Thread: S-fund Challenge! Guesses

Originally Posted by Stoplight

Still pushing the veal, eh ? "Management" must like you !
From what little I understand, NN is just too alien to me...although I'm interested, I suspect my head would explode if you tried to explain it to me...just like it does while trying to understand Ebb's "patterns" Just show me what works, and what doesn't work ! That's why this "competition" is interesting to me !
RE: the old engineering joke...Would you believe my Grand-Niece asked me that EXACT question the other night ? Made me LOL !!! I think she went out on a date with a mechanical engineer...poor Woman !
Stoplight...

Depends upon how the NN is explained as to whether your head explodes

The 'competition' will separate the wheat from the chaff...over time.

IMO, Ebb doesn't explain his charts well...at least not how I would explain them if I understood them. (Hint, Hint, Ebb) Ebb would have to be willing to undergo my rigorous cross-examination (not many can handle such questioning) before I could say I understand his charts. After that, I would be willing to post a user-friendly description/explanation of the charts on behalf of Ebb.

Originally Posted by userque

Depends upon how the NN is explained as to whether your head explodes

The 'competition' will separate the wheat from the chaff...over time.

IMO, Ebb doesn't explain his charts well...at least not how I would explain them if I understood them. (Hint, Hint, Ebb) Ebb would have to be willing to undergo my rigorous cross-examination (not many can handle such questioning) before I could say I understand his charts. After that, I would be willing to post a user-friendly description/explanation of the charts on behalf of Ebb.

I keep telling folks it's not rocket science. I just tally the results of the C, S, and I-fund for the patterns (been doing it for 7 years). I only use addition, honest! Only one pattern gets updated every day. Has anyone tried comparing the difference between yesterday's chart and today's updated chart? Does anyone understand how I get the signals (short, cash, or long) by just looking at the Strategy parameters?

Originally Posted by userque

Attachment 28186

Chart legend update forthcoming.

20140415b.jpg

Thought I would help Tom out by deleting old 'attachments' via my settings and reducing my server footprint...had no idea doing that would delete the attachments in my prior posts (Thought they were redundant.)

Here's the latest post again. I have the older ones stored locally if anyone in interested.

Originally Posted by ebbnflow

I keep telling folks it's not rocket science. I just tally the results of the C, S, and I-fund for the patterns (been doing it for 7 years). I only use addition, honest! Only one pattern gets updated every day. Has anyone tried comparing the difference between yesterday's chart and today's updated chart? Does anyone understand how I get the signals (short, cash, or long) by just looking at the Strategy parameters?

Wow, missed this post earlier

Ebb, I'm certain your charts are simple, you just haven't explained them well. 40,000 Frenchmen aren't wrong. If you're game, I will read your explanatory material, and ask you about what doesn't make sense to me until I understand it or you say "UNCLE!"

Originally Posted by Stoplight

Still pushing the veal, eh ? "Management" must like you !

From what little I understand, NN is just too alien to me...although I'm interested, I suspect my head would explode if you tried to explain it to me...just like it does while trying to understand Ebb's "patterns" Just show me what works, and what doesn't work ! That's why this "competition" is interesting to me !

RE: the old engineering joke...Would you believe my Grand-Niece asked me that EXACT question the other night ? Made me LOL !!! I think she went out on a date with a mechanical engineer...poor Woman !

Stoplight...

I hope this simple analogy doesn't 'splode your brain I looked briefly on the net for one, but couldn't find one simple enough, imo.

There have been some great recent posts here regarding April 15 statistics going back to, let's say 1955 or so. Typical statistics will add all the percentage changes for April 15 (or for where April 15th falls on a Tuesday, etc. etc.) for all those years and achieve a bias for that day. An average. I think Ebb does something similar.

Someone willing to put in a lot more work could work out the averages for each year, individually, from 1955-2013, analyse the data, and possibly determine that the bias varies by a predictable amount each year, and then formulate an algorithm or formula to predict the bias as it relates to each year. Such a person would have an advantage over someone simply using a simple average over all the years to predict a 2014 bias for April 15, given only prior years data without an algorithm explaining how the bias is changing over the years.

Suppose differently that someone looks at the averages based upon what day of the week April 15 falls. Like above, this can easily be done with simple statistics. They would conclude to the effect of: bias when it falls on a Monday is this....on a Tuesday is that, or the other....

Now suppose they wanted to know the bias when the previous close was 'up' versus 'down,' AND when the close before that was up or down; OR instead of just up or down; when its up 0-0.5%, versus 0.5-1.0%, versus ....., 0-0.1%, (as the percentage divisions approach infinity and the size of each division approaches zero). A realistically impossible task for a human.

Humans typically only compare one or two inputs at a time in these sort of cases: April 15th bias, April 15th and Monday bias, etc.

A NN, like above, could factor in the year as well as the day of the week as well as infinite price 'divisions' as well as the closes of the last several days as well as ...lunar cycles... etc. A NN running on a typical home computer can effectively handle scores of inputs, with thousands of samples/examples to analysis. Virtually impossible for a human with Excel or even an advance trading platform to accomplish (unless it properly utilizes some extremely customizable machine learning technology).

A NN is statistics on steroids...

I could go on: Why they don't "work" for some folks. Why they do work for some folks. Etc.

Hope this helps, if not, I can try again if you like.

IMO,

Q

S-Fund Challenge Tally Legend, Update

The tally now includes additional virtual traders. Virtual traders now have an '*' in front of their handle for easy identification. These traders are:

*Always Short: This trader is always short the market. I added it to balance *S-Fund, who is always long the market. Can you beat them?

*Random Walker, Sr./Jr.: Sr. trades based upon the NY Numbers game. If the midday drawing for that trading day falls >=500, then he predicts the next day close to be up, else he predicts down. Jr. likewise bases his predictions on the Illinois Pick 3 Midday drawing. Shouldn't they break even if the market is random? Can you beat them? They won't be around forever; just long enough to show whether the walk is random.

I've also added a column, "Current Cash Days in a Row." This column tracks the number of days a trader has currently been in cash. It can help observers decide for themselves whether a trader is actively trading their account.

Q

Originally Posted by userque

I hope this simple analogy doesn't 'splode your brain I looked briefly on the net for one, but couldn't find one simple enough, imo.

There have been some great recent posts here regarding April 15 statistics going back to, let's say 1955 or so. Typical statistics will add all the percentage changes for April 15 (or for where April 15th falls on a Tuesday, etc. etc.) for all those years and achieve a bias for that day. An average. I think Ebb does something similar.

Someone willing to put in a lot more work could work out the averages for each year, individually, from 1955-2013, analyse the data, and possibly determine that the bias varies by a predictable amount each year, and then formulate an algorithm or formula to predict the bias as it relates to each year. Such a person would have an advantage over someone simply using a simple average over all the years to predict a 2014 bias for April 15, given only prior years data without an algorithm explaining how the bias is changing over the years.

Suppose differently that someone looks at the averages based upon what day of the week April 15 falls. Like above, this can easily be done with simple statistics. They would conclude to the effect of: bias when it falls on a Monday is this....on a Tuesday is that, or the other....

Now suppose they wanted to know the bias when the previous close was 'up' versus 'down,' AND when the close before that was up or down; OR instead of just up or down; when its up 0-0.5%, versus 0.5-1.0%, versus ....., 0-0.1%, (as the percentage divisions approach infinity and the size of each division approaches zero). A realistically impossible task for a human.

Humans typically only compare one or two inputs at a time in these sort of cases: April 15th bias, April 15th and Monday bias, etc.

A NN, like above, could factor in the year as well as the day of the week as well as infinite price 'divisions' as well as the closes of the last several days as well as ...lunar cycles... etc. A NN running on a typical home computer can effectively handle scores of inputs, with thousands of samples/examples to analysis. Virtually impossible for a human with Excel or even an advance trading platform to accomplish (unless it properly utilizes some extremely customizable machine learning technology).

A NN is statistics on steroids...

I could go on: Why they don't "work" for some folks. Why they do work for some folks. Etc.

Hope this helps, if not, I can try again if you like.

IMO,

Q

Thanks for the NN explanation. What line of work did you say you're in?

I think that's what the Seasonality Chart does and I've seen JTH average out particular days in relation to some other stuff, also. The patterns I get from the C, S, and I-fund don't depend on dates. So it won't matter if we transition into a bear market, the ebbchart patterns will know and adjust accordingly. I have a chart that shows the performance of the patterns during the bear market crash of 2008. In the chart below, pretty much every pattern behaved as it should except for pattern 3. I can just imagine being long and short with these patterns in 2008.

Bearish green patterns (CSI win percentages): 1 (27%), 2 (39%), 4 (53%), 7 (53%).
Bullish red patterns (CSI win percentages): 3 (37%), 5 (55%); 6 (63%), 8 (58%).

Originally Posted by ebbnflow

Thanks for the NN explanation. What line of work did you say you're in?

I think that's what the Seasonality Chart does and I've seen JTH average out particular days in relation to some other stuff, also. The patterns I get from the C, S, and I-fund don't depend on dates. So it won't matter if we transition into a bear market, the ebbchart patterns will know and adjust accordingly. I have a chart that shows the performance of the patterns during the bear market crash of 2008. In the chart below, pretty much every pattern behaved as it should except for pattern 3. I can just imagine being long and short with these patterns in 2008.

Bearish green patterns (CSI win percentages): 1 (27%), 2 (39%), 4 (53%), 7 (53%).
Bullish red patterns (CSI win percentages): 3 (37%), 5 (55%); 6 (63%), 8 (58%).

'u Welcome

Yes, I've seen JTH use three inputs! Calendar day; Month; Day of Week. Doing this requires a lot of data to get a large enough and statistically meaningful sample size...hence, going back to 1955 and such. The issue with using old data is...well...it's old. Ask and answer for yourself: do the markets move today, with Super Montage, etc., like they did in the 50's and 60's...or 70's...or 80's...or even 90's.? If not, you're using "bad" data. That's another example of where a NN does well. As I've stated earlier, it can detect and account for subtle (or gross), "hidden," or complex variances. (Of course, the better and cleaner the data; the better the results.)

Basically, whatever you're doing, a NN should allow you to do it exponentially better, imo. For example:

Say you only trade using a simple moving average. Well plug several differently SMA's into the NN and see which performs best (using internal analysis tools of the software); then have the NN use it to generate your signals after you teach it what a "signal" is. (BTW, I did a study using indicators and NN's: I concluded that indicators are for humans; they simplify things. A NN would rather have the previous nine closing prices rather than a nine day SMA. Even so, NN can be bogged down by limited computer power and relatively too much data. So sometimes you can take short cuts like using indicators etc.)

Your charts seem to be working so far, but the race is long and the NN is now learning from every wrong move it makes. To the extent that the market is predictable, it can be predicted. Only time will tell.

By the way, NN was showing UP day for today after final round of tweaks and adjustments. It was late so I'm letting my original guess stand. Now that I've finished the spreadsheet, Going forward, I should have more firm guesses and have them earlier.

Q

UPDATE: NN is still crunching, but target so far is 984'ish ouch for me

4/17/14 long (caution: pattern 3 performed the worst in a bear market, not saying we're in one, though).

Originally Posted by userque

S-Fund Challenge Tally Legend, Update

The tally now includes additional virtual traders. Virtual traders now have an '*' in front of their handle for easy identification. These traders are:

*Always Short: This trader is always short the market. I added it to balance *S-Fund, who is always long the market. Can you beat them?

*Random Walker, Sr./Jr.: Sr. trades based upon the NY Numbers game. If the midday drawing for that trading day falls >=500, then he predicts the next day close to be up, else he predicts down. Jr. likewise bases his predictions on the Illinois Pick 3 Midday drawing. Shouldn't they break even if the market is random? Can you beat them? They won't be around forever; just long enough to show whether the walk is random.

I've also added a column, "Current Cash Days in a Row." This column tracks the number of days a trader has currently been in cash. It can help observers decide for themselves whether a trader is actively trading their account.

Q

Great, Short Pants should be hard to beat in bear market weather conditions.

4/17/14 long