Seasonal Strategy taking advantage of monthly transfers

[Winner]

grizzly386,

Why don't you ask the person who found this strategy "mstrzerg" that was post last year with the average return for the last eights years are 42% here is the link viewtopic.php?t=4992

[mstrzerg]

Is there a program you can analyze "maximum possible gains" historically, annually, and monthly taking into account our IFT restrictions? A program that could tell you what day and what funds the IFTs would be for in order to have obtained the max possible return.

And maybe creating a seasonal method that follows it.

This is actually very similar to what I programmed up. What I quickly found though is that using 10 years worth of data to create a seasonal pattern I was able to find a great pattern I should have used. Apply that pattern to future years and you end up with a huge variance. I was able to average about 5% year over year gains projecting out that way.

Basically what I'm saying is, heavy backtesting of a seasonal pattern (or any pattern for that matter) is no guarantee of future returns and after accounting for inflation you are likely to loose money in the long run.

[crondanet5]

mstrzerg your post brings up all kinds of examples problems with applying previous methods in today's environment in military history. In the early days of the bombing of London the British Air Commander had all his planes launch, then form up before pursuing the enemy. Lost engagement time. And there was the time over North Vietnam that B-52s lined up and made their turn at the same place, same time every day. It did not take long for SAAMs to pop up to say hello. And think of the lines of soldiers marched into gunfire at Waterloo and Gettysburg. And so too is the battle called Wall Street a theater where innovation and quick adjustments are the difference between falling down an elevator shaft or riding up the escalator. Great post.

[skiehawk11]

mstrzerg, in your backtest you were able to duplicate a 10 year rolling seasonal strategy to predict the next year's strategy? Let me know if I read that correctly.

[grizzly386]

Thanks mstrzerg,
I couldn't remember where I saw this but I wanted to give it a try. I will say the system did yield a 13% gain for 2012 so that was good. Now that I am getting more versed and paying attention to the various best guessors, I definately understand that there is no guarantee of future positive gains and will keep that in mind.
Just hoping to ride the escalator!!

[skiehawk11]

I have stated in the past, that seasonal and similar type strategies should used more as a road map/base.

[greengrass]

mstrzerg, in your backtest you were able to duplicate a 10 year rolling seasonal strategy to predict the next year's strategy? Let me know if I read that correctly.

Curious if you got that rolling 10 year for your weekly strategy tested? Does it look good?

Happy trading

[skiehawk11]

I'm learning Matlab and Python currently so I can test the strategy. I've had no luck being able to create the strategy in Excel.

[mstrzerg]

mstrzerg, in your backtest you were able to duplicate a 10 year rolling seasonal strategy to predict the next year's strategy? Let me know if I read that correctly.

Sort of... I really need to run it against just S&P500 data, using a stand-in for the G-fund. What I did:

Using data from 2003 to 2013 I developed a method for creating a seasonal trading pattern based on past years performance. This is where the crazy 20+% a year gains came from. Being suspicous, I tried a new strategy. If I created a pattern from 2003 data then applied it to 2004 data what would happen. Then take 03 and 04 data and apply it to 05. Continue until you run out of data. It worked great! Until 2008-09 where a change in market conditions caused the pattern to loose a ton of money AND miss the upside. Because of the crazy gains previously it still averaged about 7% through those years, but never recovered it's crazier high gains.

I messed around with this a bit more, trying different things, but ultimately have to recommend caution when dealing with backtesting. There is value there, but you have to be really careful how you apply it.

[skiehawk11]

It would be interesting to do a study of instead of just one year, to apply a general/average trend of the previous 4 years and then apply the step up method.

A conservative way to duplicate the G fund is assume .00011 growth per trading day. You can create an artificial share price from 1950 to now when you do your test.

[RGEN]

Did anyone follow up Skiehawk's recommendation of a 4 year analysis? Or the one year with a 4 year trend factor?

[Chulke]

I'd be interested in seeing/hearing if anyone is actually using or following this method of trading the TSP and how their returns are fairing...anyone? Right now I am following a sentiment based system and I like the return it shows (average return of alittle over 13% over the past 7 year period) but there are some years where it tanks badly and I would like to take as much risk out of it as possible, but naturally you can't remove all the risk because the whole market is completely unpredictable just like human emotion...so you never truly know what's going to happen. But, I'll take any advantage I can get...You have to in this day and age because lord knows everyone and every big business is going to take any advantage they're given as well, which usually ends up bad for the person just sitting there riding the wave...like the buy and hold guys. Bottom line...I'd like to have over a million (actually closer to $7M) in my TSP when I actually hit retirement age as I'm sure everyone would. So...I guess what I'm saying is...I'm looking for a competitive advantage...LoL!


Cheers!

[skysoldier]

Seeing this topic may let me save the trouble of contacting TSP - if you set aside part of the TSP account for monthly payments, are you not allowed to move the money around? The reason I ask is that I thought I saw a comment in this thread to that effect.
Thanks