% | $
Quotes you view appear here for quick access.

Kinder Morgan Energy Partners Message Board

  • abter1 abter1 Jul 22, 2010 2:30 PM Flag

    Trading the distribution cycle

    In a message back in May, gijoe714 mentioned that he trades the KMP distribution cycle.
    I've been mulling that ever since. This week UBS included in their monthly MLP newletter a brief historical analysis of one kind of MLP trading with the dist. cycle, which got me thinking about it again. UBS was talking about doing the opposite that gijoe did (UBS looked at buying the dist., rather than selling it), and found selling it was a loosing strategy.
    I spent some time noodling on this issue, and posted the results over at the InvestorVillage MLPs board. I have copied it here (is it plagiarism to copy yourself?)
    Like passandshoot, I was interested to see a brokerage I respect (UBS) look at distribution-chasing in the MLP world.

    UBS presented results of looking at the basic question "Can you win by just chasing the distribution, buying just before X-div day, and selling the next day (on the x-div date)". Their analysis shows (pretty convincingly I think) that no, you can't get ahead that way. Of the 26 MLPs they look at, more loose using this strategy than win in each of the 5 quarters they look at. For the Feb/March 2010 dist. payment cycle, the investor would have a total loss of 1.07% (and thats before any trading costs).

    But I think UBS is looking at it kind of backwards. The things I have read about people who encourage trading MLPs timed with the x-div dates talk about doing the exact opposite strategy. Rather than do a quick buy and sell to "capture the distribution", the people I have read talk about selling just before the x-div date, and repurchasing right after. You miss getting the distribution, but these writers argue you come out ahead. Not surprisingly, if the strategy UBS looked at looses all the time, doing the exact opposite might win. I've never been convinced of their argument, but the data suggests at least on paper it might work. I took a look at this.

    1) MY SIMPLE ANALYSIS IGNORES 2 THINGS: transaction costs and taxes. Anyone trying anything like this as an investment strategy with only 1 unit of each MLP would get killed by the transaction costs alone. But I find the analysis does help show why some people think you can win by trading to avoid the distribution.

    2) I don’t do anything like this (I don’t trade at all…I am pure buy-and-hold-forever), and I certainly don’t advocate anyone else do so. I ran this analysis just to explore the possibilities. I know some trading-oriented folks are forever looking for trading angles.

    Analysis setup:

    I looked at the same set of 26 MLPs that UBS did.

    I used the UBS data on day before x-div day and the day after for the April/May '10 distribution cycle period. I also looked at the 3 previous quarterly cycles of dist. payments. In their article UBS the used a semi-fancy "volume weighted average price" on the days in question. I can't replicate that exact weighted price for the days I need in the earlier quarters, so for the 3 earlier quarters of data I used a simpler price: closing price on each day in question.

    Approaching the end of each quarter, an investor has one unit of each of the 26 MLPs (in April/May '10 that is a total investment of $1,009, using the pre x-div day prices).

    ALTERNATIVE 1 (conventional buy-and-hold): hold onto them, collect your distributions (do not reinvest), and stay invested.

    ALTERNATIVE 2: Sell each MLP at the closing price of the day before their x-div date. Repurchase the same MLP at the closing price the next day.

    --Either way you end up coming out of the cycle with one unit of 26 different MLPs, plus some cash.


    {continued in my response to this message}

    SortNewest  |  Oldest  |  Most Replied Expand all replies