StockFetcher Forums · Filter Exchange · PORTFOLIO SELECTION AND MANAGEMENT USING RISK/REWARD RATIOS<< 1 ... 58 59 60 61 62 ... 65 >>Post Follow-up
Kevin_in_GA
4,552 posts
msg #123917
Ignore Kevin_in_GA
5/31/2015 9:57:30 PM

Before it is asked, the asset class for June is EFA.

graftonian
439 posts
msg #123923
Ignore graftonian
6/1/2015 2:10:39 PM

Thanx, Kevin

dashover
166 posts
msg #123940
Ignore dashover
6/3/2015 4:40:27 PM

Kevin, Just as long as you're putting your entire account into an asset like EFA...

Would you recommend any way of hedging? like selling if it's down 7%, or buying a 1 month put for .07 a share at 7% risk level $62...

or?

Kevin_in_GA
4,552 posts
msg #123942
Ignore Kevin_in_GA
6/3/2015 11:33:59 PM

Well, this is how I manage things like my 401k, which does not let me trade options. You may choose to do this as a hedge, but I do not do this and am okay with the risk levels and performance over the past four years. Remember that this is a strategy that one can certainly customize.


dashover
166 posts
msg #123945
Ignore dashover
6/4/2015 11:29:29 AM

Kevin, do you have the current Sratasearch backtest with graph? from 2000-2015

Thanks!!!

begotti
2 posts
msg #124193
Ignore begotti
6/26/2015 12:54:33 AM

kevin

I feeling like I been following you around for years. in fact I think you were apart of the HSM forum community for awhile. thank you for all your contributions over the year.

shillllihs
3,544 posts
msg #124205
Ignore shillllihs
6/26/2015 3:05:23 PM

Me too Kevin,. Luv you!

You've been displaying your systems so long,
i'm sure you're a mega-millionaire by now!!
But i bet TRO has you beat!

My man crush...

Kevin_in_GA
4,552 posts
msg #124222
Ignore Kevin_in_GA
6/29/2015 3:31:43 PM

Fetcher[
symlist(spy,iwm,efa,agg)
sort on column 5 descending
set{start,date(20150331,close)}
set{ch,close - start}
set{sort,ch / start}
set{sort%,sort * 100}
add column sort%
]



Given the trouble in Europe, EFA is currently lagging behind US Equities, with only IWM showing a positive return at the moment. I would wait until 7/1 to see how the Greek Tragedy plays out (they never end well), but this month is one that requires caution. It might be that all asset classes have negative 3 month returns (nowhere to hide).

I moved to 2/3rds cash two weeks ago and will wait until the start of next month to see where the filter lands. Good luck to all.

Toad
20 posts
msg #124415
Ignore Toad
7/19/2015 12:49:40 AM

I realize I may be beating a dead horse with this, but I found this idea compelling enough to test out myself. What I found was quite disappointing and I am not sure why it differs so much from other backtest results presented way back in this thread. I realize the period I am using is different so my results will be different...but the magnitude of the difference is alarming to me and it seems that a substantial number of people are using this, so I feel it is imperative to share my findings. I am going to dig through my formulas tomorrow to make sure I don't have an error somewhere (I very much hope I do have an error), and will upload the excel document…somewhere, but what I have found is as follows:

All testing was done for the period between 3-26-2004 to 6-1-2015 since this was the largest period I could get data for all funds (SPY, AGG, EFA, IWM). My testing was done in Excel using the solver to optimize, so the optimal solutions may or may not be the absolute optimal for each grouping (i.e. the solver may have converged on a local maxima instead of the global maxima), although I do believe the parameters are at their global maxima so are optimized. The basic parameters for my testing were as follows:

- Each 21 days, the filter was run and the appropriate fund selected and held for the 21 days following the run date (basic concept of the present system).
- This was run for each of the 21 possible starting days (i.e. one test run started at 3-26-2004, one started at 3-29-2004, one started at 3-30-2004, etc.)
- My optimization sought out the highest low return of these 21 possible outcomes. My reasoning for this was simple - by doing this I am both maximizing the possible return as well as minimizing the standard deviation (luck with the reallocation day you use).
- The variables used were: 6 month return, 3 month return, 2 month return, 1 month return, 0.5 month return. The fund determination function is simply adding the percentages of each (i.e. %1*a+%2*b+%3*c+%4*d+%5*e with %1+%2+%3+%4+%5 = 100%).
- Dividends were not factored in.
- Note that by account balance I am stating it as a percent of the starting balance (e.g. 200% means account grew by 100%).


I started by looking at the SPY, AGG, EFA, IWM which seems to be what most people are using. The results from this combination were (to me) very disappointing and are as follows:
- The optimal look back to use is the 3 month and nothing else (same as Kevin determined)
- The average account balance of the 21 possible reallocation days was 203%
- The standard deviation was 50% <-- disappointing to me
- The max balance was 350%
- The min balance was 142%
- The lowest period (21 day) return was -33% <-- OUCH!
- The highest period (21 day) return was 17% <-- YAY!
- 8 of the possible reallocation days had balances over 200%, only 2 had balances over 250% and subsequently 300%

To me this indicates that you need a fair amount of luck to get a decent return using this set of funds (>100% return over approx. 11 years)...you also might see some nasty drawdowns, and if you reallocate on the worst days over the course of many years you will get hosed.


I then looked at just SPY-AGG as Kevin indicated being better. The results from this combination (to me) look much better as well and are as follows:
- The optimal look back to use is 70% 6 month and 30% 3 month (a little different than Kevin is using with 100% at 6 month)
- The average account balance of the 21 possible reallocation days was 202%
- The standard deviation was 22% <-- much better than previous
- The max balance was 266% <-- Only two possibilities higher with the 4 fund variation
- The min balance was 176% <-- Even with shitty luck (me) you will be ok
- The lowest period (21 day) return was -16% <-- Sucks but won't kill you
- The highest period (21 day) return was 14%
- 11 of the possible reallocation days had balances over 200%, only 1 over 250% (the max)

To me this indicates that you still need a fair amount of luck to get decent returns using this set of funds (>100% return over approx. 11 years), but you are much more likely to avoid large drawdowns, and even if you reallocate on the worst possible day over the course of many years, you will still do ok.

As a comparison (not accounting for dividends):
- Hold SPY 3-26-04 to 6-1-15 --> 190% balance
- Hold IWM 3-26-04 to 6-1-15 --> 218% balance
- Hold EFA 3-26-04 to 6-1-15 --> 144% balance
- Hold AGG 3-26-04 to 6-1-15 --> 105% balance

From my results so far, I am not compelled to subject my funds to this strategy (SPY-AGG) yet, and certainly won’t be doing the SPY-IWM-EFA-AGG. To me it seems that pretty much all it will do is not subject your account to huge drawdowns that just holding SPY or IWM would. Better from a psychological point of view, but not really making you any richer, and not really any better than a diversified portfolio which should give even less drawdown. I hope to find an error somewhere in my study and will update this if/when I do. To me reallocating funds based on recent performance (for an ETF) makes sense....and people have backtested this before me…so there has to be an error somewhere in my study...hopefully.

Toad
20 posts
msg #124420
Ignore Toad
modified
7/19/2015 2:47:18 PM

Well, I am stumped...I can't find an error in my study. If anyone would like to take a look at the study, you can download the excel file I created/used for it here:

http://www.mediafire.com/view/1z1ngbyn7pjjlqd/Kevin's_Mutual_Funds_Screener_Validation.xlsx

I hope there is an error in there somewhere...I can't find one if there is though. I even tried smoothing the selection ranking out by averaging the ranking from the reallocation day as well as the previous 3 days...this helped to reduce the standard deviation some for the 4 fund variation but didn't change the two fund variation much.

For anyone who decides to look for themselves, the optimization results are summarized around cell AJ1054 in all sheets. The raw close data (obtained from yahoo finance) is in the first columns (B-E), and the various "look back" periods are in the columns next to the close data (G-Z). Note that I manually modified some of these for IWM and EFA to account for the splits those funds had. The manually modified cells are highlighted in green. IWM had a 2 for 1 split on 6-9-05 and EFA had a 3 for 1 split on 6-9-05. Ranking formulas are the next set of columns (around AC), and the fund selection screen and resulting return for the following 21 days for each possible start is shown in the following set of columns...I have shown the fund that should be selected, the period return from the following day to 21 days forward, and the cumulative return (really account size not return).

The first sheet is the SPY-IWM-EFA-AGG, the second sheet is the SPY-AGG, the third sheet is the SPY-IWM-EFA-AGG smoothed, and the fourth sheet is the SPY-AGG smoothed. I didn't spend much time trying to optimize the smoothed ones since at that point it seemed to me this strategy was a lost cause.

If anyone takes a look at this and sees an error please let me/everyone know...until then I remain unconvinced that this strategy has merit which is odd considering there seems to be a good number of people using it. Believe me, I want this strategy to have merit and there to be an error in my study but I just don't see it...perhaps I am misunderstanding the basis for the screen but it really is quite simple so I don't think that is it.

If for some reason someone is unable to obtain the file I have uploaded, please let me know...I am not sure if mediafire has a time limit on uploads.

StockFetcher Forums · Filter Exchange · PORTFOLIO SELECTION AND MANAGEMENT USING RISK/REWARD RATIOS<< 1 ... 58 59 60 61 62 ... 65 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus