all my traded pairs too…but as you know i only can backtest in Excel (with 1min data from HistData.com | Download Free Forex Data)…so my results will not be exactly the same, but should at least point in the same direction…

Lets look a bit harder into this. First Lets breakdown Mondays.

I think it’s clear what is happening here. False breakouts at weeks open. Therefor first new rule based on statistical analysis. Don’t trade the first three hours after market open. I’ll go as far and say don’t bother until at least 7am.

Also of note is that the system breaks down at London close (4pm)

Breaking down Friday

This time there appears two distinct no trade periods. Prior to London open and the London/New York overlay.
Whats going on here I don’t know. Backing and filling before London and balancing the books during the overlay. Regardless it seems to indicate some underlying volatility that can whip the market around and take out stops.

Just some thoughts

1 Like

the turtle system is a long term trend following system…whether it goes down after the london open or the asian open doesnt really matter.

cheers

Thats the turtle system bro. Just one method of donchian channel trading. What we’re looking at is hourly charts so it does matter in this instance.

cadchf broke the 55 day low so one could put a sell stop at .7694 and put a stop at 2 times the atr20 and then use the 20 day high as your exit…easy peasy

[quote="_bob, post:408, topic:71676"]
G’day DC lovers.

Back to get the ball rolling. Over the past week I have rebuild a bot and back tested data for 5 years and the results are in.

First the rules.
These are very simple mechanic rules and should be easy enough to follow

Markets Test pairs are EURUSD,EURJPY,GBPUSD,USDJPY,GOLD.
Timeframe 1 hour chart.
Variables Channel high, channel low, ATR and delta “n”. Delta “n” as introduced by the turtles is the measure of current volatilty and is expressed as a multiple of the ATR indicator. For the turltles on the daily chart that was 2x. Variables are recalculated at the start of each hour.
Money Management For purposes of testing I have used a fixed perecentage(2%) of a notional account balance of $100 000. Why is this important. It eliminates the effects of compounding that can rapidly distort test results. Each trade will risk only $2000 each time making it easy to compare results.
Entry Rules Very simple. If the Ask price breaks high than the channel high enter market long. If the Bid price breaks the channel low enter market short. Conditions Only one trade is to be active at any time in a specific market. If a current trade closes during the current bar then a new trade can only be entered into after the start of a new bar. If no order exist the next entry signal must be taken.
Stop Loss Stop loss on a long order is entry price minus (ATR x Delta “n”). Stop loss on a short order is entry price add (ATR x Delta “n”).
Trade Management I tested a few trade management systems and came back to one of two choices. To trail the stop immediately or set a break even then trail. As we are aware different instruments beat to a different drum and no singular rule can constantly be applied across the board. Testing demonstrated on of these two options was the best fit. To trail Immediately. On the start of each new hour recalculate stop loss according to the stop loss rules. If stop loss is greater than the current order stop loss on a long order or less than on a short order modify the current order to the new stop level. To set break even then trail. Here we introduce one more rule. The first adjustment of the stop can only occur once that level is equal to or greater than the entry price on a long order, equal to or less than on a short. Then trail as normal. This choice will be expressed as a true/false Boolean of trail immediately

Optimized values.
I have back tested and optimized for four variables. DC period. ATR period, Delta “n” and trail immediately. Optimization was preformed with the goal of a return 20% a year with draw down kept at a minimal. Our goal after all is account preservation. Here are the final values
[/q

very great job… Ill start to follow on paper how its work … seems well… I use the 200 EMA for the direction of trade …
Hope this tread is still open

Hi, I was going to do some testing on a demo account and the ea is not working as expected. there was a breakout in the GBPUSD last night and the trade did not enter. nothing was added to the journal/ea tabs at all.