Research: trading pullbacks in CL - futures io
futures io

Research: trading pullbacks in CL

Discussion in Trading Journals

      Top Posters
    1. looks_one Georgii with 100 posts (38 thanks)
    2. looks_two missionatsea with 4 posts (0 thanks)
    3. looks_3 robert880 with 3 posts (0 thanks)
    4. looks_4 Big Mike with 2 posts (0 thanks)
    1. trending_up 10,104 views
    2. thumb_up 40 thanks given
    3. group 23 followers
    1. forum 113 posts
    2. attach_file 97 attachments

Welcome to futures io: the largest futures trading community on the planet, with well over 125,000 members
  • Genuine reviews from real traders, not fake reviews from stealth vendors
  • Quality education from leading professional traders
  • We are a friendly, helpful, and positive community
  • We do not tolerate rude behavior, trolling, or vendors advertising in posts
  • We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community.  It's free and simple.

-- Big Mike, Site Administrator

(If you already have an account, login at the top of the page)

Search this Thread

Research: trading pullbacks in CL

(login for full post details)
  #101 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Analysis of Results

Clearly, based on our sample and our chosen rule set for entries and trade management, this first test shows us we don't have a profitable pullback system. After making some good money in January it becomes a net loser in February and March, taking back all profits and then some.

The culprit could be:

* Poor trade selection, need to refine entry rules.
* Not taking second entries when provided.
* Poor trade management rules.
* Bad luck due to small sample size.

We also notice a few other things:

* The winners appear larger, overall, than the losers.
* This exit strategy allowed all 13 "immediate win" trades to reach their profit target unimpeded.
* There can be as many as 8 losers in a row (late January - early February).
* Our initial profit targets are skewed above 1.0 R, which gives them a positive skew.

Reply With Quote
The following user says Thank You to Georgii for this post:
(login for full post details)
  #102 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

System analysis continued: Was it bad luck?

Let's first address the issue of 'bad luck', that is the system as it stands with its current rules is actually good but our sample was poor, we stumbled into a bad period (indeed, had we just looked at February and March, this system would be a total dog).

Here we have a printout of a histogram and our 'descriptive statistics' from Excel.

The histogram (trades on the y axis, %R on the x axis) shows that we do have a positive skew to the distribution. The statistically inclined will have already have noted that in the skewness measure in the table on the right.

Taking a look at our descriptive statistics box on the right, there are two things of note (outlined in red). First, the standard distribution is very high - a total of 0.91R! That's a pretty violent swing in either direction.

What should be our target average %R? I would argue that I'd like a trading system with an expectancy of 0.2R net all trading frictions (slippage and commission). This would amount to a 60% probability of being correct if we normalized things for a 1:1 risk reward ratio net frictions. That's not a small achievement for day trading where the spread and commissions can eat a significant percent of that %R (sometimes up to half of it).

Can we based on our data already understand that our trading system is no good, or is it just a bad patch in what is in fact a 0.2x %R system?

The second circled item, confidence level, tells us that with our current mean and standard deviation we have a 95% chance of being sure that the actual real expectancy of this system with its current rule set is in between +0.21x and -0.31x %R. Hardly a comforting thought, but this is how the numbers play out with this given sample size and standard deviation.

How can we increase our level of confidence in our system?

First of all, increasing the sample size. However, this may cause another issue - while large sample sizes are good in that they provide us more comfort, they will likely stretch over different market "regimes" (a mode of behavior, what people usually mean when they say 'a different market'). We can try to minimize this problem by defining these different modes of behavior, e.g. we can take a read on the trendiness of the market or the overall levels of volatility (ATR, etc). Indeed, a pullback system should work well when markets are trending, and poorly when they are rangebound. This would require altering our entry and/or exit rules to adapt to these regimes, a task that requires quite a bit more thought and research that we haven't gotten down to yet.

Second, reducing the variance of the system's performance (its %R returns). Perhaps by adopting another set of trade management rules we can accomplish this (one possible way is to manage it on a smaller time frame chart such as the 512 tick). We may miss out on some of the bigger wins, but we will more importantly minimize the losses. The lesser the variance in returns (which directly translates into a lower standard deviation), the less samples we need to ascertain if a system if working or not.

However, it's important to be careful not to get too cute with reducing variance. The system may perform better over the long run if we allow it to have greater variance, or it may not. This is something we'll have to figure out.

Bet at least a few people reading this thought initially it wouldn't be so complicated, right?

Reply With Quote
The following user says Thank You to Georgii for this post:
(login for full post details)
  #103 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Exit Strategy: Thoughts

We've analyzed our first iteration of a pullback trading system, which includes a simple, somewhat primitive exit strategy (trailing the stop under/over the latest candle). This exit strategy was just a random choice, as a starting point. The results of our system did not live up to our expectations - the system had a lot of variance and we couldn't tell after 48 samples if we had a profitable system or not when we take into account the confidence interval that we're dealing with.

Before we go into addressing the other possible causes of our ill performance (not the least of which could be poor trade selection), I want to see if we can possibly give our exit strategy some more thought.

If we think back conceptually to what our pullback trade is, it is a failure of a counter-trend trend. It is a counter-counter-trend trend, if you will

We've entered at the point at which we believe a new trend has begun, which simply resumes the original trend.

If we're mistaken, we've likely entered when the counter-trend trend was retracing as opposed to reversing (in other words we've entered in a miniature pullback in the wrong direction). In such a case we'll likely get stopped out pretty fast, within the first bar or two. The next possibility is that we've entered when the counter-trend trend was retracing in a more complex manner (a mini complex pullback). Here we might see a little more initial follow through before we're stopped out. Then we might enter into a situation where neither happens, when the market enters into a directionless consolidation which reduces the directionality and thus the probability of follow through (it may even increase the chances of a reversal, which is what some reversal traders look for).

The diagram below displays our several options when exiting (this is not meant to be comprehensive, but these are the parameters I'm used to):

The first responds to market structure, in this case it trails stops (the red line) above swing highs in a bearish trend.

The second moves the stop in relation to how much the market has moved (in this case, going in 10 point increments, but it can also be in percentage risk or percent ATR) regardless of what patterns price displays.

The third is what we used to test this system with, it simply trails to each candle extreme, which is sort of similar to the first idea but it treats every candle extreme as a swing which is not exactly the same.

The fourth idea is a time stop which can either trail a fixed distance based on how much time has passed (e.g. every bar or second bar), or it can simply exit after a number of bars have passed (e.g. exit after 7 bars) without trailing our stop (obviously leaving the original stop in place), or it can combine the two.

With methods two and four, we can trail the stop in a linear fashion (moving it a fixed amount of points or risk units each time) or logarithmically (beginning less aggressively and then becoming more aggressive, or beginning aggressively and then trailing less aggressively).

In fact, we can easily combine several of these ideas into one (for example the market structure method with the time stop). The trick here is to see what works best with the market and our personalities as well.

Reply With Quote
The following 2 users say Thank You to Georgii for this post:
(login for full post details)
  #104 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Failed Pullbacks: An analysis

Let's take a look at these failures and see if we can draw any healthy conclusions from them.

Our first category of failures are failures that happen on the same bar we enter the trade on. I've found three of these, and we display them below (5 min above, 512 tick below). The purple line is the entry, the red line beneath is the initial stop.

These appear to be DOA trades, in other words they just take us out within five minutes of being in them. Not a pleasant situation for those who've been in it. You feel like you never even had a chance, dammit!

Now, we do notice that only in one case did the market immediately reverse and go against us after triggering our tipping point entry (purple line), that is trade 14 026 (the middle one). The other two did advance somewhat - not much perhaps, but enough to possibly take some action, at least in the case of 14 006 (left most trade). If you look at the 512 tick version of the chart, we see there is a pink line just above the initial stop line drawn in. There is also an encircled level to the left of the entry, which the market broke through. We could hvae trailed our stop under the last bullish tick bar where the pink line is drawn once the trade broke the level on the left, which would have saved us maybe 3-4 points when the stopout happened. Not a treasure by any measure, but I'll pick up $30 off the street any day. It's small stuff like this that can sometimes make a big difference in how a strategy performs.

Granted, there may be cases where such management may take us out of what otherwise would have been a good trade, so we have to see if over the long term this pays off or not.

Let's also note that all three of these did have triggers that could have allowed us to enter a complex pullback. Going from left to right we see in the first trade (14 006), entering this complex pullback would have given us maybe a small profit or a medium one depending on how we trail our stops, in the second case (14 026) the complex pullback would have completed beautifully, and in the third case (14 018) we would have triggered, but with a tight stop management policy would have ended up either slightly negative or near breakeven as the strong negative excursion shook us out before the trade completed.

Theoretically it would have been possible to flip short during the stop out and possibly grab enough points to break even, but this would require a very modest take profit and very aggressive management.

Reply With Quote
(login for full post details)
  #105 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Before I go into analyzing more failures, I want to bring two categories to the fore, the "ragged successes" and the "not quites".

What I did here was I outlined the swing highs/lows of these trades, so that it would be possible to theorize about how the stop could be trailed based on price structure.

What we do notice is that the market should theoretically take off pretty fast if we have something that has a chance of getting anywhere near our initial profit objective. It should not 'revisit' the area from where it took off from. There should be theoretically no 'double bottom/double top' action. So the first hypothesis is that we want to perhaps start trailing right after the entry candle, by at least a bit. We should take just a little risk off the table as soon as the market shows it has intention to move our way.

The next thing we notice is that we do get some situations where the market nearly touches our profit target. It should stand to reason that whatever methodology we use, it should be absolutely unacceptable for us to sustain a loss or to even come out with a break even trade if the market is within a few points of our profit target. Our primitive candle trailing rule could, theoretically, allow such an unwelcome scenario. Consequently, some sort of rule that dictates once we are perhaps 75% on our way to our profit target, we should at the least be at a break even stop.

Now, I know that break even is not a condition the market sees or cares about. However, if we are in a 1:1 risk reward trade, which is our minimum criteria, we shouldn't be in a situation where the market moves 0.75 R and then retreats into the negative column. Unless you're trying to grab a very long move of many R multiples, which we're not, we don't want to assume that risk anymore, we simply won't see the payoff that would make it worthwhile. Sure, once in a while it might cost us, but we need to see what happens over a large sample of trades (and once again, let's not mislead ourselves into thinking that these 48 samples are giving us a good picture of the CL 5 minute pullback universe!).

Another point to think about is being careful not to choke the trade as it gets to the profit target. Clearly the profit target is a magnet. It is also a place where some traders are exiting early, and some counter-trend traders are scaling in. We want to give them a little room to dance, and those who've traded crude know how crude can do a pretty intense dance as it approaches a level.

So perhaps we can look at our stop trailing methodology as starting off aggressively, then giving some room for the market to do its thing, watching the technical levels as they emerge and making decisions off of those. We need to be sure the market will have enough room to get to its target and hit it without letting it get away from us and cause problems.

We haven't thought yet about when we might want to exit at market without waiting for our stop to get hit, or when we should perhaps pull in our take profit order.

This is all very general and its a discretionary process. It would be difficult to code and test this, but we can at least try to outline the parameters so we can figure out how we might test it.

Lastly, I want to indicate that I'm particularly happy we started with this kind of sample size. Yes, it's not enough to draw proper statistical inferences from, but it gives us some clues as to the variance we're likely to encounter and guides us toward asking the right questions, while remaining manageable. I can't imagine what it'd be like if we had 250 samples on our hands right now.

Reply With Quote
The following user says Thank You to Georgii for this post:
(login for full post details)
  #106 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Test Results: A new trade management system (B)

I've done a bit of thinking and came up with a new trade management system based on our current sample set (January - March 2014 pullback trades, no re-entries).

The entry criteria (stop entry at the break of the candle extreme) and initial stop criteria (stop is under the extreme of the signal bar) are the same.

If you remember the original trade management criteria was simply trailing the stop behind each candle extreme as the trade progressed (we'll call this trade management system "A")

The rules for this new trade management system "B" (which also operates off of five minute candles) are as follows:

1. When triggered into the trade, trail the stop one tick behind the extreme of the 'entry bar' (the bar that triggered us into the trade) once the entry bar closes.
2. Do not trail the stop until a significant pullback happens. It is not easy to define a 'significant pullback', but we can define it as a retracement in the trend we are trading that retreats at least 50% from its peak, and then moves back into our direction, leaving behind a point which we could trail our stop to. It is important that price leaves that point before we trail our stop. How much is 'leaving'? Till it clears through the extreme of the retracement candle back into the direction of the trend.
3. If the market returns to test the last significant retracement a second time, the initial take profit is shortened to the extreme of the retracement.
4. When the market reaches 75% of the take profit target and the candle closes, the stop must be trailed to break even. It is important that the candle closes before we make our assessment. If after the candle closes the market has retraced into negative territory, the stop is trailed just behind the candle.

What this exit system does is it begins by trailing the stop immediately to the extreme of the entry bar. This is exactly how our more primitive system A begins as well. This has the effect of cutting a little pain off of the losers that fail quickly. Unfortunately we can't save the ones that fail in the first bar, but both system A and B shave a little off those losers that fail shortly after that.

Then, unlike system A, system B gives the market significant slack to move to the profit target. The main condition we have is to watch for significant retracements, and respect them. We also have to watch if the market is giving us more than one retracement, at that point we have reason to consider an earlier exit to acknowledge that we may be running out of directionality. Lastly, we want to be sure we don't go into the negative column if we have already gotten 75% into our take profit zone.

There are no time stops in this system, though we could add one (e.g. exit at market or tighten stop and take profit after the trade has gone on for 7 bars).

Let's take a look at the results and see how they compare to our more primitive system A:

While our system is still not performing well in February and March, we are definitely more in the plus column with system B overall, being ahead by 4.18x %R or 61 CL pts ($610 for one contract). I'll take that.

Some things worth noting: point 1 shows a less hostile PnL dip in system B. System B also manages to have twice the peak PnL over system A at point 2 (8X R versus 4X R on system A). Pretty interesting.

Let's see some more detail from the underperforming periods:

As we can see, the under performing periods are both very similar, and both end up going down by the same amount in this segment (approx -4.3X %R). So the main contribution to PnL of the new trade management system appears to have been made in January when we were doing very well in both cases, the remainder of the time we seem to have been getting smacked a lot no matter how we managed our trades.

Next, let's see what our trade by trade situation looks like:

More good news. System B offers fewer consecutive losers over system A (a total of 5 versus 8). That is a pretty good thing to see, because consecutive losers can be a real emotional burden.

Next, we look at trade distribution:

System A is on the left, system B on the right. System B significantly reduces variance in the losses as you can see, while the winners get a little boost. There is still a visible positive skew.

Now finally, the stuff nobody likes to look at, STATISTICS:

As we compare numbers, we see that the average per trade performance of system B is better though if we look at the median, overall we see a number that's unpleasant - a negative one (which is smaller thank goodness than system A's number). Ideally we'd like to see both a positive mean AND median.

Interestingly enough the standard deviation of system B is in fact higher than system A, not by terribly much though.

Skewness here in fact decreases significantly, which is interesting. However, it is still a positive skew.

Lastly, the confidence level INCREASES, which is not really what we'd want to see from a standpoint of judging statistical significance. This means we have even less of a clue here if our system is really profitable or not, as compared to system A. True, the difference is not VERY big, but ideally we'd like to see it less so that we can be more sure of our results.

Reply With Quote
(login for full post details)
  #107 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Points of departure

We took a look at a sample of possible pullback trades based on three months worth of data. We tried two different stop trailing approaches, one that can be described as very aggressive, the other considerably less so. We've seen that for this sample the more laid back approach paid off, and it stands to reason: trailing indiscriminately behind each candle without regards to volatility of movement and price structure has little logic, while our second method had some kind of market logic.

However, in both cases the system was not truly profitable over a three month time span, experiencing a very similar peak to trough draw down of 8.6x %R or $1200 (one CL contract) for exit system A, and 8.4x %R or $1170 for exit system B. Not a comforting situation to say the least. That would suggest that our exit strategy is likely not the main culprit during this specific time period, unlike in January when it virtually doubled our earnings.

It also stands to imply that a lot of our trades are failing on us pretty quickly during this period as both method A and B begin by taking the same step - trailing the stop at the entry bar - and they both have almost identical draw down in this case.

There are several possible conclusions here:

1. We are simply having a string of 'bad luck' and need to look at a larger sample size.
2. We are doing a poor job of trade selection and need to become choosier, developing some filtering criteria.
3. We may still be able to further refine our exit strategy by looking at lower time frames (512 tick).
4. We may want to permit second entries.

Before we expand to another sample size or begin parsing the details of our failed trades (a time consuming endeavor), we should take a look at second entries as a possible remedy to our problem. This is effectively using a complex pullback if we get a valid setup for one, which can win back our profits in the event of a loss.

Reply With Quote
(login for full post details)
  #108 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received


I wanted to note that I made one error in my testing, I exited one trade 12 points too favorably when testing system B. My actual result for system B should have been 12 CL points less and 0.8x %R less (beginning with trade 33 which is at the beginning of March).

This does not significantly impact our statistics (though now the performance of exit system A and B during the peak to trough draw down period puts system A ahead by 9%), but it underlines the importance of checking your work carefully. In my first run of the second system I noticed that I accidentally entered an exit price 100 points too favorably. Looking at the numbers and noticing outliers is very important to catch situations like this.

Reply With Quote
(login for full post details)
  #109 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Analyzing Second Entries

Let's try to see what life would be like if we allowed second entries.

The rules are as follows:

1. A second (complex) entry is an entry that occurs lower (if in a bullish pullback) or higher (if in a bearish pullback) than the previous swing. Below is an illustration in a bullish context:

2. We enter the same exact way as with a simple pullback entry: stop entry after the counter-trend bar is broken, initial stop is at the last counter-trend bar, the initial take profit is one tick inside the last trend extreme.

3. We are not interested in third or fourth entries, just second entries.

Trade management we have not yet discussed here.

Below is our pool from the same sample period as before, 1/1-3/30/14:

As we can see, this gives us 14 extra potential opportunities.

Using exit strategy B, eyeballing the trades I have come up with the following conclusions:

Successes (S): 3
Failures (F): 2
Small profit (+0): 4
Small loss (-0):6 (note, the second collection of complex entries includes one error, 14 030 should be -0, not +0)

As we can see, a situation that doesn't swing strong one way or another. Without a proper test, we won't know the answer, so let's dig into Excel and do it...

Reply With Quote
(login for full post details)
  #110 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Second Entry, Stats

The results please...

Using the rule set we've been adhering to, we are now separately analyzing just second entries (we didn't mix it with our simple pullback, no re-entries system), as if we were trading it as a system on its own.

We are using exit system B.

As we can see, using second entries based on our criteria would have helped juice up our PnL by a whole 22 CL points. If we add this to our prior system using exit system B, we increase our PnL from 29 to 51 CL pts (a whole 75%).

This is encouraging, although it pays to note that this is a VERY small sample size at just 14 observations. We would need to see a bigger picture if we really wanted to get comfortable with whether re-entry is a good policy or not.

Also, this system would have eroded our general PnL during its initial iteration. This could be of course the classic 'portfolio approach', one zigs while the other one zags. As long as the zigs are greater than the zags in the long run, we're in business.

Note: An earlier version of this post had an incorrect diagram posted that stated the system had 8 consecutive losses. This was a mistake. This system has no more than 5 consecutive losses which is within expectations but again, this is a small sample size.

Reply With Quote
(login for full post details)
  #111 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Recap: what we've learned so far

Let's just re-cap what we've learned to date...

* Our pullback system as it currently stands gives us approximately 48 opportunities over a 61 trading day time span, or 62 opportunities if we permit re-entry (complex pullbacks). This includes both regular pullbacks and breakout pullbacks with decent range size.
* We've been able to establish that over our chosen sample size, a more sophisticated stop trailing strategy can produce significantly better results than a primitive 'trail the last candle' approach.
* We've seen some evidence that permitting second entries might enhance our PnL.
* We're experiencing serious underperformance in 2/3rds of the sample, showing a downward trend with both exit management systems, suggesting either poor trade selection or a bout of 'bad luck'.
* We haven't been able to make any statistically significant conclusions about the expectancy of our strategy due to its variance and small sample size.

All in all, my target is to have a system that offers a 0.2 %R expectancy net of frictions (my %R calculations, incidentally, do not include frictions while my point PnL does - but this does not affect our work to date).

If we take a look at the best performance we've gotten, it has yielded 29 CL points net of frictions, and 51 CL points if we include re-entries. The broker, meanwhile, made the equivalent of 24 and 31 CL points respectively, and 48 and 64 CL points went to slippage. That means our profit is merely 40% and 53% respectively of frictions. Now you see what they mean when they say how difficult it is for day traders to get over the 'vig'. This is what we're up against.

Clearly we don't have a satisfactory system yet, but at the very least these exercises have given us a helpful point of departure.

Next on our agenda we have to explore the area of trade selection. If we could just get rid of 2 1.0x %R losers we could increase our PnL by a factor of 2. Of course, this also means we may have to eliminate some winners as well, so good trade selection may not immediately prove its worth.

Naturally it will also mean we'll have to be disciplined enough to let go of those opportunities when it comes time to trade, especially if we're going through a draw down like we were in February and the market becons us with tempting signals that may have followed through once or twice.

A plus we can look forward to is we may even discover trades that work well in the opposite direction at times, including opportunities where we may want to reverse our position. It is important to remember, however, that not every failed trade should be a reverse, and this should be approached as if you are developing a separate strategy (note, for example, that we can likewise treat our 'second entries' as a different strategy and refine its criteria, too).

We can also work at improving our exit system further, possibly using the 512 tick chart to get a finer read on things.

Lastly, once we're ready we can increase our sample size and see what kind of real variation is out there. That may force us to change our rules again and it may also give us ideas how to change as the market changes.

Reply With Quote
(login for full post details)
  #112 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Screening Trades: Applying a Filter

Let's see if we can get a picture of what is going on on the larger scale of things. I've created this interesting chart that features the trading signals provided by the system over our three month sample. The candles represent a trading session (9:00 - 14:30 EST), with the trades of the day on top of the candle, with re-entries on the lower end of the candle.

If you see what the key says, the red plusses are the successful short trades, and the green plusses are the successful long trades, while the red minuses are the unsuccessful short trades, and the green minuses are the unsuccessful long trades. I know, it's a bit confusing at first, but try to disassociate the color red and green from PnL and think of it in the bull and bear context only, with the + and - sign substituting for PnL.

Below is the Average True Range, an indicator that lags but gives us a hint as to volatility.

Keep in mind when you're looking at this, the trades are made without the benefit of knowing how the candle closes for the day. This is a very important aspect in our testing, we have to be on the look out for future information creeping into our analysis.

One thing struck me right away when I saw this: long term bias. Clearly crude was in a down trend, then in an up trend, then reversed down once more, then up again. There are also various degrees of trendiness and session overlap. Could success or failure be hidden in taking pullback trades in less than favorable trend conditions?

I decided to look at trend bias at first. I used the 21 period EMA as a filter. In Excel, I created a condition that if the PRIOR DAY closed above the 21 pd EMA on the higher time frame, then only long pullback trades were permitted going forward. The same condition held for the short side, only in reverse: if the market closed below the 21 pd EMA, then the market was a short for the day. If the market closed at or very close to the EMA, no trades were permitted.

The results I got were very interesting. Instead of having 48 trades on the no-reentry system, I got 18 a mere trades. Applying the filter to the re-entries, I got an extra 5 trades instead of an extra 14. In each case, the signals went down by approximately two thirds.

The performance of the system, using our "B" exit criteria (from now on this will be our default exit system), we were up 68 CL points with the no reentry system. Reentries added another 15 CL points to our PnL, giving us a total of 83 CL points (interestingly enough this is less than what we would have gotten had we taken unfiltered second entries, but with just five second entry samples in the EMA filtered system we can't draw any solid conclusions). All numbers here are net frictions.

Wow, that's awesome! Our signals went down in number but we experienced a dramatic improvement in performance at the same time.

In case you're wondering, adding long or short trades when the prior day closed AT the moving average slightly degraded performance for both subsystems.

Here is our performance on a trade by trade basis:

Looks interesting, but our only problem is that we now have a small sample size again, so that makes it difficult to draw any statistically significant conclusions.

Still, we have potentially honed in on an interesting concept - a filter that may have drastically reduced our trading signals (now we'll be getting as few as 1/3rd of the previous opportunities), but more than DOUBLED our PnL!

Sure, we got rid of some profitable signals for certain. But more important, we got rid of the stinkers which were eating away at our precious account, and now we're richer for it, even though we're much more bored since we have to sit and watch trades pass by.

Now does this mean we can idly slap on the 21 pd EMA, wait for a qualified pullback, and then feel alright? It's not that easy.

We may be able to still get rid of some more stinkers that fit under the EMA filter, and we may also find some nuggets from the trades we've filtered out.

If we take all the profitable trades that DON'T fit the EMA filter from our no re-entry system, we find that we have 12 out of 30 rejected trades (40%) which could have contributed to our PnL. Those profitable trades total to 162 CL points. If we add the profitable trades from trading the re-entries, we get an additional 47 CL points, totalling 209 points of good trades we can't enter simply because they fail the filter.

If we could grab just a QUARTER of that currently rejected PnL via better entry criteria, we'd be up an extra 40 points for our simple pullback system and 52 points when second entries are included. That is a 58% and 62% potential increase in PnL.

We could also, theoretically, go the other way - trade only EMA filtered trades and find other criteria to further pare them down to increase profit. We might get an even tighter performance out of our system. However, this runs the risk of seriously depleting our opportunities and also introduces the risk of becoming too curve fitted. Caution is required here.

Reply With Quote
(login for full post details)
  #113 (permalink)
New York, NY, USA
Posts: 105 since Jun 2013
Thanks: 2 given, 46 received

Points of Departure

Now we're at a juncture where we need to make some decisions as to which direction to take our research in.

Our initial goal was to cast a fairly wide net of trade setups and not be too picky about it. There are several reasons for this.

First, when operating with the benefit of hindsight, it's very easy to say 'hey, this is a successful trade, of course there was a signal here", and the opposite when we see a failed signal "but of course that was a bad trade, should've been obvious". I tried very hard to look at the criteria I established and apply them as broadly as I could.

Second, I don't want to ascribe selection criteria preemptively without seeing evidence of what makes for a good versus a bad trade firsthand.

What we need to start doing now is becoming more specific on why something fails and why it succeeds, so we can begin to properly parse our A|B trades (the trades that conform to our system), and our C|D trades (the trades we exclude from our system).

We're separating the profitable trades into A's (system trades that work) and C's (non-system trades that work), and the unprofitable trades into B's (system trades that failed) and D's (non-system trades that failed).

We can begin by creating the following sub-systems:

1) Pullback trades.
2) Breakout pullback trades.

and in turn, each of these sub-systems can be further broken down into:

a) Simple entries.
b) Complex entries.

We can illustrate them like so:

This might or might not be helpful for us, but I think it makes it a bit easier to separately look at some specific things that we otherwise wouldn't pay attention to. For example, we would want to look at the strength of a breakout:

Notice how in the example on the left, which succeeded (albeit on the second try), the bullish breakout was fairly strong. In the example on the right, which failed quite drastically, the bearish breakout (A) was considerably weaker.

Also, we may want to see circumstances in which it would be wiser to wait for a complex pullback as opposed to a simple one, for example when the trend shows signs of weakening.

Once we look at enough samples, we may begin getting ideas for a counter-strategy - that is knowing when to stand by and wait for a failure:

A failed pullback can be thought of as a double top/bottom, or what Al Brooks calls a 'higher low/lower high major trend reversal'. It might not be necessary to analyze this for the sake of possibly trading it, but it would help in terms of being able to not enter or exit early if we notice the environment is conducive to a counter-trend move.

Now, in our last analysis we discovered what might be an effective trade filter, a moving average, that reduced the number of trades but increased the profit by a factor of 2.8. The rationale is simple: we're taking only those trades that are in alignment with the longer term trend, which is signaled to us by the moving average.

However, two caveats: 1) We did not test this filter over a very large sample size, and 2) We are leaving some potentially good trades on the table. If we can find a way to separate out those good trades that do not fit our 'filter' but merit being taken, we may add to our PnL.

To this end, our next step should be to better categorize our existing trades and see if we can do some weeding out of inferior signals. This is not an easy task for one reason: it requires a lot of discretion and as a result involves subjectivity. That makes it difficult to test, as we are going to be even more prone to hindsight bias as we're trying to optimize the criteria on observable trade samples. During back-testing we will be able to get a better idea of what we could reasonably expect, but even that will never give us the real picture until we enter the live market.

Does this make analysis fruitless? Absolutely not. It just means that we need to be aware of our limitations going forward so we're not disappointed later on.

Reply With Quote
The following 4 users say Thank You to Georgii for this post:
(login for full post details)
  #114 (permalink)
Sydney, Australia
Posts: 1 since Nov 2012
Thanks: 1 given, 0 received

Absolutely fascinating thread thank you. I am also heavily into pullbacks and think that you would find anything written by Lance Beggs extremely interesting.


Reply With Quote

futures io Trading Community Trading Journals > Research: trading pullbacks in CL

Last Updated on October 19, 2014

Upcoming Webinars and Events

NinjaTrader Indicator Challenge!


NEW BlackBird Features + FOREX Support w/Jeremy Tang @ SharkIndicators

Elite only

Our 12-year anniversary w/ $$,$$$ prizes (check soon)


Copyright © 2021 by futures io, s.a., Av Ricardo J. Alfaro, Century Tower, Panama, Ph: +507 833-9432 (Panama and Intl), +1 888-312-3001 (USA and Canada),
All information is for educational use only and is not investment advice.
There is a substantial risk of loss in trading commodity futures, stocks, options and foreign exchange products. Past performance is not indicative of future results.
no new posts