Skip to content

Improving Trend-Following Strategies With Counter-Trend Entries

July 29, 2011

It is possible to make minor adjustments to strategies that can both improve their backtest performance, but also reduce the real costs of trading. The trend-follower  will always pay more to enter a trade due to the increased slippage and market impact costs. A further “tax” in the modern era across all markets is the effect of short-term mean-reversion– as markets tend to follow through less often, a trend trade will often be triggered at a short-term high or low. This leads to poor entries that constrain the potential for the trade to make a profit. The obvious solution to this problem is to incorporate a short-term counter-trend/mean-reversion entry and exit within a given trend-following strategy. The key is to use an entry that occurs frequently enough to avoid missing trades and gets in as quickly as possible. For exits, it is important to have a quick and timely signal, and it is also critical to preserve the  nature of trend-following strategies which is to reduce tail risk. In other words, we do not want to create an exit into strength that may never be filled and leaving the trader faced with a significant drawdown.

The most significant impact of a counter-trend entry exit is typically found when combining it with a short-term moving average. The value of improving the performance of a short-term trend signal cannot be overstated because it allows you to better track the trend of the underlying with substantially less lag. As the SPY/S&P500 has become fundamentally mean-reverting since 1998, most traders have resorted to using longer-term moving averages and trend filters such as the 200sma or a golden crossover. The problem is that in today’s market– that is rife with shocks and unpredictable rallies– the tail risk above the long-term moving average has gotten higher.  The cost of lag can mean a nasty drawdown, as has been experienced by most trend-followers this year.  If you can more accurately follow the short-term trend, you can reduce this risk substantially. Let’s take a look at the performance of a short-term trend strategy— buying when the close is above the 10-day simple moving average and selling when the close is below the sma. The “mean-reversion tax” is quite evident, and the actual trading costs with slippage and commission are not even factored in:

Now lets look at a counter-trend overlay on the same strategy. In this case, we will look at a short-term range mean-reversion indicator– the DVB, and set it to the shortest period for maximum response which is one period. The indicator is normalized so that the frequency of entry/exits is stable, and the DVB in contrast to say an RSI requires far less true mean- reversion to change values. In this case, we will use the bottom quartile for entry and the top quartile for exit. Thus the strategy would be to enter when both the close>10sma but only when DVB1<25, and the exit would occur when the close is <10sma but only when DVB1>75.  Effectively we are buying into the first available weakness following a new short-term trend and selling into the first available strength once that trend has ended. Remarkably, this makes the losing strategy of following the 10sma into a winning proposition:Clearly a substantial improvement– not sexy, but it gets the job done. What is most beneficial about this strategy is that it is very quick to catch rallies and exit into weakness while boosting the % winning trades to 41% from 33%, and boosting the average trade from -.09% to .73%. The difference in the two backtests would be even more extreme with real-world trading costs. But what about tail risk?  Lets take a closer look to see if that has increased:

 

It seems like there is virtually no difference in terms of value at risk (the 5th percentile of daily returns), and the worst trade for counter-trend is marginally higher. Looking at the drawdowns, we see a much clearer difference– trend following is really risky when you get chopped around. The drawdown is almost 3x higher for the conventional strategy. This method is robust to moving average length, and appears to improve performance universally. While not shown here, even a 200sma strategy is substantially improved by such a minor change in entry and exit.

About these ads
7 Comments leave one →
  1. lantama permalink
    July 30, 2011 9:09 am

    David, for me the entries look more like Mean Reversion Strategy with a Medium Term Trend Filter. But interesting anyway imo as the comination makes sense to improve results. I am trading quite similar systems in Bunds, Eurostoxx50 and S&P Futures.
    lantama

    • david varadi permalink*
      August 2, 2011 9:23 pm

      hi lantama, thanks for the comments. actually it is not a mean-reversion strategy with a trend filter because it holds the position without an exit until the trend has ended. it simply uses a delayed entry and exit that is based on the short-term mean-reversion strategy.

      best
      david

  2. July 30, 2011 2:17 pm

    Interesting take on this. It would be interesting to see the results over a wide range of parameters (do you actually adjust the length of the DVB when changing the length of the moving average? ie DVB(2) for SMA(20), etc.)

    I ran a similar study on my blog a while back, but with a different “three-phase” breakout system. The main idea was to avoid slippage by using stop-limit orders (i.e. only enter position at or better price than breakout signal). Adding an “extra retrace” parameter to the system (waiting for the price to retrace a fraction of ATR away from the breakout level) actually increased both CAGR and MaxDD.

    Futures Mag are publishing a version of that test in their August edition, but to see it quicker, you can go to:

    http://www.automated-trading-system.com/slippage-take-3-stop-limit-orders/
    ;-)

    That study did not look into improving exits (based on the idea that you just want to exit as soon as the trend breaks down), but I think there is potential in evaluating how a mean reversion indicator could be used during a trend to manage positions (ie lighten up a long during extreme overbought readings) and reduce drawdowns.

    • david varadi permalink*
      August 2, 2011 9:29 pm

      hi Jez, the articles you are highlighting are excellent and very thorough. I think that the main point of departure here is that we aren’t looking to get in close to the entry point dictated by the trend strategy (ie 10sma or a breakout in the case of a donchian channel). furthermore as you suggest, we are using the same approach for exits. the effectiveness of this strategy will be a function of the profitability of the mean-reversion itself to a large degree. In the articles that you wrote that show the actual impact of slippage it is interesting to see that you have found results in the same direction. I think the concept has a lot of mileage in a wide variety of directions. good work!
      best
      dv

  3. Emil Tingström permalink
    July 31, 2011 7:14 am

    One way I think you could capture this is by weighting a moving average so that the most recent days are given negative weights, and the longer back you go the weight becomes more positive upon till a certain point. This could probably be done in an adaptive manners so that volatility and auto correlation is always accounted for even if it’s a long term trend following strategy.

    • david varadi permalink*
      August 2, 2011 9:30 pm

      Hi Emil, excellent thought—readers are aware of this variant in the form of the AggM and AggZ indicators that you can look up on the blog. Good catch!
      best
      dv

Trackbacks

  1. Short Term Trend-Following within Mean-Reversion Trading | Adaptive Trader

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 538 other followers

%d bloggers like this: