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

American Capital Agency Corp. Message Board

  • reits_r_us reits_r_us Mar 16, 2013 11:45 PM Flag


    Hey Everyone,

    Hope your weekend is going well. I am bringing this continuing conversation on entry and exit points for ARR forward as the other thread is getting crowded.

    There was some back and forth about which entry and exit dates were best on ARR between Jess and me and I agree that the best fit exit date historically is X+17 as Jess pointed out.

    Jess, I went back and cranked out the difference between the entry dates of X+2 and X+6 , using the close on each and I simply added those closing PPS values for the 27 months with a starting date of Dec 20, 2010 for (X+2) @ 7.60, in one column, alongside the starting date of Dec 27, 2010 for (X+6) in the next column.

    I got a total of 192.63 for (X+2) and 192.66 for (X+6). Pretty much a dead heat. More importantly though is the frequency of the hits and the std deviations of each.

    For (X+2), the number of wins was 16 out of 27. While 11 out of 27 were wins for (X+6). Important also was the average win was 10 cents/win on (X+2), and 15 cents/win average on (X+6).

    Well for others what the heck does all that mean? It means that when (X+6) won it hit a bigger payday than when (X+2) won, but it did not hit a payday as often as (X+2). If that 27 month sample size was a 100% predictor of future results, my plan would be to use (X+6) because you are in the market for a shorter time period each month, and in the end(across all 27 periods) they performed the same.

    Having said that, I would like to have a larger sample size as 11 out of 27 wins(40.7%) is significantly lower than 16 out of 27(59.3%). The fun part is that all who entered on (X+2) last Friday, can enter again on (X+6) next Thursday(X+6), if the PPS is lower....;-)



    SortNewest  |  Oldest  |  Most Replied Expand all replies
    • bump................

    • To repeat my earlier post - I'll re-enter on Monday (X+3) IF at all.

      • 1 Reply to walrus8
      • Lock and load Walrus...Yippee-ei-o! I am ready to make a boat load of cash, (or not this month)...but probability over time is Wayyyyyyy on our side.

        I kind of feel sorry for the buy and hold crowd. Think of the great loss all those millions of shares suffered that were sold last Thursday and Friday at 6.17/share....probably bought at 6.50-6.60 at the spo, last month, then dumped for a huge loss. Let's see.... buy and hold(lose money) or make monthly profits...I'll get back to you....;-)


    • I am exiting MAIN on the close tomorrow the 18th(DBEX) and plan on entry of CLM(thanks Diva!) and CFP on Tuesday, the 19th, but only if GLAD is met. MAIN will be up for entry on Monday the 25th if GLAD met. Tuesday, the 26th is FSC entry(no GLAD) and PSEC exit on 25th-26th.

      Nice, might have 4 stocks on board this Tuesday, of the Sweet Six(SS).

      Lets make some money!!


    • Thoughts on trading short term cycles
      By definition, buying at a cyclic low point and selling at the high point provides maximum profit for a single trade. Unfortunately, because we live in an imperfect world and have misfits for a congress, (just threw in congress) the absolute high and low of a given cycle are generally imprecise. However, using history for a guide does provide an edge, if followed long term. That is, history is not of much value for predicting a single cycle, but if used over long periods, it must provide gains. The trick is to make your bankroll extend long enough to make the averages kick in.
      Identifying the cycle high point allows, by definition, all cycle trades prior to the high point to be profitable. If we identify the average cyclic high point long term, it follows that most trades entered prior to the high point will be wins. Buying each trading day prior to the high point, and selling the accumulation at the high should produce a net profit over time. Since the high point for a future cycle cannot be absolutely identified, some trades will be losses. Formulating entry point criteria (like Doc has done with GLAD) so as to reduce losses can maximize profits.
      For instance, if the high point for ARR historically was determined to be EX+17 (27 periods between Dec 16, 2010 and Feb 13, 2013) , and trades were entered every day between EX+1 and EX+16, there would be a total of 432 trades. 322 would be wins, 110 losses. The net gain would be 50.62 for the entire 27 periods (0.117 net gain per trade).
      This net gain/trade can be improved by several methods. The easiest is to buy only during the first 9 trading days of each cycle (EX+1 to EX+9) and selling all at EX+17. This gets the cream of each cycle and produces a total profit of 38.60, raising the average gain/trade to 0.159 (243 trades). Combining this with a entry formula that reduces losses would be even better. Identifying the formula takes some work.
      There are some disadvantages to this system. Averages can be misleading. For instance in the 63 periods of CFP, four of the periods account for almost 90% of the gains. Buying multiple trades in one cycle may introduce significant trading costs, depending on your broker.
      Got to quit, supper calls.

      • 2 Replies to jess_1554
      • Hey Jess,

        That was like listening to one's favorite music. Well said. Just a comment. CFP did not meet the requirements for average daily volume before 2012. I went back to 1/26/2010 when looking at the returns over that three year period in order to allow a large enough sample size for observation, realizing that the results would be somewhat skewed because of volume.

        Your results are not just skewed but leads one to erroneous conclusions when you go back to CFP's inception. At that time the average volume was less than 5,000. If we look at a smaller sample size since avg daily volume has been over 200,000(01/2012), starting at 12/27/2011, we find 14 sample periods or cycles returning 100 % consecutive profitable months using GLAD/3% and X+4 as the first possible entry if GLAD met. and exiting on X-1 close.

        Results:(Every month was profitable)


        If we throw out the highest(.63) and the lowest(.02) we get 3.00 divided by the 12 remaining months gives .25 avg/month. I don't care to work out the std deviation but observing the data will make even the casual observer agree how close those remaining 12 months are to the mean.

        My point, we need to trade stocks with at least 150,000+(preferably 200,000+) volume to even out the monthly std deviation.

        Hope dinner was good...;-)


      • I should have stated more clearly that the number examples above apply only to ARR. Each stock history will produce it's own averages as to high/low points and gains/losses.

    • Bump....

    • Bump................

17.84-0.23(-1.27%)Nov 25 4:00 PMEST