Seasonal Calculator Launch

[tmj100]

Hey mjedlin66, I will start an orderly review of the Minimum Trade Strategies I know of and apply the variations to see what other results pop up from other - in / out tunings.

Thanks for all your hard work.

** The 1993 Super Mario Brothers Movie was great !

[mindofmush]

tmj100,

I've been looking for a daily/monthly strategy with no negative months and a return of at least 12%, but I don't think that is possible. Most of the daily strategies average 30 negative months (2004-2017) with 6-9 of those months losing <1%. It is hard to maintain confidence in a system that loses a lot, especially when the losses continue for 2 or even 3 months in a row.

I just looked at #7858 and counted 42 negative months with 23 of those <1% loss and 2 months that were over 10% loss.

Thanks for a simpler system but I misunderstood your explanation.

[nrialto]

If you find a strategy without a single losing month, please let us know. All things considered, I think overall, it's more important not to have any losing years; the top strategy right now has none and averages an almost 38% annual return (understanding 2008-2009 were outliers, it still averages pretty well).

[LONEWOLF]

nrialto......what is the top strategy right now that has no losing years and averages 38% ??????????

[mindofmush]

#7653 = 37.62% as of 7/12/2017

[tmj100]

mindofmush:

I should have clarified, no negative months was on the Monthly Average summary line across the bottom. The individual months vary up and down and there are always negative months somewhere in any strategy that I can find. So, sorry I should have been more specific about that.

[tmj100]

Here are some updates on the standard seasonal mixes when just slipping IN and OUT dates
=======
TSPCENTER basic #35 as compared to slipping dates
http://tspcalc.com/seasonal.php?ID=8200&years=04-16
14.30 %
Total Compound Return: 424.80 %
Time In Stocks (C,S,I): 38.40 %
no down years
=======
Jahbulon's basic #27 as compared to slipping dates
http://tspcalc.com/seasonal.php?ID=8112&years=04-16
18.62 %
Total Compound Return: 750.87 %
Time In Stocks (C,S,I): 60.40 %
1 down year
======

So, just by adjusting the move IN/OUT dates a little, one picks up several percent and reduced time in stocks. So, an initial review suggests, if that is the case and hard dates can do this much upward push then no need to bother watching MACD and other indicators daily to figure out best times to move.

More to come as time permits on the following:

TSPCENTER FGCC-CFFF-FFCS
Jahbulon's Basic Mix SSCC-CFCF-FCCS
Crispyz Mix FSCC-CFCF-FCCS
Sell in May Mix CCCC-GGGG-GCCC
6in6out with S/F Mix SSSS-FFFF-FFSS
Timboslice 2015 Mix FCSI-CFCF-FCSS
Clapper Mix FSCI-CFIF-FCCS
Boltman Mix FFSS-SFIF-FCSI

=======

[ravioli817]

The top 100 right now is actually the top 50. If you notice, the strategies all have a counterpart in the leaderboard with one minor change, November return to G on the 20th day or the last trading day. Thats the only difference between 7653 and 8042(the top two), and the same difference between the next two. The rest also have that same counterpart. This is visible if you chart them all out like this:

Capture.PNG

I circled some of them, but the rest are also visible. This isn't really a problem but it would be nice to have a filter to take out strategies that only have a single change and perform worse. for example 8042 would be taken out because it only has a single change from 7653 and 8042 performs worse.

The reason i graphed out the above was because i was trying to see how strong of a relationship there is between average return and compound return. There is clearly a positive relationship but it is loose( judging from the r^2 value).

[ravioli817]

I was playing around with the april and may area and got strategy 8535.

The biggest difference is that 8535 is in stocks 58.4% of the time vs 60.8% of the time for 7653. in summary:

---------------- 7653----------8535-----------difference
avg -------------37.73---------37.38----------0.35
med-------------36.17%------36.56%--------0.39%
compound-------6995.84%---6993.94%-----1.90%
st. dev----------26.81---------24.02-----------2.79

so now im kind of conflicted between these two... thoughts?

sorry, i dont know how to properly format that

[mmmmmbeer]

I was playing around with the april and may area and got strategy 8535.

The biggest difference is that 8535 is in stocks 58.4% of the time vs 60.8% of the time for 7653. in summary:

---------------- 7653--------8535-----------difference
avg -------------37.73-------37.38----------0.35
med-------------36.17%----36.56%--------0.39%
compound------6995.84%--6993.94%-----1.90%
st. dev----------26.81-------24.02-----------2.79

so now im kind of conflicted between these two... thoughts?

sorry, i dont know how to properly format that

Ya bro I'm doing the 7653 right now, but I definately see the merit of being less exposed in the market for relatively small difference. Absolutelu something to keep in mind come April. Probably with be a gutcheck on March 30th hahah..

[KWG_5]

First let me say the calculator is really awesome. It is certainly saving a lot of people time trying to calculate returns by hand.

The only feature I would like to see that it doesn't have is the ability to make IFTs based on a set number of days from the end of the month. Given that months can have a different number trading days from year to year I would like to calculate IFTs based on X number of days before the last day of the month (in addition to making IFTs on a certain trading day of the month).

I'll explain why. My hand calculating showed that the last 4 days of the month and the first 2 days of the month were the best performing days on average. Based on this I would expect allowing the calculator to make an IFT 5 days before the last day of the month could give even better returns (although probably only minimally). Right now the calculator gets close to letting you do this, but not quite.

Maybe I'm the only one who would like to see this or maybe this would be really hard to add. I just think this added flexibility would increase returns slightly. My hypothesis is that making an IFT during the second half of the month would be better if it were made in relation to the last day of the month rather than the first day of the month.

[mjedlin66]

First let me say the calculator is really awesome. It is certainly saving a lot of people time trying to calculate returns by hand.

The only feature I would like to see that it doesn't have is the ability to make IFTs based on a set number of days from the end of the month. Given that months can have a different number trading days from year to year I would like to calculate IFTs based on X number of days before the last day of the month (in addition to making IFTs on a certain trading day of the month).

I'll explain why. My hand calculating showed that the last 4 days of the month and the first 2 days of the month were the best performing days on average. Based on this I would expect allowing the calculator to make an IFT 5 days before the last day of the month could give even better returns (although probably only minimally). Right now the calculator gets close to letting you do this, but not quite.

Maybe I'm the only one who would like to see this or maybe this would be really hard to add. I just think this added flexibility would increase returns slightly. My hypothesis is that making an IFT during the second half of the month would be better if it were made in relation to the last day of the month rather than the first day of the month.

Thank you for making a feature request and explaining why you want it. I would appreciate more feature requests, because I am guessing what people want. I was already proven wrong by the poll on this thread. I thought people would prefer graphs to 2017 data.

With the way the current calculator is written, that would require a MASSIVE rewrite. It would take me at least 3 months because it is a lot of work. I only work on the website when 1) My wife is working evenings and 2) I want to.

With that said, the coming version 2.0 has already undergone a massive code rewrite which would make this feature a lot easier to implement. Still rather involved, but not impossible. The original calculator passed the "trading day" to every function, and each function figured out what calendar day to make the trade. So the results table, the trading calendar, the graphs (hint to the future), and others were all simultaneously converting trading days to calendar days.

The 2.0 code is a lot better. Now there is a dedicated function for converting trading days into calendar days. All of the other functions (results, trading calendar, graphs) now accept only calendar dates instead. So with my 2.0 code, the feature that you are asking for would be implemented in one function instead of several. It would still require changes to the database and changes to form handling/post processing code.

It is possible. Just don't expect it in 2017. There are a lot of things that are higher priority right now, such as finishing 2.0, installing graphs, e-mail IFT alerts, a much more robust and filterable leaderboard, etc.

[KWG_5]

My idea probably wouldn't be worth the effort if it would be a huge undertaking. I just noticed you had an option to make an IFT on the last trading day so I thought your code might be written in a way that second-to-last, third-to-last, etc., wouldn't be too difficult to implement.

I look forward to seeing version 2.0. I hope you don't get burned out and keep impressing us with your calculator.

[mjedlin66]

The "last day" is really easy. It actually submits the number 30. The functions automatically determine how many trading days are in each month. If the trading date is later than the number of trading days in each month, then it trades on the last day instead.

So for a month that only has 20 trading days, such as July 2017, the "20th trading day" "21st trading day" and "last trading day" will all be auto-corrected to 20.

[MrSnrub]

mjedlin66 will you release your code on GitHub?