TSP backdating date range

General TSP Discussion.

Moderator: Aitrus

Post Reply
mill0168
Posts: 2
Joined: Tue Sep 22, 2015 12:58 pm

TSP backdating date range

Post by mill0168 »

So I am currently considering daily seasonal calendar #16517 but i have a question about why we are using the entire 2004-2017 data instead of optimizing 2010-2017 data. As far as I know seasonal strategy are trying to find seasonal trends that exist from all our backdata. My assumptions are that the closer to the current data the more relevant it is and there is no way to predict market crashes like 2008-20009. Right now our data includes the big anomaly of 2008 into our predictive model which we aren't trying to predict. For example if there was a big crash in Jan 2018 then all of our strategies would decrease the funds that crashed in 2018. But this is no indication that in Jan 2019 there will be a similar trend. Is there something wrong about trying to only predict bull market scenarios (while stilling being aware that the crash could happen). I have an excel file that shows weekly average returns from 2004-2017 and 2010-2017 with some notable differences in the trends. Am I wrong to try and optimize 2010-2017 data and then reacting to the next crash when it happens?

EpicWin
Posts: 45
Joined: Fri Sep 10, 2010 10:17 am

Re: TSP backdating date range

Post by EpicWin »

I think it makes sense to exclude outlier years.

Chulke
Posts: 420
Joined: Thu Aug 01, 2013 10:25 am

Re: TSP backdating date range

Post by Chulke »

I can agree with what you are getting at but, If you remove the low end bad year of 2008 from you data sets then you also should remove the high end year too, whichever year that might be. That way you get a truer prediction of the averages.
Cheers!

Current Strat: Loosely following 152300 and 85660 more the former rather than the later
Current PIP: 24.04

User avatar
TSPsmart
Posts: 212
Joined: Fri Nov 14, 2014 12:24 pm

Re: TSP backdating date range

Post by TSPsmart »

Both date ranges are wholy inadequate.

I assume you are only going back to 2004 because TSP started posting daily prices in 2003. The indexes they track have been around much longer.

TSP C fund has good data back to 1950 if you use the SP500 data and the TSP S fund has good data back to 1987 if you use the index it tracks. You have to find the discrepancies in any historical data base to fix the issues which most seasonal calculators do not do this.

I know because I built almanacs for both funds and provide a subscription service for access.

Yes 2008 really skews the data on the monthly/weekly/daily levels from both the final bear market plunge and the rapid bounce back. I eliminated bear market years (2001-2003 & 2008-09) from the 29-year look back for both funds. I did not eliminate it from the 67-year look back for the SP500 (TSP C fund). If it made a difference in a 29 year look, it really skews 12 years of data.

Cheap sells pitch: If you do not have much money in TSP, it probably is not worth the cost to you. But if you have a $19K in TSP, then a 1% additional gain would pay for my almanacs if you are a seasonal investor. BTW, the TSP F fund has an interesting pattern not seen in the AGG ETF which is interesting since they both track the same index.

Qball
Posts: 11
Joined: Tue Apr 04, 2017 11:00 pm

Re: TSP backdating date range

Post by Qball »

Chulke wrote:I can agree with what you are getting at but, If you remove the low end bad year of 2008 from you data sets then you also should remove the high end year too, whichever year that might be. That way you get a truer prediction of the averages.
Agree. If you remove the lowest, you should also remove the highest.

User avatar
evilanne
Posts: 2067
Joined: Thu May 14, 2015 6:52 pm

Re: TSP backdating date range

Post by evilanne »

mill0168,

You are not wrong in weighting more recent data more or excluding certain data that may not be relevant, it is a matter of perspective. Each financial fund prospectus claims that past performance does not guarantee future performance. Aitrus calculates performance for different time periods, since inception, last 15, 10 or 5 years. There is a lot of data out there, how you choose to weigh it is up to you. Looking at the recovery from the 2008-09 recession, I could argue that older data is no longer relevant given the Fed actions & unprecedented QE. Some are not willing to use one of the daily strategies because they don't feel there is enough data. What strategy each person uses chooses is a matter of judgement based on their knowledge of the data available and their individual risk tolerance, experience and current situation.

mindofmush
Posts: 353
Joined: Mon Jul 02, 2012 1:38 pm

Re: TSP backdating date range

Post by mindofmush »

That's the beauty of the TSP calculator: YOU decide what factors are the most important to create a plan that works for YOU regardless of what others think.

Some other members have already done what you mentioned and even limited the timespan to the most recent few years. Some people place more importance on a Low StdDev, others think more data years makes a better model and mj is working on a rolling averaging 5 year (or so) plan.

I think your plan has value but when I tested some good short term strategies against 2008, they flopped, (big negative returns) so for me, it's not good. I'd rather settle for a lower average return than have any negative years. (I do realize that the next Great Recession may not occur in any way like the last one but a strategy that has no negative years with all the data possible makes me feel like I won't lose my shirt in the next big bear.)

The only problem I see with TSPcalc is too many choices...but that is a good problem to have.
mo meng, mo ching (which loosely means: no money, no life)

Travis1984
Posts: 102
Joined: Wed Jun 03, 2015 1:24 pm

Re: TSP backdating date range

Post by Travis1984 »

mindofmush wrote:That's the beauty of the TSP calculator: YOU decide what factors are the most important to create a plan that works for YOU regardless of what others think.

Some other members have already done what you mentioned and even limited the timespan to the most recent few years. Some people place more importance on a Low StdDev, others think more data years makes a better model and mj is working on a rolling averaging 5 year (or so) plan.

I think your plan has value but when I tested some good short term strategies against 2008, they flopped, (big negative returns) so for me, it's not good. I'd rather settle for a lower average return than have any negative years. (I do realize that the next Great Recession may not occur in any way like the last one but a strategy that has no negative years with all the data possible makes me feel like I won't lose my shirt in the next big bear.)

The only problem I see with TSPcalc is too many choices...but that is a good problem to have.
The thing is, with most of the “good choices” they ar epretty close to in and out of the market at the same time. Tough to go wrong.

At this point, i think the only question is: do you want a 40 or 60’ cabin cruiser?
"Fear of loss is the path to the Dark Side."

Yoda

oldjack
Posts: 124
Joined: Sun Oct 17, 2010 9:24 pm

Re: TSP backdating date range

Post by oldjack »

I think you're hitting on what I consider to be the logical fallacy to the "seasonal" calculations. They are 100% backward-looking only. Who cares what happened in the past you want to predict what is happening this year/month/week/tomorrow.

If you think about it, investing is not at all like it was 50 years ago. Retail investors didn't exist, the brokerages controlled all the trades even the big money listened to brokers with seats on the exchange, and this forum fortunately has replaced the ticker tape. Electronic trading is so much faster. I don't think we should expect the same patterns as 50 years ago.

In this respect I think one of the better things to do is focus on the Leading Indicators and let them guide you on near-term projections for the big indexes (which is all we can invest in with the TSP).

OTOH, throwing theory away and just looking at the proofs, well, the Seasonals are clearly working. I'm late to the Daily Seasonals and sort of following #15536, which is up 22.97% this year while the C Fund is up ~17%. That's a very good beat, I'd take it every day of course!

mindofmush
Posts: 353
Joined: Mon Jul 02, 2012 1:38 pm

Re: TSP backdating date range

Post by mindofmush »

You are correct, oldjack, that seasonal strategies are "backward-looking" only. But I don't know of any system that "predicts" the future without historical data.
The market index funds have outperformed 80% of the actively managed mutual funds (according to Motely Fool). The TSP has given us limited but good funds with which we can at least match the market. The TSP think us as simple children by limiting our choices and our IFTs, but we complicated the game by observing patterns in the seasons and developed strategies to take advantage of those cycles which yielded returns greater than the index average.
There are hundreds of strategies outside the TSP that can provide higher returns but you have to play the hand you're dealt.
mo meng, mo ching (which loosely means: no money, no life)

SnareMV17
Posts: 176
Joined: Thu Jan 10, 2013 9:06 am

Re: TSP backdating date range

Post by SnareMV17 »

If you normalize the historical data by eliminating actual historical events (high and low years), all you end up with is growth period data. It depends on personal flavor, but I don't think most people are interested in finding algorithms for investing real money that ignores when the market goes down. All you'd end up with is growth period numbers. How is that smart?
"Get your money for nothin', and your chicks for free."

Following TSPCalc strategy #64902.

paulweath
Posts: 7
Joined: Thu Feb 20, 2014 4:44 pm

Re: TSP backdating date range

Post by paulweath »

The calculator actually does that by factoring in the performance and deviation but depends on your personnel choice of each. I enjoy the 20 plus percent returns. Until that fails (which it hasn't since the available data), I will return to my monthly plan. Nice thoughts on this thread though.

Post Reply

Fund Prices2024-03-27

FundPriceDayYTD
G $18.14 0.01% 1.00%
F $19.09 0.26% -0.68%
C $82.11 0.87% 10.42%
S $82.19 1.48% 6.61%
I $42.68 0.56% 6.21%
L2065 $16.38 0.84% 8.36%
L2060 $16.38 0.84% 8.36%
L2055 $16.39 0.84% 8.36%
L2050 $32.73 0.71% 6.94%
L2045 $14.91 0.67% 6.56%
L2040 $54.37 0.63% 6.20%
L2035 $14.34 0.58% 5.77%
L2030 $47.66 0.53% 5.35%
L2025 $13.14 0.31% 3.40%
Linc $25.60 0.24% 2.79%

Live Charts

Pending Allocations

Under development. For now, you may view Pending Allocations by going to "fantasy TSP" and selecting "Leaderboard sort" of "Pending Allocations".