Trading pullbacks – stunning results

Looking out at the Kuta surf, I can see surfers catching waves. They’re looking for troughs because they know that each trough means a wave is following. This reminds me a lot of trading pullbacks.


After my recent laborious stint in Papua New Guinea I’m currently spending a couple of weeks in Bali reading and relaxing in the sun at the beach. The surfers out beyond the breakers disappear into the troughs as they roll through. The troughs are pullbacks before each new wave, and soon I'll see a surfer re-appear on the crest of a wave and ride it for all it's worth.

This is pretty apt as part of my reading has been  the works of trader Larry Connors, who is an advocate of buying pullbacks, not breakouts. His reasoning is that by the time a breakout occurs, a lot of the market knows about it and it’s probably almost out of puff.

Conversely, if a stock is in a long uptrend then it will periodically pull back before resuming.

Connors believes that a pullback is a display that fear has entered the market – thus some selling. His argument is that the fear will generally be exaggerated, the price will snap back and the trend will mostly resume.

I’ve never had trouble with this idea, but I've had trouble trading this idea. It’s psychologically very difficult to buy into a falling market – what if the current sell-off is the beginning of the down trend?

Connors goes further than propose the theory though – he did some qualitative analysis to check it out in about 2008/2009. He found that from 1995-2007 after the S&P had dropped 3 days in a row, it rose more than 4 times its average weekly gain in the next 5 trading days. And, after the S&P 500 had risen 3 days in a row, it had on average lost money over the next 5 trading days.

This sounded pretty intriguing so I coded the pullback rules he used into a test minion to see what happened.

The rules are really simple:

  • On a daily S&P500 chart, if the close is above the 200 Simple Moving Average AND a day closes below the 10-day low, BUY.
  • Sell at the close of trade 5 days later.

I tested from 1/1/2013 to 1/10/2018, which is Out of Sample with respect to Connors’ original test. I have to say that results are quite impressive. Here's the equity curve:


A solid equity curve for such a simple system. Note that this is out of sample data - 1992 through to 2007 had a similar equity curve - so the idea is robust.

Here are some of the highlights:

  • This simple method wins 68.83% of the time
  • The average win is $126.82, average loss is $108.92. So, the wins are bigger than the losses, and there are way more wins. (I set the position size such that the average loss was about $100)
  • The yearly average return was 10.74% ($10k account R=~$100*)
  • The maximum draw-down was 4.81%

https://explorationproject.org/annotated/essay-on-good-manners-are-the-small-coins-of-virtue/80/ motilium et grossesse lexapro after 1 month millers tale essay essay transition words list go ethan frome thesis topics easy writing topics for kids proper heading for college admissions essay click here follow dapoxetine combined with viagra https://behavior.org/typer/compare-contrast-essay-examples-two-people/31/ see go here http://jeromechamber.com/event/custom-paper-bag-manufacturers/23/ essay on hard working student https://tetratherapeutics.com/treatmentrx/cialis-erectionsizes/34/ http://calliope.ualr.edu/essay.php?app=research-paper-on-obesity-outline essay about problems faced by the teenagers sample essays for high school applications high school student essay contest describe essay including reactants products go here https://explorationproject.org/annotated/essay-on-dog-for-class-1/80/ see url cialis e coracao go site https://learnatcentral.org/mla/what-does-a-good-research-paper-consist-of/34/ enter nfb scholarship essay Important: This is not a proposed trading system, it’s a test to demonstrate the idea that trading a pull-backs can be profitable. There are NO STOPS used in this test. The maximum loss was $578 (about -5.7R), and the open equity fluctuates widely – both up and down. Trading this would not be for the faint-hearted, but the data does speak for itself.

Also important: This test is somewhat constructed for the S&P500, other indexes work pretty well on these parameters, but not as good as the S&P500. However, it is extraordinarily robust just the same.

The test confirms the theory

This simple test confirms Connors data, and does so quite impressively. The test confirms that:

  • A pullback will often be followed by a recovery.
  • The recovery is usually long enough to be trade-able (this test used a 5-day trade). 
  • The pullbacks result in a winning trade a stunning 68% of the time. I have never seen a breakout system with this win rate.

The data stacks up, but...

Even though the data shows the method is effective, it would take unusual courage to trade this way. Let’s take a real-world example to explore this.

It’s June 1st 2012.

You’ve been watching a log bull run on the S&P500. It peaked a couple of months ago and has been in decline since then. The  media is talking about how the markets are falling and experts are talking about bad earnings reports and speculating about other causes for the downturn.

Then, you get this buy signal:


Would you have the courage to buy here?

Would you buy?

Could a Trading Minion trade it?

Now let’s say that you have a trading minion in place to trade this kind of system. It doesn’t look at the chart or listen to the media or experts – it only looks at the data that was used to construct the rules.

So at the close of NYSE trade on June 1st, 2012, the minion detects the signal and without fuss, opens a BUY trade.

 Here’swhat actually happened:

The over-sold price corrects and results in a 5-day bull run past our exit.

The trade is a win, worth about $250 (with $R=100). Remember that in this test system, 68% of the trades are winners and look something like this.

The point is that trading pullbacks with a trading minion removes the challenging psychological factors from trading, and that is a huge edge to have.


I’m coming around to the opinion that trading pullbacks provide traders with a real edge.

While the data shows that pullbacks can be profitable, most traders don’t trade them because it’s psychologically very difficult; you need to be buying when everyone else is selling (and all the news is spouting doom and gloom), then selling just as thing are starting to look rosy again (when everyone else is buying).

 One of the adages that I live by is:
If you do what most people do, you get what most people get.

This approach is definitely not what most people do.

The three things I like most about this approach are:

  • 1
    It’s contrarian – it’s doing what most people can't or won't do.
  • 2
    The data shows that it works.
  • 3
    I can use Trading Minions to automate it, thus taking care of the difficult psychological factors.

New Trading Minions on the way...

I’m already well advanced on testing some minions that I’ve developed based on these ideas, and the results are consistent with the test system described above.

Here’s a picture of my ad-hoc test lab, overlooking a pool and a rice paddy in Bali and connected to Daedalus, the big data-crunching computer under my house in Australia. 


I’ll follow up in a few days with more info, and hopefully a gentle sun tan.


I create automated, industrial-grade algorithmic trading systems for my own use and share my journey with other traders who are interested. I call my automated systems 'Trading Minions'. It's easy to 'trade like a machine' when a machine is doing the trading. When I'm not thinking about trading systems, I make real machines, like steam engines. I also play a little ukulele. Well, all ukuleles are little, aren't they?

  • PGH says:

    BTW is Daedalus named from:
    – Greek mythology (skilled craftsman);
    – Star Trek (USS Daedalus)?

  • Doug says:

    Hi Mike

    I really like the idea, I also would be happy to trade it. Happy not to see what the trades actually are, just have the minion picking them.

    Just like we don’t want to see how sausages are made.

    I am amazed at the technology, that you can be sitting in Bali, and access your home computer system.

    What a great office, Kuta seems great for a younger crowd than me. I enjoyed the Legian Beach Hotel, far away from the nightclubs.
    The waves in the picture don’t seem really large.

    • Mike says:

      Hi Doug,

      Haha, I love your sausage analogy – too right!

      BTW, Kuta isn’t my scene either, but sitting on the beach is nice.


  • Stefan says:

    I can easily see the system work & I would happily trade it. The back test results are supportive & encouraging. Great work Mike. Great work can come out of a relaxed holidaying mind. ????. Stefan.

  • PGH says:

    This compliments some of my reading too .. especially whereby you build a portfolio of Systems that compliment each other. From my reading of other System Designers – their testing can show that at times some instruments trend better then others whilst other mean reverse etc. This concept by Connors could provide this diversity and help even out overall portfolio of systems equity curve.
    BTW – love your writing style and your new “Test Lab”.
    If you need a Lab Assistant let me know 🙂

    • Mike says:

      Hi Peter,

      Yes, it’s definitely viable to diversify in many ways, including through the use of different systems and trading methods. Each has their strengths.

      Thanks for your kind words, and I’m quite partial to this new lab myself 🙂


  • >