Diversified Trend Trading Approach

Performance Summary - Discretionary Systems - Trend Following Retracement Entry Technique (DTTR)




Overview
This portfolio crawls along when compared to the multi-timeframe discretionary breakout strategies. The very selective trade entries combined with the restriction to the D1 timeframe results in few and far between trades, but what a breath of fresh air. This strategy reminds me of some of my favourite long term wealth building techniques like DIBS and the Equity Millipede. We could be holding some of these positions for years providing that market structure is maintained.

Despite the apparent inactivity and low trade activity, unrealised profit is progressively ratcheting upwards so now we are sitting on a great nice war chest. This just goes to demonstrate that we really don't need to trade lower timeframes. A far more peaceful and profitable trading life can be enjoyed out in the longer term timeframes. This is also evident when looking at the most successful CTA's in the industry (eg. Dreiss and DUNN) who trade the weekly timeframes. Once again, it demonstrates that over-trading is the greatest enemy for the trader. It unnecessarily introduces drawdowns into the system and also significantly increases the trading cost base associated with trade frictional costs (spread, SWAP and slippage).

Given the mean reversion nature of this strategy back into the primary trend, it is anticipated that this system will prove a useful addition to the discretionary system portfolio arsenal which so far is focused on trend breakout strategies. When trend breakouts are few and far between, this strategy is anticipated to shine attributed to mean reversion activity.

Trade Statistics - Closed Trades
There is not sufficient data to make any informed decision relating to trade statistics. The low Pwin% of 28.4% YTD is very deceptive as the strategy only commenced in late April and as a result, the statistics associated with closed trades only relate predominantly to losing trades as the majority of profitable trades currently are still open and not currently reflected in the results. Closed trades for the month of August have resulted in s small residual drawdown but given the unrealised profit sitting in the portfolio, there is no reason for concern at this stage.




Equity Performance & System Health
Currently with the limited trade history, not many conclusions can currently be drawn regarding equity performance or system health. Needless to say, things appear to be tracking Ok.









Trade Highlight
Nothing to report for August so will simply focus on the good result achieved recently with Silver which has enjoyed a nice bull run and has allowed us to take half our position off the time with a nice profit of 3xR. The remaining position is maintaining it's value and we have have the ability to benefit from any additional upside.

For those who do not like seeing their unrealised profits vapourising away before their eyes as the trend 'bends' then this is an option to ease the pain somewhat. While in the long term this will probably be detrimental to our overall returns, the benefits to cashflow for such a long term strategy need to be considered.





Current Open Positions
Good unrealised profit is currently sitting in the portfolio allowing us to maintain an active aggressive stance in the market and not be concerned with portfolio heat.





What's Trending According to this Technique

GBPUSD





CADJPY




AUDJPY




NZDUSD




That's it for the week. A detailed spreadsheet of analytics is attached.

Best of trading for next week.


Cheers

C

Trend Following Performance Report - July 2016

Inserted Video

Overall performance - Trend Following (TF) Index



Fund Performance Table



Best CAGR - Jan 2002 - Jul 2016




Worst Drawdown - Jan 2002 - Jul 2016




Lowest Drawdown - Jan 2002 - Jul 2016




Best Performer (Adjusted CAGR) - Scaled to Drawdown Exposure - Jan 2002 - Jul 2016



Best Performer - Annualized Sharpe - Jan 2002 - Jul 2016



Best Performer for the Month of July 2016



Best Performer for the Last 12 Rolling Months to July 2016



Cheers guys

C

This is superb C, and proves that the trend filter was a good addition. In fact, I'm not sure it's a mean-reversion system any more because it's entering in the direction of the short-term trend after a retrace

Just out of curiosity, have you tried this exact setup with no SL? Can't imagine it's going to be any better, but you never know. And was the 50 SMA better than the 100 and 200, I wonder?

You're well ahead of me here mate. Great work!

EDIT. Damn!!!!!! When it appears to be too good to be true it usually is. Disregard following post as an error was found in programmatics.

NT

You are right. They work beautifully as short term trend (momentum) traders. The SMA addition has been a windfall.

It just keeps getting better. I am now testing a variation to his Chapter 2 but simply applying a setup if the 3rd bar has the lowest low, has a 50 SMA filter applied and holding the position for 50 bars.

Obviously there is something to these short term trend traders with retracement entries and time exits .....Have a look at the results below. Now I am starting to salivate.

The basis for the setup and entry signal is as follows:
1. We need a lowest low on the 3rd bar;
2. We buy on the open of the 4th bar;
3. We are only ever in 1 trade at a time per market;
4. Position sizing uses 2 X ATR (14 day period);
5. Initial stop is placed 2 x ATR (14 day period) below entry price less commissions;
6. We hold the trade for a period of 50 bars and exit on the open of the 51st bar OR we exit on the open of the bar following the triggering of the initial stop; and
7. This is a long only strategy.




Analytics conclude the following:
1. It is profitable on 37 of the 40 instruments in this universe. This is an excellent and highly unusual result so it has broad application across any liquid instrument in any sector;
2. It produces a CAGR (no profits reinvested) of 8.34% with a max drawdown of 5.47%. This is a superb result and very scalable;
3. Profit factor across the 16.5 years is 1.70.....WOW!!!
4, Great positive expectancy of 0.39R
5. If profits are reinvested and we scale up to a max drawdown over the 16.5 year period of 20%, we achieve a CAGR of 26.72%.

NT....I am a bit stunned at the moment but what a great way to end the night.

Obviously we need to spend more time on Reschke's strategies mate.

aveagreatweekend :-)

C

EDIT. Damn!!!!!! When it appears to be too good to be true it usually is. Disregard following post as an error was found in programmatics.



It just keeps getting better. A simple variation that reduces the holding period to 30 bars.

Time for jubilation.



Analytics are as follows:
1. It is profitable on 39 of the 40 instruments in this universe;
2. It produces a CAGR (no profits reinvested) of 8.54% with a max drawdown of 5.95%;
3. Profit factor across the 16.5 years is 1.60
4, Great positive expectancy of 0.29R
5. If profits are reinvested and we scale up to a max drawdown over the 16.5 year period of 20%, we achieve a CAGR of 26.98%.

Attached Image (click to enlarge)

Cheers NT :-)


C

Great equity curve, keep up the Good work.

I spoke too soon Bass. Disregard typical optimistic jubilation. It was great while it lasted though......back to the grind. I wish I could blame my cat but unfortunately it was a strictly human error :-)

Its achivable, don't give up, the other guys that did it are not smarter than you or me, they are only better equipped and better financed.

Impossible dosen't exist in our vocabulary, keep It going .

Best,
bass

Thanks C. The great-looking equity curves are the ones we tend to check the most, and rightly so. But I wonder how many systems we've rejected over the years due to an initial poor-looking equity curve which was caused by an undiscovered coding error

I finally managed to do a bit of backtesting yesterday and am composing a post about it right now. I'll post it here when I can, but work beckons so it might not be today sorry. Watch this space!

So C, so I managed to spend a couple of hours on one of Reschke's setups on Sunday and have made a little progress.

The first thing I realised is that you have far more data over different asset classes than I do, so the main backtesting should probably be done on your data using the powerful Excel beast that you and RedLineFred developed. So instead of running the tests on daily bars over many years, I decided to run them on hourly bars over just the last two years. I didn't expect much profit from these systems on hourly bars, mainly because:

a) time of day might be an issue
b) intra-day spikes caused by news events
c) cost per trade is larger because slippage and commission is the same as for daily, and I deliberately included a large slippage of 4 ticks per round turn (which is unlikely on such liquid instruments)

And I got what I expected... net losses on every run! But the profit/loss result didn't matter. The point of my testing was to try various combinations of conditions and parameters in order to see which ones would produce the best results. Then, if you wanted, you could run the best combinations on The Beast and see how it looks on the daily. Curve-fitting, perhaps? Well yes, maybe a little. But I'm talking about very general conditions here, such as whether to use a stop loss or not.

So, I decided to run the tests on a small but diverse portfolio of just five CME futures instruments:

6E (EUR.USD)
6J (JPY.USD)
CL (Crude Oil)
ES (S&P 500)
GC (Gold)

These are the basic rules that every test followed:

a) go long after 3 consecutive down bars if last bar is above SMA (or in the reverse case: short after 3 consecutive up bars if last bar is below SMA)
b) hold trade for x number of bars
c) if SL used: fix at 2 x ATR(14) from entry
d) 1 contract per trade (fixed position size)
e) commission = $4.02 per round turn
f) slippage = 4 ticks per round turn
g) test period is 21-Aug-2014 to 20-Aug-2016
h) always enter and exit at the open of the next bar following the signal

And these are some of the variable conditions I tested:

a) whether to always go long, always go short, or go either way depending on the setup
b) whether to take only one trade at once, or allow simultaneous trades if setups allow
c) whether to hold for 10, 30 or 50 days
d) whether to use SMA(50), (100) or (200)
e) whether to use EMA instead of SMA
f) whether to use SL or not
g) whether to use an additional trend filter or not: ADX(14) > 18

Almost all conditions generated at least 1000 trades over all five instruments, which is an adequate sample size. I didn't test every single permutation of course because that would have taken too long, so I used my judgement in discarding those conditions that weren't working at an early stage. And I've since realised that the included commission is a little lower than what I actually pay, but since we're only comparing results it doesn't really matter.

Now, I won't bore you with all the results and stats, so instead I'll highlight the main conclusions. "Result" below means net profit/loss:

a) "long only" and "short only" produce very similar results
b) "both long and short" yields a few more trades, but a similar result to (a)
c) exiting and reversing "both long and short" when an opposite setup occurs is not so good; better to stay in each trade until the end of its holding period
d) multiple trades per direction (one for each new setup) is unnecessary; it leverages both good and bad trades and amplifies the result
e) use of SL is ambiguous: it always produces more trades (because not all trades are held for their full holding period), but sometimes with more profit and sometimes less depending on the other parameters.
f) higher SMA periods are suited to higher holding periods
g) EMA produces virtually the same results as SMA, so use SMA because it's simpler
h) the ADX filter doesn't improve the results, presumably because the SMA acts as a trend filter

And the very best results were obtained using the following variables:

a) trade both long and short (only one trade per direction, hold each trade for its full holding period)
a) holding period of 30 bars or 50 bars
b) SMA(200)
d) no SL
e) no ADX filter

The 30-bar holding period produced 1287 trades with a net loss of $-22614
The 50-bar holding period produced 898 trades with a net loss of $-14505
(the less trades the better when running this on the hourly it seems, due to the reasons mentioned above)

It would be interesting to see how these two combinations perform on the daily over 16.5 years and 40 instruments, both with and without a stop loss and with position sizing. Over to you C, if you have the interest and time! Thanks

Hat is off to you NT. Lovely analytics. :-)

You have also picked a particularly tough period for trending markets NT (over the last 2 years) which no doubt will effect the results over that timeframe. With a difficult market plus increased trade frequency on H1 relative to say the D1 timeframe, I am not surprised with the overall results.

As soon as I get a chance I will run your two variants (30 bar hold and 50 bar hold) for the long/short combo on my universe and let you know the results.

From my current tests, the consecutive lower low variant is generating better results than the simpler low over the last X bar variant so I am also looking forward to testing the other ideas in that book.

I'll keep you posted NT and thanks greatly for this feedback mate. This style of collaboration really helps.

Cheers

C

Thanks for the kind words C. They weren't that great though, not even a picture! Yours are much better.

I was troubled by the discovery that exiting and reversing when an opposite setup occurs doesn't improve the results, in fact makes them worse. It doesn't seem logical. So I've just run a few more tests on that and can confirm that it really is the case. It always produces more trades but less profit (or in this case more loss)

This could be because it's running on the hourly though, and more trades might equate to more profit on the daily? I don't know. So if you have the inclination to try that condition on the best result you get then please do! Thanks

What we might be finding mate is that frictional costs (eg. commissions and slippage and in my case SWAP) outweigh any alpha in the strategy. If that is the case then reversing the strategy unfortunately does not necessarily work.

For what it is worth, in every strategy I trial over the long term and across a diversified portfolio, the combined commissions and SWAP exceed the profit margin, thus you need a sufficient edge to overcome these frictional costs....and finding a robust strategy with sufficient edge is the hard part of this game. In general terms I have more joy out in the longer timeframes because of this factor. Fewer trades held for longer tend to reduce frictional costs of trading.

C

Hi Guys,

Great thread! Wondering what you think of the following chart? is if possible to prove the 84% mentioned here?
If yes, surely that could be used to trade as an edge?
What are your thoughts?

Great point timster. To win in this market you need to consistently produce different systems that offer statistical arbitrage opportunities (or alpha) until of course they don't. That means that you need to continuously develop new systems that are appropriate for current market conditions and have clear rules for when you tun off a system after the alpha has been exploited.

If you approach your system development in a probabilistic way for each material element of the strategy such as the setup, the signal, the hold period and the exit, and assess the individual probabilistic outcome for each element, you can then deliberately construct a system with an edge. I note that this is the preferred approach for some successful system developers such as Ernie Chan.

Effectively you need to become a system factory whereby your efforts are invested in testing as many different systems as you can that respond to a variety of different market mechanics and simply turning those systems into live traders as soon as they are discovered...... and then crossing the fingers that the out of sample results continue to achieve desired performance metrics.

The three line strike bearish reversal is interesting timster. I assume that the 84% success relates to the Pwin rate. If it does, then it definitely is worth testing. The question of course is whether the 84% success in the setup translates into a robust profitable system when risk:reward metrics are applied by selecting a signal, hold period and exit condition. If we applied a short term time stop of say 4-8 bars after entry, then the closer we can get to a symmetric risk:reward profile, the closer we can get our strategy expectancy to be representative of the Pwin ratio.

I will have a think about this and might test it across the portfolio when I can get to it.

Cheers

C

Thanks for posting these probabilities timster. I'm a bit confused as to what they're actually measuring though. In the case of the bullish reversal (top left), is it saying that the probability of the next bar after three consecutive down bars (and lower highs and lows) being a bullish reversal bar is 84%? If so, which instruments, periods and timeframes was this measured against? It's hard to believe in such a high figure without some context.

Having said that, it would be easy for someone to test it using Excel, NinjaTrader, MetaTrader, or anything else. Personally I'd be surprised if the probability were as high as on that chart you posted. And even if it were, there's the problem of where to put your SL, as C suggested, since the target bar has a lower low than the trigger bar. But it's worth a try I suppose.

In a similar vein, I've finally got some positive net profits out of the H1 testing I've been doing (without resorting to curve-fitting hopefully). How? By using three consecutive higher highs (or lows) rather than three consecutive up (or down) bars as a setup. Here's a recap of the best results using the old way:

a) H1, two years
b) trade both long and short (only one trade per direction, hold each trade for its full holding period)
c) close of last bar must be above SMA(200) to go long, or below SMA(200) to go short
d) no SL
e) full commission and slippage included as before
> 30-bar holding period : $-22614 (1287 trades)
> 50-bar holding period : $-14505 (898 trades)

And here they are using the new way:

> 30-bar holding period : $8626 (1641 trades)
> 50-bar holding period : $435 (1062 trades)

Still not the holy grail obviously, but more trades and more profit with such a high cost per trade is encouraging indeed.

And here's something else I found just before I had to stop for the day. If you use my BADL indicator (price and volume) instead of just price to measure the trend, then you get much better results. Basically you can only go long if the last bar's BADL is above its SMA(200), and short if below. Here are the new results above using BADL SMA instead of price SMA:

> 30-bar holding period : $31887 (1664 trades)
> 50-bar holding period : $20932 (1067 trades)

These results are hugely encouraging, and I'm going to check them in detail tomorrow to find out why they've suddenly improved so much. It's also personally encouraging to know that my BADL idea might have an edge after all (I was beginning to think it was a bit useless)

Something else I've just realised while typing this... the higher highs (or lows) setup I tested doesn't care about whether the other ends of the bars are higher (or lower) too. Will adding this filter make a difference I wonder? It will make the bars look like the ones in timster's chart if so. Worth checking I reckon. I'll let you know.

Great feedback guys.
I too have a few q's re these % quoted. I'm aiming to complete some code during the next few weeks to test it.
I'll be assuming that a trade would commence as a buy entry after the completion of the engulfing candle. S/L would be below the low of the engulfing candle.

Really not sure how else to trade it live!

I'll get back to you re my results.

Just wondering are there any other candlestick tests we could review? Anything else you have seen with % expectations?

Some recent interesting studies on Trend Following.

Enjoy

C

Trend Following: Equity and Bond Crisis Alpha


In this paper we have studied the performance of trend-following strategies since 1960, a time period that includes extended bull and bear markets for both equities and bonds. We have shown that a strategy that closely matches the BTOP50 performed consistently over the 56 years covered. It also has a number of compelling risk characteristics: positively skewed returns and strong performance in the worst equity and bond market environments, which we refer to as equity and bond market crisis alpha respectively. Despite 56 years of supportive empirical evidence, it is natural to ask the question whether momentum strategies will continue to be profitable. While past performance does not guarantee a similar future, it is worth noting that academic papers on the topic date back to at least Jegadeesh and Titman (1993) and performance has persisted for the 20+ years since then. While a meaningful amount of capital is dedicated to exploiting the momentum phenomenon, there is evidence that there are very large players that have a tendency to take the other side (knowingly or unknowingly), possibly ameliorating concerns that too much capital is chasing momentum profits. For example, Lou, Polk and Skouras (2016) present evidence that stock momentum is different from other trading strategies in that professional, institutional investors tend to “trade against the momentum characteristic”.



Trend Following Works Weakest After Crises

Study Results
Hutchinson and O’Brien’s data set for the global analysis consisted of 21 commodities, 13 government bonds, 21 equity market indexes and currency crosses derived from nine underlying exchange rates covering a sample period from January 1921 to June 2013.

Their results include estimates of trading costs as well as the typical hedge fund fee of 2% of assets and 20% of profits. Following is a summary of their findings:

1. Time-series momentum has been highly successful over the long term. The average net returns for the global portfolio from 1925 to 2013 was 12.1%, with volatility of 11%. The Sharpe ratio was an impressive 1.1 (a finding consistent with that of other research).
2. A breakdown in futures market return predictability occurs during crisis periods.
3. In no-crisis periods, market returns exhibit strong serial correlation at lags of up to 12 months.
4. Subsequent to a global financial crisis, trend-following performance tends to be weak for four years on average. This lack of time-series return predictability reduces the opportunity for trend-following to generate returns.
5. Comparing the performance of crisis and no-crisis periods, the average return (4.0%) in the first 24 months after the start of a crisis is less than one-third of the return (13.6%) earned in no-crisis periods. The performance in the 48 months following the start of a crisis (6.0%) was well under half the return in no-crisis periods (14.9%).
6. Results were consistent across stocks, bonds and currencies. The exception was commodities, where returns were of similar magnitude in pre- as well as post-crisis periods.
7. The authors found a similar effect when they examined portfolios formed of local assets during regional financial crises.

Performance Summary - Discretionary Systems - Breakout Techniques (DTT, EDTT, EDTT1)

It's a nasty time for the trend followers out there and symptomatic of what happens immediately after a crisis event (eg. Brexit). This is when central bank intervention steps in and trend followers start entering protracted drawdown periods. It is a sea of red for trend followers for the month of August and early September. This is the time to pull out the Shiraz, chill and watch from the sidelines if you can until volatility returns.



Overview
Following the Brexit glory in June for trend followers (the crisis), the markets have become far more subdued with mean reversion once again becoming the flavour of the month as central bank intervention kicks in. This is typical of periods following crisis events as discussed in the previous post. July was a flat performance month and August/September is now seeing a building drawdown. During these more difficult periods over-trading becomes your enemy. To be safe as opposed to sorry, I am now taking proactive steps given the discretionary nature of this strategy to protect capital by extending the range of pending orders to keep clear of current volatility. What we therefore want to see statistically in our trade frequency for August and September (until volatility rears it's head again) is a noticeable reduction in trade frequency that demonstrates the efficacy of our trade management decision.


Trade Statistics - Closed Trades
August/September is experiencing a low Pwin rate of 27.7% and 12.5% respectively.....yuck.....this is a hostile market for trend followers. Look at the current Pwin levels compared to the long term average of 39.1% which indicates the current mean reverting nature of this market and relative lack of trend trading opportunities. As a discretionary trader, we can therefore take proactive measures to minimise drawdowns during these unfavorable periods. While we never know when volatility will resume, we are prepared to take some future profit off the table with delayed entries to preserve our capital at all costs. You never know how long these unfavorable conditions will last....so it's defend the castle at all costs. We therefore need to place our entries well outside the current normal range of volatility and wait until the market comes to us if volatility and momentum pick up.






Equity Performance & System Health
The equity curve is now starting to retrace as the drawdown is building. Now is the time to get defensive to keep the powder dry and closely monitor trade frequency to ensure that you are not tempted to over-trade. It is preferable to reduce trade frequency by stretching entry conditions beyond normal market volatility. This way you can avoid the 'noise' of the market when mean reversion becomes the name of the day. While this may result in some 'give-back' of potential profit on entry, a risk management mechanism such as this is more likely to assist in preserving capital.

The Win% and Average Win $ chart highlights the under-performance of August and September with respect to the historical cumulative average. Obviously when both these metrics sit above the cumulative average, things are rosy........ so at the moment you can see that market conditions are currently less than optimal for trend trading in general. You will also clearly see how our capital protection trade management decision is taking effect with significantly reduced trade frequency compared to the historical average. This is one of the tools we have up our sleeve as a discretionary trader that many automated strategies don't have at their disposal.



The Trade Profit Frequency chart highlights the positive skew of the current strategy which is what we are after with this technique, namely lot's of small losses that are more than compensated by a few large wins which is symptomatic of trend following strategies in general.




Trade Highlights
For the month of August, the only trade so far worth highlighting was USDCHF. Very slim pickings this monthyet beggars can't be choosers.

Attached Image (click to enlarge)
http://www.forexfactory.com/attachme...1&d=1471661299



Current Open Positions
The unrealised profits sitting in the portfolio of approximately $4K or 5.14R are too small to allow for aggressive game play. We need to keep things tight during these tough times. Given my experience in trend following, this period following a market shock event is obviously a very troublesome time for trend followers and a time when we start noticing that most trend following programs quickly go into max drawdown territory. Stay tuned for the Fund Performance results that highlights this factor across the board of the trend following CTA's.....August has been ugly. A sombre moment for reflection on the price you pay for trend following.






What's Trending According to this Technique
Not much on the radar at the moment.

EURJPY







USDJPY





Below is the analytic spreadsheet for those that want to get into the detail.


Cheers

C

Performance Summary - Discretionary Systems - Trend Following Retracement Entry Technique (DTTR)





Overview
This portfolio crawls along when compared to the multi-timeframe discretionary breakout strategies. The very selective trade entries combined with the restriction to the D1 timeframe results in few and far between trades, but what a breath of fresh air. This strategy reminds me of some of my favourite long term wealth building techniques like DIBS and the Equity Millipede. We could be holding some of these positions for years providing that market structure is maintained.

While trend trading in general has been very difficult over the past month or so, unrealised profit in the portfolio is holding at current levels and we have a nice war chest for the future. This just goes to demonstrate that we really don't need to trade lower timeframes.

Given the mean reversion nature of this strategy back into the primary trend, it is anticipated that this system will prove a useful addition to the discretionary system portfolio arsenal which so far is focused on trend breakout strategies.

Trade Statistics - Closed Trades
There is not sufficient data to make any informed decision relating to trade statistics. The low Pwin% of 33.7% YTD is very deceptive as the strategy only commenced in late April and as a result, the statistics associated with closed trades only relate predominantly to losing trades as the majority of profitable trades currently are still open and not currently reflected in the results. Closed trades for the month of August and early September have resulted in a small residual drawdown but given the unrealised profit sitting in the portfolio, there is no reason for concern at this stage.





Equity Performance & System Health
Currently with the limited trade history, not many conclusions can currently be drawn regarding equity performance or system health. Needless to say, things appear to be tracking Ok.











Trade Highlight
Not a cracker over the past 2 months but silver is holding up well which has enjoyed a nice bull run and has allowed us to take half our position off the time with a nice profit of 3xR. The remaining position is maintaining it's value and we have have the ability to benefit from any additional upside.

For those who do not like seeing their unrealised profits vapourising away before their eyes as the trend 'bends' then this is an option to ease the pain somewhat. While in the long term this will probably be detrimental to our overall returns, the benefits to cashflow for such a long term strategy need to be considered.






Current Open Positions
Good unrealised profit is currently sitting in the portfolio and has been stable over the difficult trending period of the last few months following Brexit. This allows us to maintain an active aggressive stance in the market and not be concerned with portfolio heat.







What's Trending According to this Technique

GBPUSD






CADJPY






AUDJPY







NZDUSD





A detailed spreadsheet of analytics is attached.


Cheers

C