Seasonal Calculator Launch

[mjedlin66]

TSPCalc counts veteran's day as a holiday. You are correct that there are a maximum of 20 trading days, so the calculator allowing 21 is redundant.

This was pointed out months ago but I haven't gotten around to it. It is not a big deal to make the change and then re-validate all of the existing strategies to delete all the ones that trade on the 21st. I just haven't done it yet.

The current project is e-mail and SMS alerts. These are going to be offered under a 19.95/year subscription. To do that, I have to build a user system and integrate it with both a biling system and a payment processor. It is a massive project.

By the way, everyone who has donated will have those funds applied towards a subscription service.

[repinda808]

Just a question about november, Is it possible for november to have more than 20 trading days? There are two holidays that are always observed (veterans day and thanksgiving). These two days always have a non trading day even if veterans day is on a weekend. I think as far as i can see november can have a maximum of 22 weekdays. So 20 maximum trading days after holidays.

So for the strategies 20th trading day is always the same as last trading day. That would make a lot of strategies redundant. If this is right then 20th trading day shouldnt be an option for november.

Am I right or am i being dumb and missing something?

Should be right, if both holidays are indeed observed. However, I did see a year that posted results for 21 days in November.

Do you by any chance remember what year that was? I went back to 2011 and counted a max of 22 weekdays.

You didn't go back far enough. November 2006, TSP website reported prices for 21 days...

[mhayes]

TSPCalc counts veteran's day as a holiday. You are correct that there are a maximum of 20 trading days, so the calculator allowing 21 is redundant.

This was pointed out months ago but I haven't gotten around to it. It is not a big deal to make the change and then re-validate all of the existing strategies to delete all the ones that trade on the 21st. I just haven't done it yet.

The current project is e-mail and SMS alerts. These are going to be offered under a 19.95/year subscription. To do that, I have to build a user system and integrate it with both a biling system and a payment processor. It is a massive project.

By the way, everyone who has donated will have those funds applied towards a subscription service.

What is included in a subscription?

[mjedlin66]

E-mail and SMS trade alerts.

[dearsmiths]

Hey mjedlin66, I plan to donate to help out as this could be the greatest tool ever built, so thank you for going the extra mile for all of us. One question, since I cannot see all the results, what result as the lowest standard deviation and the highest mean return? I have been following #7980 and #15161 and they are low but just asking since I cannot see them all.

Thanks again, dearsmiths

[onerepmax]

Sort 2004-2016 by minimum 30% mean and 5% SD and you'll find some absolutely incredible numbers.

[MakeMe$$$$]

Sort 2004-2016 by minimum 30% mean and 5% SD and you'll find some absolutely incredible numbers.

However, if you use 10% SD there are some interesting results.

[dearsmiths]

Knucklehead here but how do you find 10% SD in the current sorting?

dearsmiths

[mmmmmbeer]

Knucklehead here but how do you find 10% SD in the current sorting?

dearsmiths

The little boxes above each column are "filtering" boxes. Click inside each one and put your numbers.

[dearsmiths]

Knucklehead here but how do you find 10% SD in the current sorting?

dearsmiths

The little boxes above each column are "filtering" boxes. Click inside each one and put your numbers.

Thanks. Does it matter if use Google, or IE?

[dearsmiths]

OK, I just donated $25 for the beer. Figured it out.

Thanks, Dearsmiths

[dearsmiths]

OK, typically you want data to be 5% Std Deviation, or less. So this gives you a 95% confidence in the data, and less chance of volitility. So check out #15981.

dearsmiths

[ravioli817]

OK, typically you want data to be 5% Std Deviation, or less. So this gives you a 95% confidence in the data, and less chance of volitility. So check out #15981.

dearsmiths

Confidence intervals and standard deviations dont work that way. the formula for confidence value for a known standard deviation is:

x +/- z(o/sqrt(n))

where x is the mean, z value corresponding to the confidence interval we want(1.96 for 95%), o is the standardd eviation, and n is the sample size.

could be interesting but I dont think it can be used in the way you described.

[evilanne]

see viewtopic.php?f=14&p=63444#p63444
Xσ for a strategy is based on the mean & standard deviation of the strategy. If you want to understand the detail read the whole thread. My understanding at a basic level from reading it all is that if you have 30% mean and 15 SD, you have 95.46% confidence that the return will fall within 2σ of the mean, i.e. chances of having negative return is <5%; 10 SD is 99.73% or 3σ; and 5 SD is 99.999998% or 6σ. With stock there is always a chance of loss any time you are in the market but the lower SD helps to reduce that possibility as much as possible but it can also limit the upside so it really depend on your goal, risk tolerance and how you achieve a balance that works for you given the data available.

[dearsmiths]

Good point and input, many thanks for bringing me back down to earth. It's just that its very interesting to see one strategy with such a low std deviation and high mean that you have to consider that if when you are correlating the data. RIght now, I am looking at #7980 but this could be one to start out with in 2018.

Thanks again. dearsmiths