How Do You Determine the Long-term Trend?

[MightyMouse 12]

And all trends, even those correctly identified must eventually reverse.

 

As things currently stand, I'm only usually wrong once at the end of the trend for this reason. Add in a few losses caused by premature entries during the trend, and I'm correct (profitable) about 7-8 times out of 10 within the typical S&P trend as I define it. I have now become quite accustomed to sustaining a significant loss after a string of smaller winning trades. So all this is fine, as things go, but problems occur when my determination of trend is incorrect, as then the inevitable losses associated with trend change often become sequential. EVEnly spread they aren't an issue at all, but clustered together I fear that sooner or later they may un-nerve me.

 

BlueHorseshoe

 

My record is somewhat opposite. I end up catching the direction, but it usually comes after multiple attempts at trying to enter, or more accurately, after several small losses. As such, my win rate is near 35%.

 

The long term trend defies logic while it is happening, which is why it makes it so difficult to spot or difficult to ride. I find that stepping back and simply looking at the chart and determining its visual slope over the last 200 bars ( lower left corner rising to upper right for an uptrend, upper left to lower right for a downtrend, etc) will give you the best assessment of the long term trend. Studying candles or bars is the surest way to let myopia set in and get out early or, worse, call a top or bottom and try to reverse.

 

I cannot lie. There are few things in life that are better than catching a trend or move and riding it. Patience, discipline, and confidence.

[Obsidian 10]

Hi Obsidian,

 

I have never really looked at Ichimoku. I seem to remember deciding it was very complicated at some stage, and I prefer simple, but if what SIUYA is saying and it is just the middle of a Donchian channel then that's straightforward enough. Thanks for the suggestion - definitely something I'll investigate further!

 

BlueHorseshoe

 

just use the clouds part...i.e. trend is up when prices are above the cloud, down when prices are below the cloud and flat when they are in the cloud...

 

here is a weekly s&p chart..attached clouds and 200sma

[BlueHorseshoe 10]

My record is somewhat opposite. I end up catching the direction, but it usually comes after multiple attempts at trying to enter, or more accurately, after several small losses. As such, my win rate is near 35%.

 

As I'm sure you're aware, the markets have a habit of evening things out. So the fact that I have more winning trades than losing trades is simply a function of the fact that I have small wins and large losses. In other words, nothing for me to brag about . . .

 

One of the advantages of how you trade, I imagine, is that with every successive entry signal following a losing one the market will have moved further in the direction you anticipate, therefore further confirming the trend you wish to enter.

 

BlueHorseshoe

[MightyMouse 12]

As I'm sure you're aware, the markets have a habit of evening things out. So the fact that I have more winning trades than losing trades is simply a function of the fact that I have small wins and large losses. In other words, nothing for me to brag about . . .

 

One of the advantages of how you trade, I imagine, is that with every successive entry signal following a losing one the market will have moved further in the direction you anticipate, therefore further confirming the trend you wish to enter.

 

BlueHorseshoe

 

Sometimes I end up getting in at a worse position. Sometimes I end up getting in better positioned in the direction of the major trend. As long as the direction remains the same I will keep trying to get positioned. I can and have gotten hacked up trying to get positioned. No reason to pretend that I haven't; however, overall my winners are big enough to cover all the crumbs I give up.

[SIUYA 11]

One of the advantages of how you trade, I imagine, is that with every successive entry signal following a losing one the market will have moved further in the direction you anticipate, therefore further confirming the trend you wish to enter.

 

not for eveyone, and potentially dangerous, but just as food for thought after you determine a trend and applicable only to some styles.......

 

people have mucked around with variations of this, by missing the first signal after a winner or other such filters.

there is also the idea that you increase your size with more losers.....

eg; trade 1 size =1, lose....trade 2 size=2....lose, trade 3 size=3...win

 

You need to make sure you dont pyramid and add to the losers you need to still cut them.

downside to this is if you get it right early your size is small and has to start very small at the beginning! but if using this to build bigger longer term positions there is probably some ideas here.

 

The other thing that needs to be remembered if you only go one way - ie; with the direction of the trend - you need to ensure you are right otherwise it might be a slow death unless your entries are very good.

[BlueHorseshoe 10]

You need to make sure you dont pyramid and add to the losers you need to still cut them.

downside to this is if you get it right early your size is small and has to start very small at the beginning! but if using this to build bigger longer term positions there is probably some ideas here.

 

The other thing that needs to be remembered if you only go one way - ie; with the direction of the trend - you need to ensure you are right otherwise it might be a slow death unless your entries are very good.

 

Thanks - some very useful ideas there.

 

Unfortunately my entries aren't especially great - most of the time I will sit through several days of adverse excursion. I have a list as long as my arm of ways that this can be avoided and my entries improved; unfortunately every one of them significantly reduces opportunity and net profit. I decided that I'd prefer regular opportunity and greater net profit to good entries, and that I could control the effects of adverse excursions through position sizing.

 

I'm just playing around with code to try and determine what happens if I increase the number of contracts traded with successive losing trades (as with a martingale position sizing method), but simultaneously decrease the stop loss per contract to limit risk to the same percentage of equity (so as to avoid an actual 'doubling down' scenario). For example:

 

Losing Trade 1: 1 contract, SL $200 per contract (total risk $200).

Losing Trade 2: 2 contracts, SL $100 per contract (total risk $200).

Losing Trade 3: 3 contracts, SL $66 per contract (total risk $200).

 

So with the fourth trade, assuming it was exited for a profit, and assuming a stop loss of $50 per contract was not hit, then the position size would be four contracts. If the profit target remains constant at $200 for each trade, then the above scenario yields a $200 profit, as opposed to a $400 trading a fixed 1 contract with a fixed $200 stop.

 

Of course, the more the stop is reduced (assuming the outcome of each trade is independent of those before it), the greater the probability that it will be hit before the target is elected. The markets tend to be very efficient in this way. What I am really interested in is whether strings of sequential losing trades within the same market analysis structure are indeed independent, and whether they are dependent to such an extent that an edge can be gained, whereby the probabilty of a stop-out does not actually increase as the stop-loss per contract is reduced.

 

Not that any of that has anything to do with determining trend!!!

 

BlueHorseshoe

[BlueHorseshoe 10]

Back on the subject of trend . . .

 

And in answer to my own question, my strategy now considers the ES to be down-trending. Assuming the market holds above about 1316 into the close, then I will short it. If it holds above 1320 I'll be happier. The low volume today is also encouraging, as I interpret this as lack of buying interest as we drift higher, although this doesn't form any part of an actual decision making process for me.

 

Not that I intend to make a habit of posting live trades . . .

 

BlueHorseshoe

[SIUYA 11]

I'm just playing around with code to try and determine what happens if I increase the number of contracts traded with successive losing trades (as with a martingale position sizing method), but simultaneously decrease the stop loss per contract to limit risk to the same percentage of equity (so as to avoid an actual 'doubling down' scenario). For example:

 

Losing Trade 1: 1 contract, SL $200 per contract (total risk $200).

Losing Trade 2: 2 contracts, SL $100 per contract (total risk $200).

Losing Trade 3: 3 contracts, SL $66 per contract (total risk $200).

 

So with the fourth trade, assuming it was exited for a profit, and assuming a stop loss of $50 per contract was not hit, then the position size would be four contracts. If the profit target remains constant at $200 for each trade, then the above scenario yields a $200 profit, as opposed to a $400 trading a fixed 1 contract with a fixed $200 stop.

 

 

To a certain extent this could be used to determine trend, imagine an imaginary measure clocking up various levels using a similar method....

 

I also think you need to take the position sizing a step further....assume your total risk you might like to take is -- R, and you are willing to go at something X times

eg; Losing Trade 1: 1 contract, SL 10 ticks per contract (total risk X/R).

Losing Trade 2: 2 contracts, SL 10 ticks contract (total risk X/R).

Losing Trade 3: 3 contracts, SL 10 ticksr contract (total risk X/R).

 

Then amalgamate this with something that pyramids into the winners as well.

 

Then you are not so interested in picking great entries IF you pick the trend right.

(this is more for building positions and trend following than day trading - plus I dont do it this way, but thought if going fully automatic this would be an interesting test. For me this works when you have a trend and you only go one way on it.... all food for thought)

[BlueHorseshoe 10]

just use the clouds part...i.e. trend is up when prices are above the cloud, down when prices are below the cloud and flat when they are in the cloud...

 

 

I thought I would test Obsidian's suggestion, as the 'cloud' is fairly simple in construction, and the results may be interesting to others here (although I'm only going to test it for a particular type of strategy, so the results will by no means be conclusive for everyone). I'll test a simple strategy of buying pullbacks in trends on a daily chart, not too far removed from how I trade.

 

I don't believe that numbers have any kind of special value or that there are magical, timeless numbers, so the numbers that are used as inputs for lookback periods and projection periods within Ichimoku are in no way sacred to me.

 

Ichimoku uses three inputs. I would never optimise this many inputs - I would be reluctant to have this many within an entire strategy, let alone just the trend determination element of it - there would be too high a risk of curve fitting.

 

However I do want to optimise one parameter to facilitate a fair comparisson with the optimal SMA value. As the projection period input is processed after the lookback period inputs, it is the latter that I will optimise. This then creates a situation not too dissimilar to a neural network or markov model, with the projection period acting almost like a weighting measure with which to modify the output from the functions with fixed lookback periods.

 

With dollar optimisation of the projection period (typically a value of 26 is used by Ichimoku followers), the optimal value tested was 97, and the 10 year ES results were as follows:

 

All Trades Long Trades Short Trades

Total Net Profit $89,637.50 $39,125.00 $50,512.50

Profit Factor 2.29 2.04 2.6

 

Percent Profitable 70.30% 73.05% 65.66%

 

Avg. Trade Net Profit $336.98 $234.28 $510.23

Avg. Winning Trade $849.80 $629.82 $1,262.69

Avg. Losing Trade ($899.68) ($857.10) ($956.44)

 

Max. Consecutive Winning Trades 12 9 12

Max. Consecutive Losing Trades 4 3 5

 

Max. Drawdown (Trade Close to Trade Close)

Value ($4,612.50) ($4,612.50) ($5,825.00)

 

Incidentally, the Ichimoku strategy has been written in such a way as to treat the zone within the 'cloud' as neutral, or trendless, and allows both long or short positions to be initiated within this space.

 

Here's how this compares to using the directional slope of a Simple Moving Average to determine the trend, once again with the optimal value, which would have been 124:

 

All Trades Long Trades Short Trades

Total Net Profit $90,347.50 $41,775.00 $48,572.50

 

Profit Factor 2.24 2.17 2.31

 

Total Number of Trades 270 164 106

Percent Profitable 70.74% 74.39% 65.09%

 

Avg. Trade Net Profit $334.62 $254.73 $458.23

Avg. Winning Trade $854.83 $635.25 $1,243.08

Avg. Losing Trade ($959.54) ($893.13) ($1,033.33)

 

Max. Consecutive Winning Trades 12 9 12

Max. Consecutive Losing Trades 3 3 5

 

Max. Drawdown (Trade Close to Trade Close)

Value ($5,200.00) ($3,875.00) ($8,000.00)

 

The differences between the two approaches, with each operating at its optimal value, is in most respects negligble. I would always therefore choose the latter (SMA), as it is a much simpler function and the results are more likely to be robust.

 

Hope that's useful/interesting to someone.

 

BlueHorseshoe

Edited May 24, 2012 by BlueHorseshoe

[BlueHorseshoe 10]

I had a message request for a picture of the strategy above on a chart. Not sure how that could possibly help anyone, but it's attached (in the ludicrously small size that TL allow!) for anyone who's interested (its the last few months, in which the market trended way above the optimal 'cloud').

 

BlueHorseshoe

Ichimoku.bmp

[BlueHorseshoe 10]

Back on the subject of trend . . .

 

And in answer to my own question, my strategy now considers the ES to be down-trending. Assuming the market holds above about 1316 into the close, then I will short it. If it holds above 1320 I'll be happier. The low volume today is also encouraging, as I interpret this as lack of buying interest as we drift higher, although this doesn't form any part of an actual decision making process for me.

 

Not that I intend to make a habit of posting live trades . . .

 

BlueHorseshoe

 

I exited this position on the close today. Notionally, the entry short was 1322.50 and the exit was 1308.00, for a 14.5 point profit, or $725 with a single ES contract.

 

Because of how I trade (with CFD prices that never quite correspond) and because of slippage when trying to enter around the close, I came away with slightly less than this.

 

Well there we go - the first and last time that I ever intend to post a live trade on here!

 

 

BlueHorseshoe

[Obsidian 10]

I had a message request for a picture of the strategy above on a chart. Not sure how that could possibly help anyone, but it's attached (in the ludicrously small size that TL allow!) for anyone who's interested (its the last few months, in which the market trended way above the optimal 'cloud').

 

BlueHorseshoe

 

if you could post screenshots in .jpg that would be better. you wouldn't worry about file size

[WHY? 16]

My posts # 230,261,322,324 on the Beyond Taylor thread may useful for helping to determine the longterm trend. They discuss and illustrate the stages of longterm trends using the cash S&P.

[fast 10]

The 200 period simple moving average is an excellent long term trend indicator, and what it's indicating is the existence (or nonexistence) of a long term trend.

 

The 200 period simple moving average (200 SMA) is not limited to mere trend identification. It can also be used to determine the direction of the long term trend (if there is a long term trend). For example, if the 200 SMA is angling upwards, then the underlying security is in a long term uptrend (even if the short-term trend or intermediate term trend isn't), and if it's angling downwards, then it's in a long term downtrend. If it's flat (or mostly flat), then no, it's not trending sideways over the long term; Simply put, if the 200 SMA is flat, then the underlying security has no long term trend (when the 200 SMA is used as the long term trend indicator).

 

Keep in mind that stocks, for example, are not always trending over the long term, just as they're not always trending over the short term or intermediate term.

 

Also, timeframes have no bearing on the long term trend. Yes, an underlying security may very well be in a long term uptrend on one timeframe while the very same security may be in a long term downtrend on another timeframe, but we must distinguish between the existence (or nonexistence) of a long term trend and the direction of any long term trend that may exist (be it in an uptrend or downtrend). The 200 SMA isn't merely a simple moving average. It's a period simple moving average, so regardless of the timeframe it's plotted on, the 200 SMA still identifies the existence (or nonexistence) of long term trends, so although knowing the timeframe in which one speaks goes to whether or not a security is in an uptrend or downtrend, if it's trending at all, the timeframe has no bearing on whether or not the 200 SMA is indicative of the existence (or nonexistence) and direction (if there's a trend) of the underlying security.

 

I always have the 200 SMA on my charts, but I could care less whether or not a security is or isn't in a long term trend, since only the short term trend and intermediate term trend play an integral function in my trading methodology. The reason I even have it on my chart is because it provides another indication that I find particularly useful. It identifies strong support and resistance levels that many people and programs react to. If it didn't, I wouldn't even have it on my charts.

 

Your first question was, “How do you determine the long-term trend on a daily price chart (?),” and my answer is the same as expressed in post number fifteen by Tams: the 200 SMA! –well, except that I don’t use lowercase letters for abbreviations.

[TimRacette 11]

Aside from moving averages, simply looking at where we are in relation to the last swing high or low on a daily or weekly chart is an easy way to determine the larger trend. The simpler the better.

[BlueHorseshoe 10]

Aside from moving averages, simply looking at where we are in relation to the last swing high or low on a daily or weekly chart is an easy way to determine the larger trend. The simpler the better.

 

Hi Tim,

 

Thanks for your reply. I think for day-traders, especially of more trending markets such as currencies, then your solution is as good as any I've come across. It's certainly a more adaptive approach for short term trading. The only advantage of MAs is that they can be programmed and tested easily.

 

Is there anything unique about how you would identify a swing high/low pattern?

 

Thanks,

 

BlueHorseshoe

[TimRacette 11]

Hi Tim,

 

Thanks for your reply. I think for day-traders, especially of more trending markets such as currencies, then your solution is as good as any I've come across. It's certainly a more adaptive approach for short term trading. The only advantage of MAs is that they can be programmed and tested easily.

 

Is there anything unique about how you would identify a swing high/low pattern?

 

Thanks,

 

BlueHorseshoe

 

I prefer top down. Identifying the larger time from the daily chart, then going to the smaller time frame. For very short term intraday trading I've found it can be more effective to just look at a 5-min chart (in my case) and base everything off that one chart without accounting for the larger time frames because the multiple time frames can contradict one another.