Strategy 2 – Monthly End-of-the-Month (MEOM)

The 2nd strategy is a the Turn-on-the-Month implementation applied to a universe of 26 ETF’s. Excellent risk vs reward results are achieved, reflected by an annualized sharp ratio of 5.56. Details below. Signals of this strategy will be provided through this blog starting  end of June 2010.

Strategy Concept
The seasonal strategy is based on the Turn-of-the-Month anomaly. This effect is discussed on various forums and Michael Stokes  recently confirmed its validity in his analysis.

Universe
The strategy will be applied to a universe of 26 ETF’s representing a broad range of indexes across various asset categories. ETF’s are used as they show lower volatility compared to shares and offer low-cost access to index trading for small investors. ETF’s included in the universe are: DIA, EEM, EFA, EWH, EWJ, EWT, EWZ, FXI, GLD, GSG, IEF, ILF, IWM, IYR, QQQQ, SPY, VNQ, XLB, XLE, XLF, XLI, XLP, XLU, XLV, XLY, XLK.

Entry & Exit
The strategy enters a long position, at the last day of the month, in ETF’s that are trading above a medium term moving avarage (WMA89) of the closing price. It will hold its position for a maximum two days or closes the position after a upday.

Money Management & Ranking
Initial Equity is assumed to be 10.000$ as we are looking for to run strategies for small investors. The equity is split into 2 positions of 5000$. Going for smaller positions would increase transaction cost% to an unacceptable level.
To determine what two ETF’s are traded, ETF’s are ranked by a calculation that multiplies short-term 2-day returns with long term 2-day returns.  The top 2 ETF’s in that ranking are selected.

Results
Below is a graphical summary of the simulation over the period 1-1-2003 through 22-6-2010. Transaction fees based on Interactive Brokers fee structure have been included. Slippage costs have been kept at 0% as the strategy uses MOC and MOO orders.

23 thoughts on “Strategy 2 – Monthly End-of-the-Month (MEOM)

  1. Hello QD – Michael from MarketSci here – just found your blog. Always GREAT to have more quantitative types in the blogosphere. Love what I’ve seen so far and looking forward to more big things. ms

    • Hi Michael,
      Thanks for the encouraging words. I am pleasantly suprised to see that you have found my Blog. As I am still in the phase of setting up I was not yet prepared to see visitors of your posture 🙂
      I expect to post soon some more progress and directions I want to take for the Blog. In the mean time I am lurking in your Blog as well, a true source of inspiration for me.
      Regards, Michel

  2. hi.
    great work….came over to your site from marketsci.

    i was looking at this strategy and wanted to ask if “short-term 2-day returns with long term 2-day returns” is correct. Did you mean to say 2 month.

    I am just starting out in this field…most helpful stuff is the blogs like yours and marketsci. Thanks for your efforts.

    • Hi, thanks for your question.
      When talking about short-term 2-day returns I actually mean: a short term average (Mov Avg) of the 2 days returns (c/ref(c,-2) ).
      In amibroker formula language: ma(c/ref(c,-2),5)
      Hope this helps, Michel

      • Thanks that makes sense. I was using ROC(C,2) for short term return and ROC(C,40) ( approx. 2 months) for long term.

  3. Hi, I don’t understand this average:WMA89, 89 is period? if 89 is average period why89?
    Best reguards

    • Hi Luka46,
      WMA89 is the weighted moving average with 89 as period. 89 was selected as mid-term oriented period. I could have selected 80, 90 or 100 but choose 89 as it happens to be a fibonacci number (an old habit).
      Regards,
      QD

    • Hi tstudent,
      At the moment I do not plan to share the code publicly. I think most of the logic is available in the posts. From here I think you should be able to build a model that resembles the MEOM strategy in AB (or any language that you master). Send me an email if you get stuck with certain topics and I can give you some more hints & tips.
      QD

  4. Not sure what this means exactly: “It will hold its position for a maximum two days or closes the position after a upday.” My interpretation (in an example): If position is opened on a Wednesday at close, you exit on Thursday at close, if Thu close > Wed close; else, you wait until Fri close to exit. That is how I read it, but then your avg. days in trade stat is 2.3 days, which suggests a longer hold time than your rules allow (max. 2 day hold). What am I missing?

    • Hi Evo34,
      Regarding your questions, it will exit on Friday open, when thu close>thu open. The 2.3 days is caused by additional days when bank-holidays do not allow exiting on first or second of the month.
      Hope this helps,
      – QD –

  5. May I repeat Evo34’s question? I too read your rules to mean a maximum 2 day hold. I assumed you enter MOC ont the last day and exit up to two sessions later but maybe one session if that is a positve day.

    Also what short term calculation is used? I am not asking for code but just the math you are using.

    Thanks for your great work

  6. Hi, first of all great work. You and some other Quants from your blog roll are doing a great job.
    I am constantly looking for similar short term approaches like RS2/DV2, seasonals etc.. . So it is great new input for me in terms of what do others to generate some edge to widely known trading systems. I fully understand the approach to filter for the most dynamic markets (e.g. ROC(2) vs. ROC(40)) as well as for in general uptrending marktes (MOV(89)) as seasonality trade signals alone might be tricky. If think in addition you try to higher the win ratio by the early exit rule (Exit on Upday), but as far as I understand this, it should worsen the profit/trade and the cagr? It was always the case when I tried things like that in the past.

    Thank you,
    lantama

  7. Hello.
    Thank you. I enjoy your site and am working along similar lines. My questions is how did you determine your ETF universe? I could not see this explanation, but apologies if I missed it.

    Regards

  8. Hello QD,

    I recently came across your blog and I must say it is very informative. This particular post got me thinking and while I understand the premise behind the strategy, I am a bit confused about the *ranking* criteria.

    You multiply a moving average of two Rate-of-Change measures (one short and the other longer term). What happens when both measures are negative? The multiplication is positive, which could also be the case when both measures are positive.

    For instance

    ROC1 = 2 and ROC2 = 5, then ROC1 * ROC2 = 10

    Now say if

    ROC1 = -2 and ROC2 = -5, then ROC1 * ROC2 = 10

    Should both situations receive the same treatment in terms of ranking? I must be missing something in the logic. I would kindly appreciate your feedback.

    Keep up the great work here!
    sp

  9. Hi QD-
    Just discovered your excellent blog.

    The MEOM ranking description says it combines a short-term 2-day return with a long term 2-day return.
    But shouldn’t the long term return be more than 2-days?

    George…

  10. I implemented the system for the S&P500, using data from 2007 to 2015.

    Looking at the performance report, I wouldn’t recommend trading this system with the aforementioned entry/exit rules.

    I guess, the entry/exit rules aren’t robust enough to withstand the change of time.
    Though, I think that the turn-of-the-month (TOM) effect still exists.

Leave a Comment