Systematic Trading research and development, with a flavour of Trend Following
Au.Tra.Sy blog – Automated trading System header image 2

Monthly vs. Daily Trading – updated

March 13th, 2012 · 5 Comments · Backtest, Equities, Futures, Strategies

EndofMonth CraftyGoat

2011 offered challenging market conditions for trend followers (as illustrated by both the State of Trend Following and the Trend Following Wizards reports). The second half of the year in particular threw a few “curve balls”, in the form of wild volatile moves with inter-market correlations spiking up.

With this backdrop, I wanted to revisit the daily vs. monthly trading comparison run a bit over a year ago, to check the impact of these volatile moves against a system trading on a monthly basis (2008 had proven “costly” to the monthly trader; would 2011 be similar?).

DAILY V. MONTHLY GOLDEN CROSS

As a reminder, in that post, I compared two versions of a moving average cross-over system (Golden Cross) tested on daily data. Both versions generated the same trading signals, but the first instance executed signals on a daily basis (as new signals were generated) while the second instance executed them at the end of the month.

The results did show that there was little difference in the monthly version versus the daily one, with alternating periods of under-performance and over-performance; the daily version ultimately coming out on top.

Re-running the test with updated data (until the end of last month) still gives similar results:

Performance Stats Daily Monthly
CAGR
18.02%
17.09%
Max DD
36.41%
36.25%
MAR 0.49 0.47
Trade Number 774 677

 
Daily vs. Monthly Equity Curves 2000-2012

Note that I “deleveraged” the system to reduce the drawdown (from 50%+ in previous test) to more manageable levels, so the performance numbers are not directly comparable.

IMPACT OF EXTREME MARKET CONDITIONS – ZOOM INTO 2011

One of the empirical observations from last year’s run was the impact of 2008 and its dramatic market action, on the two systems. During that time, the daily system clearly over-performed the monthly one. One possible interpretation is that the monthly system was “too slow” to follow the extreme moves of that period.

Did the same thing happen in 2011?

By zooming in on the 2011 performance, it appears that the answer is.. No.

Performance Stats Daily Monthly
CAGR
-10.79%
-8.99%
Max DD
26.93%
28.25%
Trade Number 123 114

 
Daily vs. Monthly Equity Curves in 2011

Well, let’s declare it as a tie between the two: similar results for both systems, but the monthly edges out the daily on the CAGR stat, while it is the other way around for the Max Drawdown stat.

As I was preparing this article, I came across this video from Mebane Faber about his Tactical Asset Allocation Model. In one of the points, he highlights a similar over-performance of his monthly model vs. the daily and weekly ones.

A possible explanation could be that monthly trading allows the system to side-step some of the whipsaws brought by short volatile trend reversals (although – surprisingly… or not – the trade count does not vary a lot between the two versions).

MONTHLY TRADING A VIABLE OPTION?

So it seems that “monthly trading” still held up as a viable option through 2011. Of course the same caveats from last year’s post still applies: most importantly that these results are only derived from a single test for example.

Additionally, a very good point made in a comment by reader Pumpernickel on the original post highlighted the fact that the futures back-adjusted contract time series used in the back-test did imply daily “roll-over” monitoring and trading, since they can roll any time in the month (whenever liquidity shifts). Additionally, monthly roll-over might be all but impossible for these financial instruments that completely roll over a few days every month.

So, monthly trading of futures might not be a practically viable option. However, the main take-away from this comparison is that the trading signal execution frequency does not seem to have a strong impact on the results.
It could be assumed that the same stands for weekly trading, which was suggested by other readers and might represent a better compromise for futures trading

Yet, another alternative could be to switch from trading futures and trade ETFs instead, a subject covered in Anthony Garner’s practical guide to ETF trading systems. Futures act as good proxies to their corresponding ETFs – something discussed in the book and tested here – so the monthly-versus-daily comparison in this post should be transposable in the world of ETFs.
 
 

Picture credits: CraftyGoat via flickr (CC)
Related Posts with Thumbnails

Tags:

5 Comments so far ↓

  • Ross

    Jez,

    I enjoyed the follow up article on monthly vs daily trading. I would be interested to see the results for signals generated by the close price crossing a moving average, instead of a moving average crossover. I would expect to see a significant difference in # of trades and subsequently transaction costs. I have run that test on a handful of FX pairs and equity ETFs, but would be very interested to see the results using your larger portfolio with futures.

  • Investment Warrior

    Moving average penetration alone vs. moving average crossover system of similar time horizons will produce more whipsaw trades and will trade more often.

    The offset is that moving average crossover systems tend to be slower, and will have a greater drawdown from market tops. There is always a tradeoff when it comes to MA systems, whether it be the time horizon you determine, or frequency of trading.

    The more trades though, the more slippage, and of course trading costs.

    Fascinating that trading once a month is nearly as effective as daily monitoring. However, I figure that a 100 year “black swan” event coming in the middle of a month would change that result in favor of daily.

    On the other hand, since market timing really is about avoiding large losses as its strength, I have yet to figure out a system that will defend against the “End Of The World”! LOL!

  • Jez Liberty

    With regards to the MA penetration system vs. dual MA cross-over system, this is more a system-specific question and I prefer to stick to more generic concepts with this blog.

    Regarding the 100-year black-swan event, entering with stops in the market to protect positions (as mentioned by Pumpernickel as a comment in the initial post) might be a possibility to mitigate against strong adverse moves.

  • Miroslav Krajcir

    Hallo Jez, i am interested in the risk management used by you.

    So actually you didnt use any SL or exit just reverting the positions? How many contracts were you buying? (5atr= 1% of bankroll?).

    Any limitations on number of positions, or is it theoretically possible to be 50 positions at one moment?

    Thanks. Mirec

  • Jez Liberty

    Mirec – yes, just “plain vanilla” MA cross-over system, which is always in the market (ie reversal system). The position sizing was 0.5% of 4ATR.
    Re: the number of positions, it should stay by the nature of the reversal system (always in market for all instruments)

Leave a Comment