newbie here, looking for seasonal investing advice link

[Wind]

Hi, new to this site and forum but have searched around enough to find some bits of info for seasonal investment strategies but mostly seems quite old.

I'm not able to find combo of clear 'how to' and current best available data supporting a current seasonal investment strategy. Hopefully, this doesnt require a lot of new typing in response - just hoping for some helpful links would be appreciated

As far as new typing goes - are people having good results with seasonal investing ytd for 2021-2025?

Thanks,
Wind

[Aitrus]

Welcome! Always to see new blood here.

A good starting point is to take a close and thorough look at the Seasonal Musings thread in my sig block. Take your time with it - there's a lot to cover if you've never encountered the concept before, or are new to TSP investing as a whole.

The thread covers the basics of what's going on when it comes to a monthly seasonal approach using TSP. The first dozen or so posts will get you caught up (basically to the end of the Funds and the Mixes that I used to post about regularly). Feel free to ask questions from there - most answers will assume that you've covered that material as a baseline. You asked for "current" and what's in that thread is about as current as it gets as far as theory goes.

The next step is to go deeper into how that concept can be applied more thoroughly in TSP. A forum member created www.tspcalc.com to apply the seasonal investing concept, but using the 2 IFT limit per month to the max. Basically, it's a calculator / experimenter for testing out various ideas / repository of already-created ideas that have already been created. On that site is also more current returns for the core TSP Funds and Mixes I discuss in my thread. If you've read my thread and understand the contents, then it's a simple matter of going to TSPCalc and looking at the up-to-date returns.

There is also a Facebook group connected to TSPCalc that discusses seasonal investing

[Wind]

thank you kindly! i may be running the pages' inputs all wrong and/or drawing wrong conclusions but it appears this strategy has been good in 2021-2024 (and longer time period)
https://www.tspcalc.com/seasonal.php?ID ... sortby=avg

but is running quite below average for this year so far.
https://www.tspcalc.com/seasonal.php?ID ... sortby=avg

does that seem about right?
if i'm reading columns right - it's still doing better than my account. just sayin',..
thanks,
Wind

[Aitrus]

The one you list seems to be optimized for 2021-2024. If you look at 2004-2020, it doesn't perform very well.

Compare the strategy to something like 141091. See how the compare from 2021-2024, then 2004-2020, and finally 2004-2024. Then look at the 2025 number thus far.

I prefer something that performs across a range of investing conditions - bull markets, pullbacks, times where it just sort of meanders, etc. Having a lower STD is nice, but not mandatory. Somewhere around 10 or so is fine. Others aim for nothing above 3 or 4, while others don't care about STD so long as the return is as high as possible.

That's why my Seasonal Musings thread talks about using as many years of data as possible. The fewer years you use, the more impact any individual year has on the final result. Once you have about 20 years of data, you begin to see the long-term average in a way that lets you have an idea of the kinds of returns (+/- the STD) you can expect.

[Wind]

interesting stuff!

I can see how a 'strategy' designed around Not being in certain funds through some of their historic big losses and being in others during big gains might be able to skew a historic analysis and look pretty good when run over a long range, but possibly not help in the future (assuming those historic big losses that were avoided during the strategy's design phase, and/or future big losses have no anchorage to season). what do you think - is this a natural development of historical optimizations?

Although i can enter 141091 in an input box in the 'seasonal calculator' section and run it over different ranges of years in the 'results' section, i could not figure out a sort in the 'seasonal strategies' section that would get it to display there due to its 100 strategies display limitation. am i missing something?

Are there people that have trusted and used these strategies with noticeably-good results over a long haul? I'll have to figure out how to use the fantasy link, but would you suggest a couple strategies to experiment with going forward?

Thanks again,
Wind

[Bubba]

interesting stuff!

I can see how a 'strategy' designed around Not being in certain funds through some of their historic big losses and being in others during big gains might be able to skew a historic analysis and look pretty good when run over a long range, but possibly not help in the future (assuming those historic big losses that were avoided during the strategy's design phase, and/or future big losses have no anchorage to season). what do you think - is this a natural development of historical optimizations?

Although i can enter 141091 in an input box in the 'seasonal calculator' section and run it over different ranges of years in the 'results' section, i could not figure out a sort in the 'seasonal strategies' section that would get it to display there due to its 100 strategies display limitation. am i missing something?

Are there people that have trusted and used these strategies with noticeably-good results over a long haul? I'll have to figure out how to use the fantasy link, but would you suggest a couple strategies to experiment with going forward?

Thanks again,
Wind

I think most of us use the strategies as guides and often use multiple strategies. There are many variations you can select. In my case, I try to do one strategy and buy-and-hold at the same time. 50% to each. Having said that, I've totally ignored my strategy for this year and done what I perceive will happen. I'm somewhat satisfied with my results, but nothing is certain!

[Wind]

thank you -

i guess my question was - 'will a seasonal strategy that is inherently based on historical data which is both seasonal and non-seasonal in nature provide useful results for future use?'

i noticed 141091 is marked as created feb 2021. if i read results correctly - it averages 27%/yr for 2004-2021 (i assume is the period of historical data to which the strategy was fitted), then 24.91% for 2004-2024 - pretty sweet!

but, how did it perform 2021-2024 inclusive? if i read the results table correctly, and did some minor calculations accurately, it looks like avg yield for 141091 strategy 2021-2024 was 12.55%/year whereas i think that the s and p 500 averaged over 15%/yr for that period.

does that seem about right?

Wind

[Aitrus]

With a seasonal approach, the idea is not to try to beat the market at every moment or even every year. That's impossible to do.

However, what IS possible is to beat the long term market average. That average is around 10-11% since the Great Depression. Once you factor in the -3% cost due to annual inflation, you reach the 7% conservative ballpark number that everyone uses to plan for retirement. So we aim for beating that mark.

So while 2021-2024 only has a 12.55% increase for 141091, it still beats the long term historical average of the stock market. By the standard I discuss in my Seasonal Musings thread of returning an average of at least 12% a year, 141091 is a winner for that time period.

141091 was based off another mix: 85660. That one was created in mid 2019 (a couple of years before 141091). The only difference between 141091 and 85660 is that the latter uses the F Fund at some points, while 141091 uses only the G Fund when not invested in stocks. The returns from 2020 - 2024 for both still beat the long term average even though we went through a massive swing in the market due to COVID.

One of the reasons that a Seasonal approach is used is that it allows you to ignore the gyrations of the market - wars, politics, etc. Fundamentals are fundamental. The market has a rhythm that doesn't repeat perfectly, but it's close enough to plan around in general terms. There will be swings outside the norm - COVID, tariff wars, etc - but those swings always fix themselves back to the fundamentals driven by everyone working and buying stuff to fill their daily needs. The last time we had a wild swing downward that wasn't stopped and headed back upward after a year or so was the Great Depression. If we ever have another one of those, we'll have far worse things to worry about than our retirement accounts. Even the Great Recession started in late '07, bottomed in '08, and was on a firm trend upward by '09.

Yes, it's true that "past results are no guarantee of future returns," however, people forget that it applies to the regular market as well. People bank on using 7% because the long-term average has been 10% with 3% inflation, and they bank on that going into the future. You pointed out that 2021-2024 had an average of 15%. It certainly won't be that high going forward.

These strategies keep us out of the market during times when the market tends to do poorly - June, August, parts of other months, etc. If we're not in the market 100% of the time but still matching or slightly exceeding the market's returns, then we're winning because we're avoiding the potential of a downturn at times when those historically happen most often and thus lowering our risk.

People opposed to this approach often say "If you're in the market 100% of the time, then you always catch the 10 best days of the year and not miss them". They seem to forget that being in the market 100% of the time also means catching the 10 worst days of the year as well.

[Wind]

I think I understand the reasoning but I still wonder which strategies have beat the S and P going forward since their creation, rather than beat some average of the (past + future) but used hindsight to beat the past as part of their creation.

It looks like 85660 was created June 2020, then averaged 11.39%/yr for 2021-2024 whereas S and P averaged 13.75% for that period. To be fair - 85660 seems to be beating S and P so far in 2025!

Thank you for your patience in explaining the basis, reasoning and value of the strategies,
Wind

[Bubba]

I think I understand the reasoning but I still wonder which strategies have beat the S and P going forward since their creation, rather than beat some average of the (past + future) but used hindsight to beat the past as part of their creation.

It looks like 85660 was created June 2020, then averaged 11.39%/yr for 2021-2024 whereas S and P averaged 13.75% for that period. To be fair - 85660 seems to be beating S and P so far in 2025!

Thank you for your patience in explaining the basis, reasoning and value of the strategies,
Wind

I was trying to say that the strats do not always beat everything. A few years ago, I was playing with the top 10 strategies. I found that from Jan to May (at the time), that buying and holding would do better than most of the top 10 strats. It was post May that the strategies killed it. The point of the strategies, however, is to remove emotion and just to follow a plan. In this line of thinking, if I had followed my strategy this year I would have been down by a bit over 0.05%. Since i used my brain and a bit of buy-and-hold and a few other strats, I've gained a bit instead of losing. That's not always the case, but this year is special in my mind. Whenever the inversion happens of the 10yr vs. 3 month, then all bets are off. So, I'm trying to time the market, which is a terrible long-term approach, but working for now.

Good luck!

[Wind]

Ok, but I don't think buy and hold C fund (or a chunk of a diversified portfolio including C fund) is emotional unless you look at it while C fund is losing :O

All theories aside - I'm starting to wonder if any seasonal strategy has beat a simple average of the C fund's yearly % returns over the span of years that followed After creation of the strategy?

If yes - I'd like to look at that strategy closer for sake of getting better returns

Thanks,
Wind

[Aitrus]

I think I understand the reasoning but I still wonder which strategies have beat the S and P going forward since their creation, rather than beat some average of the (past + future) but used hindsight to beat the past as part of their creation.

It looks like 85660 was created June 2020, then averaged 11.39%/yr for 2021-2024 whereas S and P averaged 13.75% for that period. To be fair - 85660 seems to be beating S and P so far in 2025!

Thank you for your patience in explaining the basis, reasoning and value of the strategies,
Wind

A quick note: the site underwent an overhaul in June 2020, so any that say "created before June 2020" we can't know for sure. The calculator / seasonal stuff was added to the TSPCalc site in 2017, and some checks on the Wayback Machine can give some educated guesses about how many were in existence at given points.

No apologies needed. We're all here to learn, and you make a fair point: where's the evidence that it's working (preferably the longer the better)? We like it when people poke holes in the ideas we're working with - it forces us to take a fresh look at things.

Here's a few examples of where it's working:

136510 - created on 16 Jan 21, average from 2021 - 2024 was 16.88%, C's average was 15.44%.

143396 - created on 2 Mar 21, average from 2021 - 2024 was 16.37%. For 2022 - 2024, 143396's average was 14.26% vs C's average of 11.03%.

138857 - created on 17 Jan 21, average from 2021 - 2024 was 16.69%.

89490 - created before June 2020 (Wayback says it was in existence in Jan of 2020, so it's at least that old), average from 2020 - 2024 was 16.91%, C's average was 16.01%.

The way some people use seasonal is not to beat the C over the short term, rather, they want stability. They prioritize consistent annual returns over high gains / beating the average. To do this, they don't aim for the highest returns, they just want a consistent average that's reliable. So they look for a low STD (standard deviation), usually something around 3-5%.

Who might want something like this? Maybe a retiree or almost-retiree who wants something that's stable but doesn't trust the bond market due to the Fed's meddling, and the G Fund won't keep up with inflation. So they have to be in stocks at least some of the time, but don't care about making gains. They just want to stay ahead of inflation, and want the returns to be consistent so they don't die early of market-induced heart attack.

For this, there's lots of recent ones that have low STD and high returns, but I wanted to show you some that are older and managed to keep to the idea from then until now.

84611 - created before June 2020, 2004 - 2019 average was 6.63% with a STD of 2.60. 2020 - 2024 the average has been 7.03% with a STD of 3.4. During that same 2020-2024 period, C returned 16.01% with STD of 17.41.

68577 - older than the last one, 2004-2019 average was 6.64%, with a STD of 2.51. 2020-2024 the average has been 6.59% and STD of 3.92.

47495 - Older still. 2004-2019 average was 6.71%, STD of 2.88. C was 10.54% and STD of 15.99. From 2020-2024, the mix returned 5.03%, STD of 2.92. This one might have even been created in 2018, so the 2004-2018 numbers were 6.92% / STD 2.85, and 2019 - 2024 was 4.77% / STD 2.73.

Other reasons: maybe someone doesn't want to use one of the Funds for whatever reason, which eliminates the use of the L Funds. So they can use the calculator to experiment with approaches that don't use the I Fund, the C Fund, etc.

[Wind]

Thank you kindly!

[Bubba]

I think I understand the reasoning but I still wonder which strategies have beat the S and P going forward since their creation, rather than beat some average of the (past + future) but used hindsight to beat the past as part of their creation.

It looks like 85660 was created June 2020, then averaged 11.39%/yr for 2021-2024 whereas S and P averaged 13.75% for that period. To be fair - 85660 seems to be beating S and P so far in 2025!

Thank you for your patience in explaining the basis, reasoning and value of the strategies,
Wind

A quick note: the site underwent an overhaul in June 2020, so any that say "created before June 2020" we can't know for sure. The calculator / seasonal stuff was added to the TSPCalc site in 2017, and some checks on the Wayback Machine can give some educated guesses about how many were in existence at given points.

No apologies needed. We're all here to learn, and you make a fair point: where's the evidence that it's working (preferably the longer the better)? We like it when people poke holes in the ideas we're working with - it forces us to take a fresh look at things.

Here's a few examples of where it's working:

136510 - created on 16 Jan 21, average from 2021 - 2024 was 16.88%, C's average was 15.44%.

143396 - created on 2 Mar 21, average from 2021 - 2024 was 16.37%. For 2022 - 2024, 143396's average was 14.26% vs C's average of 11.03%.

138857 - created on 17 Jan 21, average from 2021 - 2024 was 16.69%.

89490 - created before June 2020 (Wayback says it was in existence in Jan of 2020, so it's at least that old), average from 2020 - 2024 was 16.91%, C's average was 16.01%.

The way some people use seasonal is not to beat the C over the short term, rather, they want stability. They prioritize consistent annual returns over high gains / beating the average. To do this, they don't aim for the highest returns, they just want a consistent average that's reliable. So they look for a low STD (standard deviation), usually something around 3-5%.

Who might want something like this? Maybe a retiree or almost-retiree who wants something that's stable but doesn't trust the bond market due to the Fed's meddling, and the G Fund won't keep up with inflation. So they have to be in stocks at least some of the time, but don't care about making gains. They just want to stay ahead of inflation, and want the returns to be consistent so they don't die early of market-induced heart attack.

For this, there's lots of recent ones that have low STD and high returns, but I wanted to show you some that are older and managed to keep to the idea from then until now.

84611 - created before June 2020, 2004 - 2019 average was 6.63% with a STD of 2.60. 2020 - 2024 the average has been 7.03% with a STD of 3.4. During that same 2020-2024 period, C returned 16.01% with STD of 17.41.

68577 - older than the last one, 2004-2019 average was 6.64%, with a STD of 2.51. 2020-2024 the average has been 6.59% and STD of 3.92.

47495 - Older still. 2004-2019 average was 6.71%, STD of 2.88. C was 10.54% and STD of 15.99. From 2020-2024, the mix returned 5.03%, STD of 2.92. This one might have even been created in 2018, so the 2004-2018 numbers were 6.92% / STD 2.85, and 2019 - 2024 was 4.77% / STD 2.73.

Other reasons: maybe someone doesn't want to use one of the Funds for whatever reason, which eliminates the use of the L Funds. So they can use the calculator to experiment with approaches that don't use the I Fund, the C Fund, etc.

I've been really big into the SD part (I targeted less than 8%, which is about the SD of a long-term bond fund) and a system that was created prior to 2020. I've combined a couple together like that (making a new one with a more recent date...ugh). I've not been pleased this year, but the jury's out. Like I said before, I found that many strats did not beat buy-and-hold from Jan to May, but afterwards they get interesting. So, we'll see how the rest of the year goes.

[Wind]

so, i guess it should go without saying that a strategy that optimizes results for a past or historical data set will be optimized for that data set by identity. but how it performs for the future is the important thing. as far as that goes, i guess strategies could be optimized using relatively-older historic data sets and tested against the relatively-less older historic data for their predictive power.

to some extent, i think that is how we got the ones cited above that did better than C index After their creation ('21-24). but i dont think any of those did well for 2025. that's not to say they wouldnt perform better than C on average during 2025-2035 - i really have no idea!

given a data set of enough thousands of strategies created before 2021, i expect a sort for those that have done well in 2025 so far, or 2021-2025 so far would have results! examining their particular IFTs to see how that happened might be interesting but i would still be suspect of their predictive power for the rest of 2025 for similar reasons that the 21-24 sort didnt produce big winners for 2025.

Wind

ps - without paying my dues on research, i kind of assume major crashes and recoveries are not 'seasonal' but have profoundly affected the historic data (and strategies that look for seasonal optimization based on that data). i wonder if hybrid seasonal/timing strategies that are optimized for more general bull (or bear) markets have useful predictive power during future bull (or bear) markets (respectively). their use would require a user to identify a current bull vs bear market and assumably make adjustments. in other words, is there techniques to strip out some of the non-seasonality of past results. i expect relatively-advanced statistical math attempts to do this and provide objective measure of its results but that is beyond my skillset