Trend following system is dead | Page 8 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Long Term Forex Trend Following System 5 replies

Is Grid Trading (combined trend following and counter trend) Profitable? 3 replies

Trend following or trend fading? 8 replies

SMA - Bollinger long term trend following system 7 replies

PBTrend A Daily Trend Following System 13 replies

Attachments: Trend following system is dead
Exit Attachments

Trend following system is dead

  • #141
  • Mar 3, 2023 8:57am Mar 3, 2023 8:57am
Quoting OutThere
I do know you trade only swap positive pairs in case of a long haul.
//-----

i intentionally chose this day from peytons account maybe last week because you can see the swap the day before those trades and the day after.....

swap really only accounts for 1 or 2 pips..... one single trade could easily cover the days swap.....

you know this but some might not, her lot size is small and targets are reasonable...... oanda's rules here in the states forces us to have a maximum of 1 trade per lot size if we use targets..... so there might be a 0.01, 0.02, 0.03 and so on..... that is merely following oanda's mandatory rule.....

she is far more interested in soccer nowdays...... nothing wrong with that.....h
//-----

Attached Image (click to enlarge)
Click to Enlarge Name: swap.png Size: 45 KB
to trade and code, keep both simple... no call to impress....h
  • #142
  • Mar 3, 2023 8:17pm Mar 3, 2023 8:17pm
Quoting hayseed
{quote} //----- i intentionally chose this day from peytons account maybe last week because you can see the swap the day before those trades and the day after..... swap really only accounts for 1 or 2 pips..... one single trade could easily cover the days swap..... you know this but some might not, her lot size is small and targets are reasonable...... oanda's rules here in the states forces us to have a maximum of 1 trade per lot size if we use targets..... so there might be a 0.01, 0.02, 0.03 and so on..... that is merely following oanda's mandatory...
Hello Hayseed. I have been studying these trades you took. I assume these were all trend trades, but I am having a hard time understanding why you took some of them and what time frame and what triggers or thought process you used with the initial trade was made for that pair. Would it be possible for you to post a screen shot of a couple of these with some descriptions why you took the trade?

THANK YOU!
  • #143
  • Mar 3, 2023 10:03pm Mar 3, 2023 10:03pm
Trend-following will always exist, you have to:

1) Reduce noise.
2) Pick the right instruments for trend-trading.
  • #144
  • Mar 4, 2023 1:07am Mar 4, 2023 1:07am
Quoting Abade
Trend-following will always exist, you have to: 1) Reduce noise. 2) Pick the right instruments for trend-trading.
Hi Abade

What noise?
What's the right instrument?

Looking at history to provide these answers is a fools opinion. Each moment in time is unique, a move may have been started by just one British banker who went Short last week in a large trade at every double top. This week you see a double top and say to yourself "Its going down like last week so I will Sell". But price this time when up - Why?

Because this same British banker called Roy who always goes short on a double top had a heart attack on Sunday afternoon and never traded in the new week.
My point is the markets move because of human's, every setup contains different people / bankers who may decide to trade something different this week.

Every move is unique and therefore the action is random in natural.
Which gets to my point in disagreeing with your statement.

Noise is Roy, its Larry, its Sue, its Abade and every moment is unique therefore you can't predict Trend when there is always a different combination of traders working, same goes for noise.

Historical action does not tell us shit, you just have to be sitting on a pair and looking for the right action.

PS. Roy was in hospital for 2 weeks and will be back trading in April

That will be another unique trading month, who knows what will happen.
Trading thin liquidity at the boundary of the charts
  • #145
  • Edited 9:40am Mar 4, 2023 9:19am | Edited 9:40am
Quoting hayseed
{quote} //----- i intentionally chose this day from peytons account maybe last week because you can see the swap the day before those trades and the day after..... swap really only accounts for 1 or 2 pips..... one single trade could easily cover the days swap..... you know this but some might not, her lot size is small and targets are reasonable...... oanda's rules here in the states forces us to have a maximum of 1 trade per lot size if we use targets..... so there might be a 0.01, 0.02, 0.03 and so on..... that is merely following oanda's mandatory...
A lot of things you are saying keep bringing back the memories of trading manual with FXtrade.
  • #146
  • Mar 4, 2023 9:31am Mar 4, 2023 9:31am
Quoting RickM
{quote} Hi Abade What noise? What's the right instrument? Looking at history to provide these answers is a fools opinion. Each moment in time is unique, a move may have been started by just one British banker who went Short last week in a large trade at every double top. This week you see a double top and say to yourself "Its going down like last week so I will Sell". But price this time when up - Why? Because this same British banker called Roy who always goes short on a double top had a heart attack on Sunday afternoon and never traded in the...
Amazing reply , its so easy to backtest trend following systems on previous 'trends' but there is no reason why that exact trend will EVER happen again, as i believe all trend are made from macro causes and/or largest participants needing to aggressively get rid/adjust their exposure etc. Both of these situations will have people trying to stop that same reaction from happening in the future. Both the smartest participants not being shocked by similar macro events in the future and the largest having more control over their exposure at any one point in time after being f*cked and causing a trend against them. This is my philosophy of why earlier in the thread it says that once a 'trend' has formed, its over.
Seeking Fair Value
  • #147
  • Mar 4, 2023 10:30am Mar 4, 2023 10:30am
Quoting RickM
{quote} Hi Abade What noise? What's the right instrument? Looking at history to provide these answers is a fools opinion. Each moment in time is unique, a move may have been started by just one British banker who went Short last week in a large trade at every double top. This week you see a double top and say to yourself "Its going down like last week so I will Sell". But price this time when up - Why? Because this same British banker called Roy who always goes short on a double top had a heart attack on Sunday afternoon and never traded in the...
If you add the question: What is volume in FX, your answer will fit that question as well. I am not thoroughly familiar with your methods but I know volume is a cornerstone of it.
  • #148
  • Mar 4, 2023 4:24pm Mar 4, 2023 4:24pm
Hi Rick! I highly disagree with these two statements:

Quoting RickM
{quote} you can't predict Trend when there is always a different combination of traders working, same goes for noise.
Quoting RickM
{quote} Looking at history to provide these answers is a fools opinion
Even if this other statement is TRUE:

Quoting RickM
{quote} Each moment in time is unique...
My reasoning:

1) Nobody is trying to "predict" noise, that's impossible by definition. By "noise reduction" I am referring to an attempt to diminish MOST of the smaller price moves and enhance the bigger ones by using different mathematical formulas which is absolutely POSSIBLE and involves no "prediction" (I show an example below).
2) You can't "predict" a future trend that's true, but that doesn't mean this phenomenon cannot be exploited or that you require knowledge of each individual participant at that precise moment to make bank, that's just a ludicrous thing to say.
3) You say that looking at history to provide answers to these questions (what noise, what instrument) is a fools opinion, but you have no problem using historical data to optimize the "frequency" of M1 price series in this post (blows my mind):

Attached Image (click to enlarge)
Click to Enlarge Name: RickM_Using_Historical_Data.png Size: 19 KB

4) Other experienced algorithmic traders use historical data to pre-select the instruments they want to trade for each specific system and trading logic. To me that makes more sense than looking for a single universal logic-type that works consistently in every instrument, time-frame, everytime; as a matter of fact that's precisely what a fool/scammer/gambler would suggest.

A small demonstration on the noise reduction concept. This picture shows the overall performance of a trend following system that I coded myself. It includes several parameter permutations over a 4 month period. It ranks performance based on sharpe ratio (MetaQuotes formula):

Attached Image (click to enlarge)
Click to Enlarge Name: Without_Noise_Reduction.png Size: 14 KB

This is the same trend following system, period, parameter ranges and instrument, but I enabled the "noise reduction" features:

Attached Image (click to enlarge)
Click to Enlarge Name: Noise_Reduction.png Size: 14 KB

Which one you think gives the best out-of-sample performance? Yes! The one with noise reduction!

But won't stop other traders if they want to use one of these instead of research and empirical evidence :

Attached Image (click to enlarge)
Click to Enlarge Name: 81gVpFjwDjS._AC_SL1500_.jpg Size: 300 KB

Peace!
  • #149
  • Mar 4, 2023 4:39pm Mar 4, 2023 4:39pm
Quoting makingmoves
{quote} Amazing reply , its so easy to backtest trend following systems on previous 'trends' but there is no reason why that exact trend will EVER happen again, as i believe all trend are made from macro causes and/or largest participants needing to aggressively get rid/adjust their exposure etc. Both of these situations will have people trying to stop that same reaction from happening in the future. Both the smartest participants not being shocked by similar macro events in the future and the largest having more control over their exposure at any...
The matter of fact is that you DON'T actually need an identical trend to form to make money from it, that's an absolute fallacy.

BTW chart patterns are terrible! You need something that is adaptable and will let you generalize future price movements without curve-fitting (fitting to noise)

Cheers!
  • #150
  • Mar 4, 2023 4:41pm Mar 4, 2023 4:41pm
"I am the trend."
  • #151
  • Mar 4, 2023 6:25pm Mar 4, 2023 6:25pm
[quote=CashBox;14347612]"I am the trend."[/quote

I have always found your charts interesting.

Simple chart from my Archives .....just for interest

R
Attached Image (click to enlarge)
Click to Enlarge Name: XAUUSDH1.png Size: 37 KB
Ronnie Renko
  • #152
  • Mar 4, 2023 8:27pm Mar 4, 2023 8:27pm
Quoting OutThere
{quote} If you add the question: What is volume in FX, your answer will fit that question as well. I am not thoroughly familiar with your methods but I know volume is a cornerstone of it.
If you ask the question "what is volume in FX", you just increase advertising revenue for Forex factory.
Let's not do that.

My strategy is as simple as it gets. Using volume from a centralized exchange like CME, I just trade momentum runs via market orders on the DOM while ensuring its attacking limit order imbalance.
The new addition for me is to use Price Squawk while trading which is sound signals of increasing Bids & Asks at the same time.
I believe the brain can react quicker to sounds than visuals.
Trading thin liquidity at the boundary of the charts
  • #153
  • Edited 9:09pm Mar 4, 2023 8:42pm | Edited 9:09pm
Quoting Abade
Hi Rick! I highly disagree with these two statements: {quote} {quote} Even if this other statement is TRUE: {quote} My reasoning: 1) Nobody is trying to "predict" noise, that's impossible by definition. By "noise reduction" I am referring to an attempt to diminish MOST of the smaller price moves and enhance the bigger ones by using different mathematical formulas which is absolutely POSSIBLE and involves no "prediction" (I show an example below). 2) You can't "predict" a future trend that's true, but that doesn't mean this phenomenon cannot be exploited...
Hi Adabe

I know where you are coming from - been there myself.

I traded EA's with a group of 15 traders from 2016 to 2020 where we developed over 100 Algo's to trade Forex markets. We found a few things in this process.

1.. Some pairs favoured Trends like many GBP pairs ie. GBAUD GBPUSD GBPJPY GBPNZD
2.. Some pairs favoured reversal strategies from AU & EU pairs. ie. AUDNZD EURJPY EURCAD AUDUSD AUDJPY
3.. Some pairs are dogs ie USDJPY

4.. Back testing is a failed theory with only a small outcome of success
5.. Only forward testing offers any value and then not much.

6.. 60% of volume executed on IC Markets comes from Algo traders. The losing rate has apparently got worst. In fact if you trade a Martingale strategy - you will get better rates if asked. ......... WHY?

That was just my experience in the Algo trading world, which is why I burned my EA's in a bonfire.

Maybe your'e better than us 15 traders - hope so.

Looking at your testing which looks great, have you considered Monte carlo analysis because the few Algo traders i found that were successful always pointing me in that direction.
Attached Image (click to enlarge)
Click to Enlarge Name: mc.png Size: 214 KB
Trading thin liquidity at the boundary of the charts
  • #154
  • Mar 4, 2023 9:10pm Mar 4, 2023 9:10pm
Quoting Abade
BTW chart patterns are terrible! You need something that is adaptable and will let you generalize future price movements without curve-fitting (fitting to noise) Cheers!
Agree
Know what you trade
  • #155
  • Mar 4, 2023 10:46pm Mar 4, 2023 10:46pm
[quote=Ced74;14347660]
Quoting CashBox
"I am the trend."[/quote I have always found your charts interesting. Simple chart from my Archives .....just for interest R {image}
Ced74, you always have something fun to try to figure out! Love it.

I figured it was Saturday so would poke some fun at trading and lighten things up a little. Thanks for joining me.

Our electric has been out for 30 hours...telling us not to expect it to be back on for another 48 hours. Thankful this place came with a permanent whole house generator and fiber internet that doesn't go down!
  • #156
  • Edited 11:22pm Mar 4, 2023 10:59pm | Edited 11:22pm
Quoting RickM
{quote} Hi Adabe I know where you are coming from - been there myself. I traded EA's with a group of 15 traders from 2016 to 2020 where we developed over 100 Algo's to trade Forex markets. We found a few things in this process. 1.. Some pairs favoured Trends like many GBP pairs ie. GBAUD GBPUSD GBPJPY GBPNZD 2.. Some pairs favoured reversal strategies from AU & EU pairs. ie. AUDNZD EURJPY EURCAD AUDUSD AUDJPY 3.. Some pairs are dogs ie USDJPY 4.. Back testing is a failed theory with only a small outcome of success 5.. Only forward testing offers...
My development process is a bit different from the stuff that is found online, perhaps one day I will share the whole process. In a pinch, this is what I do for trend-following systems:

1) I only create single position stop-and-reverse systems. I discovered that including fixed or variable targets (stop-loss and/or take-profit) from the start tends to produce very good looking backtest but they ALWAYS show very poor performance going forward because they are overfitted to noise (a fluke), so I don't include them at this point.
2) I run a single stage optimization (in-sample & out-sample) and evaluate the overall performance of a full series of parameter ranges by using various metrics. Sharpe ratio works great for visualization and it includes profit, drawdown and sample size in its calculation, so this is my usual pick.
3) I take three things in consideration when evaluating the system at this point:
______a) If most in-sample runs fall in a "positive" area (above the metric breakeven), I consider it to be a good sign of adaptability (the image I provided shows this).
______b) If the best performing parameters are inside a "stable" zone I also consider it to be a good sign.
______c) It must pass an out-of-sample test by at least one of these metrics: Profit Factor, Minimal Drawdown, Sharpe Ratio or Recovery Factor.
4) I repeat the first three steps with the rest of my algo list and rank them from best to worse based on a, b and c.
5) The best performing algorithms get tested using a multi-stage walk forward analysis and several months of unseen data. WFA grants a very large pool of trades for out-of-sample evaluation and it's also an indispensable tool for future adaptability.
6) I combine the algos that survived WFA in a basket of systems with very low risk.
7) Algos get re-optimized each month, because trading conditions and volatility changes with time.

I don't use monte-carlo analysis, maybe I should. However, the current process by itself is producing very good results when the systems are combined in a porfolio, even if a third of them start showing poor performance (I usually shut them after a week). In conclusion, the secret sauce for trend-following systems (IMHO) is:

- Using noise reduction features.
- Avoiding systems with too many parts and rules.
- Evaluate a wide range of parameter variations instead of single sets.
- Using WFA for proper out-of-sample testing and adaptability.
- Treating each instrument as a different "animal".
- Re-adapt to changing market conditions.
- Trade in a basket so winning systems "hedge" the losing ones.

(Phew that was a lot of text)

Also, martingale is a "no-no" in my book, definitely.
  • #157
  • Mar 4, 2023 11:18pm Mar 4, 2023 11:18pm
Quoting forexdave1
but I am having a hard time understanding why you took some of them and what time frame and what triggers or thought process you used with the initial trade was made for that pair. Would it be possible for you to post a screen shot of a couple of these with some descriptions why you took the trade?
//-----

hey dave...... many have said that...... really not sure how to explain..... will try......

it is almost impossible for me to remember the exact reason unless the trades have descriptive comments telling me the reason...... then i can export to excel and see...... similar to seen here.....

imagine you had indicators, scripts or ea's that could give you every imaginable buy/sell signal on every pair on every time frame....... these could send alerts, push notifications and export to excel..... similar to this one...... that's what i do..... really don't look at charts much.....

in the gbpjpy chart below......, the daily trend was down so i'm leaning towards shorts..... the 240 jfatl and zigzag countertrend up counts were nearing average length so that countertrend could possibly stall...... the history indicator is drawing the closed trades and doing the math.....
//-----

the other chart is a common example on multi time confirmation...... buy trades are entered when there is no red on any timeframe..... sells are entered when there is no lime on any timeframe..... this raises the win/loss ratio high......

i've said many times, the actual signal is really not important to me...... only the trend......h
//------

Attached Image (click to enlarge)
Click to Enlarge Name: gbpjpy-h4-oanda-division1.png Size: 23 KB

//------

Attached Image (click to enlarge)
Click to Enlarge Name: eurjpy-m5-oanda-division1-no-red.png Size: 13 KB
to trade and code, keep both simple... no call to impress....h
  • #158
  • Mar 4, 2023 11:35pm Mar 4, 2023 11:35pm
Quoting forexdave1
but I am having a hard time understanding why you took some of them and what time frame and what triggers or thought process you used with the initial trade was made for that pair.
//-----

what follows a red 14.......

the initial trade usually starts with a 0 or 1....... the idea is, if the tsr was rising yesterday but fell today, the count would be a red 0...... what follows a red 0......

counts triggers......h
//------

Inserted Video
to trade and code, keep both simple... no call to impress....h
  • #159
  • Mar 5, 2023 12:03am Mar 5, 2023 12:03am
Quoting Abade
{quote} My development process is a bit different from the stuff that is found online, perhaps one day I will share the whole process. In a pinch, this is what I do for trend-following systems: 1) I only create single position stop-and-reverse systems. I discovered that including fixed or variable targets (stop-loss and/or take-profit) from the start tends to produce very good looking backtest but they ALWAYS show very poor performance going forward because they are overfitted to noise (a fluke), so I don't include them at this point. 2) I run a...
Excellent work Abade

I agree on a few of your findings - run a basket of Algo's that are all simple with few rules and use some EA's more as a hedge for the whole basket rather than a winning EA in themselves.

Nice work
Trading thin liquidity at the boundary of the charts
  • #160
  • Mar 5, 2023 12:23am Mar 5, 2023 12:23am
A trade out of exact market trend goes bust for sure. So, follow indicators like RSI, trend lines, moving average and Bollinger bands to identify the exact market trend.
0 traders viewing now