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

American Capital Agency Corp. Message Board

  • reits_r_us reits_r_us Feb 9, 2013 12:38 AM Flag


    Hi Everyone,

    I've been data mining again utilizing some unique patterns with PSEC I mentioned in a thread about one year ago. That fact was that PSEC runs to EX from(best fit(gain) average) 10 days prior to EX. I have combined that fact with the higher high, higher low data from my FSC thread to look at the following historical runs to EX from 10 days out.

    PSEC began monthly dividends in June 2010. Prior to that they had Q'ly dividends. Here is what I found for the past 31 months(June 2010 through Dec 31, 2012).


    Count back 10 days from the EX date and buy on the close. Only trade if the day's high and the day's low is higher than the previous day's high and the previous day's low. If not, wait for the next close which fits that criteria. If no dates fit through EX date, do not trade that month(every month traded on this 31 month historical look back).

    Hold for .20 cents each month. Place limit order for 20 cents above entry price immediately after execution of the buy fill.


    1) Buy and Hold:

    Initial PPS June 14th, 2010 ...$10.50.
    Final PPS Dec 31, 2012 .......$10.87
    Dividends for buy and hold 30 x .101 = $3.03
    Capital Appreciation $0.37

    Total return $3.03 + $0.37 = $3.40 for buy and hold
    Days in market 613

    2) Ten day method:

    $3.00 total gain over the 31 month period
    Days in market 131


    Although Buy and hold did better than the ten day method I think we should look more closely at the data. Of the 7 trades in 2010 there was one losing month of -.05(5 cents). All other months won with a total gain of .98 for the 7 months. In 2011 it was a rough year. The PPS itself lost almost two dollars(10.93 to 9.29). Of the 12 trades that year we had 9 winners, and three losses of -.06, -.47 and -1.04. The 1.04 loss was not good and ended up the year with a +0.24 overall profit. Without that one loss we would have beaten buy and hold.

    Buy and hold, when adding the dividends for 2011 had a -.43 loss for the year.

    2012 was better as there were 10 winning months and two losses at only -.02 each for a profit of +1.78 on the year for the ten day, whereas buy and hold had a great year between CA and dividends at +2.65 profit.

    The point of this thread is to point out the relative "sameness" between buy and hold and the ten day method for profit and yet the incredible exposure difference between the two. The Ten day Method was in the market 131 days over 31 months or an average of 4.22 days/month.

    Buy and hold was 613 days in the market at an average of 20 days/month, almost a 5x exposure difference.

    $3.40 over 31 months for buy and hold on an initial 10.50 PPS yielded a 11% annual return, which is close to the stated return for PSEC. The deal is(on the Ten Day Method) is that you are making $3.00 or 1.16/year(31 months) based on only 4.22 days per month or 50.64 days per year. There are approximately 252 trading days per year, so your real return is 55.7% on an annual trading day basis.

    That beats the heck out of buy and hold when you are trading in a tax sheltered account.

    Happy Trading,


    SortNewest  |  Oldest  |  Most Replied Expand all replies
    • Great play Jim!

      I have an order in to match for the same.


    • You beat me to it, PSEC was going to be one of my weekend projects.

      Three hours later: dbx=day before xdividend, xdd=exdividend day
      Ok, using a rote algorithm of shorting on dbx and counting on a drop in excess of the div I found that covering on xdd or xdd+2, only a little profit was made after covering the shorted dividend. Since oct2010 in 28 months of trading, a profit of 1.21. That is only the icing that delivered an average enhancement on the trade of 0.043.
      Buying the stock at the same time as covering the short, holding long and then selling on dbx gave a profit of 6.81. Total profit 8.02 over 28 months for an average of $0.286 per month. Almost threefold over the div. This algo did not have a profit reinvest component which would have provided a significant compounding effect.

      There were some criteria on when not to short and alternate conditions for covering.
      The main short criteria was the opening price on dbx plus $0.03. If the price did not rise by at least .03 the stock was not shorted. Any stock held for capital appreciation was sold at the close on dbx.
      For 2012 there was very little price gain from the open on dbx.

      So then I looked at an alternative to months in which the short criterion of opening price on dbx+3 was not met. On the ten instances in which it was not met in the 28 month test period, I then shorted at the close on dbx and covered according to the algo. Two instances resulted in a loss. The total gain was .49 for the ten months. A .049 enhancement.
      The extra .49 per share added to the 8.02 gives a yield of 8.51/28 = .303per month average That is a threefold return compared to the average divy over the test period.

      The criterion for covering was the opening price on xd minus $0.09. If the price did not hit the criterion, then the cover was done two days later at open minus .01. There was one exception in which I looked at price momentum and covered much lower. If the chart is babysat for the day looking for the bottom, I cover at .03 on the upturn after bottoming. In the mileau, though, a headfake will appear to be a bottom, so this concept can not be back tested.

      Shares held long until the dbx were acquired at the same price as the short cover.

      I say it is a rote algorithm because it follows rigid rules. In reality a trader's sense will look at how the stock is behaving. Using Doc's second derivative analysis on price movement, several of the long positions might not have been initiated particularly in the summer of 2011.

      • 1 Reply to slegermark
      • Hi Slegermark,

        Thank you for your work! PSEC is an awesome stock to trade. On a quick check I roughly counted 5.03 profit, buying on the day after EX on the close, and holding until the close of the dbex, from Oct 2010 through Dec 2012. Is that right?

        "Using Doc's second derivative analysis on price movement, several of the long positions might not have been initiated particularly in the summer of 2011"

        Did you understand the criteria of the second algo? You count back 10 days from EX. If that high, low, close does not fit the 20% and 2.5% rule move forward on a daily basis towards EX date until the next close fits. Summer of 2011:

        Enter June 22 @ 10.17 out on dbex @ 10.07 for -.10
        Enter July 20 @ 9.89, out July 22 for +.20
        Enter Aug 16 @ 8.94, out Aug 17 for +.20

        I love the collected trials...between all of us we've got to find the best fit...;-)



    • Hi Doc,
      I want to thank you for your many cogent and informative posts.

      Just a heads up about PSEC is that last February they announced a SPO on February 22 while the
      EX was on February 27. The SP then was also at about 11.50. The last SPO was on November 1 so the timing for another SPO at the end of February is about right.

      I bought after the last SPO at about 10.50, have collected three dividends and plan to sell at around 11.50 to avoid the possible SPO.


      • 1 Reply to ronfal
      • Yes Ronfal, agree!!

        11.90-11.92 has been the magic range for the last two SPOs. I have a sell limit on my shares for 20 cents this month and I hope to get hit on Monday and then am waiting for the next SPO. Congrats on your last purchase. My plan is to quadruple down on the next one...;-)



    • Addendum:

      If you follow the complete rules I outlined for FSC, that is the 20% and 2.5% rules, then the loss of the 1.04 in June 2011 is reduced to a loss of .10, making the total gain over the 31 months at $3.94 for the Ten Day Method, exceeding the gain of $3.40 for buy and hold. The FSC rules leave the other numbers unchanged for all of the other months, changing the return from 55.7% to 72.3% annualized for 252 trading days/year.


19.97+0.13(+0.66%)Aug 23 4:00 PMEDT