Avoiding Curve fitting

[jack hershey]

The pattern is attached to show how all the pieces fit together.

B is black for trending long and R is red for trending short.

The Hypothesis set (Seen in your post) dictated the pattern and that discussion is best left to logic, algorithm and paradigm theory. The riddle of induction's solution precedes these, of course. Science and the use of deduction only is what is required.

By stepping away from the financial indutry's orientation to CW derived from induction, we arrive at a paradigm that yield the observable pattern of the interlocked frequencies of P, V and A/D.

It is expressed as B2B 2R 2B for long trends and R2R 2B 2R for short trends.

In using the system the short pattern is ubserved and it brings the ONE PAGER into use at the BO of the RTL of the short channel. It is AFTER the FTT of the short where Distribution is waited out until a "7" scores at the BO. FRV ensues with the "7" score.

An ADA hold takes the market's offer. This is the days of "duration" on the Excel sheet.

 

[jack hershey]

The synthesis of each of these parts is shown on the final attachment. Quality Assurance -----Return on Investment.

Quality is the most effective risk minimizer. return on investment is the compound interest formula where the variables are ranked in importance as: exponent, cycle profit and initial capital.

I started with 300 bucks in 1957 and my one day best for 100,000 shares in one stream of 12 streams at the time was 17 points. 31 partial fills were required.

My favorite back hand compliment was from th SEC which persisted for a while to cite me for insider trading. when the SEC was learning to use computers to detect habits of traders, I made the list using the above stuff I posted. They learned slowly how front running the herd works.

It was my custom for many years to trade limited POA's for professional people who donnated time to those in need who could not afford their services. Usually it was a 20% time service donation.

I figure some of you will catch on to this opportunity and not screw it up. Then maybe you will use some of your profits to help solve local problems.

You saw recently acrary's comments on annual earning from trading. He is correct on how capital performs. If a person uses streams of capital spread over many stocks then all the capital does turns. Going from 40 to 60 to 80 to 100 turns a year is the objective. briefly, this is called crossover trading where % gain rate determines the time of crossover. This is illustrated in "Putting the Pieces Together".

At this point it maybe possible for some people to see the difference between talking and doing the walk. there was an era before ET existed where decorum on the web occurred. That era is over and helping others encounters flak from OCD's. From those of who do help others, we regard this OCD flak and behavior as humor. What could have happened to these OCD's? It took me years to not care about it.

 

[charts]

FYI: consistant ...

Quote from SomeYoungGuy:

Wow, this is pretty consistant. Can you reverse action and rerun? (Meaning go short when the signal says go long)

Very interested to see if we can mirror this into profit. I always need more short strategies to look into.

 

[Trader666]

What a crock! Actually, what you did was try to obfuscate how poorly your "method" backtests by adding in conditions AFTER-THE-FACT... like testing your "method" on a pre-screened "universe." Unfortunately for your credibility, NOTHING about a pre-screened "universe" was mentioned in your paper which I referenced a few posts ago, which I used as the basis of my testing.

Another red herring was questioning the code I used. But as I've pointed out many times, I used Spydertrader's coding of your garbage:
http://www.elitetrader.com/vb/showthread.php?s=&postid=2496544&#post2496544
http://www.elitetrader.com/vb/showthread.php?s=&postid=2473857&#post2473857
http://www.elitetrader.com/vb/showthread.php?s=&postid=2470470&#post2470470
http://www.elitetrader.com/vb/showthread.php?s=&postid=2461451&#post2461451
http://www.elitetrader.com/vb/showthread.php?s=&postid=2460982&#post2460982

Quote from jack hershey:

Years ago we critiqued Trader666's fiasco back test. And we went further in support of his potential learning process. He ignored both our critique and our learning support. He did one further thing: he failed to report out his backtesting of the additional learning support provided to him.

 

[Trader666]

Come on Jack... your obfuscations are getting more and more lame. You can do better than that!

I've told you repeatedly that I tested your "method" on 1000 stocks from 2000 to 2005 -- a total of 5000 stock-years. Which comes out, on average, to less than 5 trades per stock per year.

Here's some real humor... your September 1st prediction

"As a trader for 53 years, this turning point tomorrow is the most significant I have seen in my life."
http://www.elitetrader.com/vb/showthread.php?s=&postid=2560747#post2560747

Quote from jack hershey:

Regarding the 24,000 trades using inital capital of 500,000 dollars; anyone can simulate the trades. Start with 500,000 dollars and subtract 20 dollars 24,000 times. Try that in real life and see how long it takes to do it. At one trade a day, it is about a century. But the hold was 5 days so that makes it a little longer. OR..... This is just humor...

 

[Trader666]

ROTFLMAO!!! In my mind? How about in reality?! (which is an alien concept to you)

ANYONE who uses spydertrader's code for the scoring over the period I tested will see there are WAY more entries (0 to 7 turn) than exits (4 to 3 turn). You don't even understand your own drivel!

And the reason for that is the market doesn't conform to your flawed perception of it.

Just like you don't really make 3X daily range as you claim, just like Neoxx wasn't doubling his money every few days as you claimed, etc.

Quote from jack hershey:

Trader 666 creates a humorous statement continually from his mind. Read it and have a laugh. In his mind there are more entries than exits in scoring some stocks.

But in trading the natural cycle there are an equal number of exits for the entries.

 

[Trader666]

Coming from you I'll take that as a compliment!

You've been the voice of sanity here on ET, with gems like these

For a while I participated with a multidisciplinary team at a learning hospital at the invitation of my psychiatrist who is also a Native American shaman.
http://www.elitetrader.com/vb/showthread.php?s=&postid=1555191&#post1555191

Even Jiminy Cricket has his old fashioned neck tie off or loose. this is a real treat for the mind.
http://www.elitetrader.com/vb/showthread.php?s=&postid=2106468&#post2106468

This is THE place in THE world to be when I post. If you are so stupid to not have 3m micorpore tape to tape your mouth shut, then use the stupid person's painful type tape.
http://www.elitetrader.com/vb/showthread.php?s=&postid=2111426#post2111426

what distinguishes me from Forbes 400 people is that they are greedy and selfish and I am not
http://www.elitetrader.com/vb/showthread.php?s=&postid=2610703#post2610703

Quote from jack hershey:

Trader666 exhibits most strongly the OCD characterisitc of very angry people.

 

[Trader666]

You didn't answer me Jack...

Is this as bogus as your claims of success for your other proteges? Neoxx and RoughTrader for example?

Neoxx -- you said Neoxx was doubling his money every three days when he was actually losing money.

RoughTrader -- your claims about his trading deflated each time I called you on it and he hasn't posted since May.

04-23-09 01:26 PM
"For SCT like trading, rough trader takes 1500 dollars to 94,000 dollars"
http://www.elitetrader.com/vb/showthread.php?s=&postid=2399521&#post2399521

04-26-09 06:51 PM
"Bottom line: looks like roughtrader makes 20,000 a month on two contracts average while scaling out of entries on his ATS"
http://www.elitetrader.com/vb/showthread.php?s=&postid=2403231#post2403231

05-22-09 07:28 PM
"Rough trader is only making a couple thousand a month on a couple of contracts. He needs to make one adjustment to make 20,000 a month. He needs to add 18 contracts to his trading cars"
http://www.elitetrader.com/vb/showthread.php?s=&postid=2437715#post2437715

Quote from jack hershey:

Currently 20K of capital is returning 4K of profits weekly with those who have been doing the weekly meeting for a matter of months.

 

[jack hershey]

Here is a general comment on avoiding curve fitting And I am glad to see that Trader666 has provided more indepth information on the backtest he performed and why he did what he did.

The vast majority agree with his views. The basis of their agreement with him is their orientation to the markets and their beliefs. Everythng turns out to work for everyone.

In PVT stock trading it has always been important to avoid curve fitting that is inappropriate.

Two realms are involved in trading: how the markets work and appreciating capital.

Stocks work in cycles and markets also have cycles. PVT trading involves picking a quality Universe on three bases: earnings, price strength, and the combo of repeatability and reliability. All of these involve curve fitting through a filtering process. Filtering in this manner is a good and positive on going process. It is now a click of a button.

In PVT trading our prime concern is timing the market and the leading indicator of price is volume. This is curve fitting. To trade stocks where a signal is generated by using an appropriate volume signal, the Universe MUST exhibit this characteristic. Again, this is a purposeful curvefitting process. The IAS is used to get this information and it is done by analysis and data taking and mathematics. There is NO way a group of stocks will exhibit this characteristic over a period of years. A person may not presume that any publically determined financial industry defined group of stocks will have an established characteristic the will enable the approeciation of capital over the long term. Trader666 proved this and I personally appreciate his proof. The mutual fund industry also proves it and so does the financial planning professional paid expertise. This is what we deem BAD curve fitting.

How does curve fitting fit into timing market trades? From a trader perspective only trading stocks that have the same behavior is an advantage. Volume changes four times in each price cycle. Thus, timing the coincidence of volume change and price chance is the name of the game. By adding a third variable which changes four times during the price cycle, the timing becomes more precise since the window is narrower in time.

To trade most effectively PVT uses the MADA routine. M and the first A, is preparation. Curve fitting applies to preparation in a good way because is creates consistency. Decision making and Action have nothing to do with curve fitting, except to say the SIGNALS are the result of curve fitting.

PVT used 3/8th of the cycle to make money (from RTL BO to the FTT of the long half cycle). Basically it is open entry trading on the day the DU ends and the FRY begins within a short period after open. And open exit trading on the day the FTT begins by volume signalling it no longer keeps up with the factual ly determined peaking volume within the day.

PVT trading trades on a fractal within the daily bar fractal. EOD data is used to filter and gain the Universe. Skillful trading is done using the ONE PAGER. A person who is employed operates of the daily data and does use EOD for his trading as a beginner. A switch to intraday fineness is made when the person goes to full time trading. Usually at that time the person also goes into SCT trading.

My switch was made within two years of my trading fees and commissions exceeding my salary on a weekly basis. At that time fees and commissions were much higher than today; it was 45 years ago.

Avoiding curve fitting is especially important when the CW is followed since it is largely inductive. Constructive curve fitting onlt comes into the picture when Monitoring and Analysis is the goal and filtering is the means.

If work is done and only one or few market variables are employed then almost no curve fitting is productive. It may be noticed that some work is done with price only. This is a pitfall.

 

[mind]

extensive testing will inevitably subject results to curve fitting. the problem of curve fitting is not merely addressable within code, it must include the human being doing the tests as well.

example: consider you test a system with merely one indicator (assume there is a system with a single variable for simplicity's sake). you decide you run 1ooo variations of the indicator. assume your output criteria is sharpe ratio. assume your minimum value for that criteria is 1. so you are running simulations of the single indicator system with 1ooo variations and you are looking for an equity curve with a sharpe above 1.

in order not to curve fit you do the same thing with random data. thus you create random equity curves and you find out that let's say three out of one hundred have a sharpe ratio of above 1.

now let us say you run your 1ooo sims and 3o of them have a sharpe above 1. you throw them away since this is perfectly in line with pure randomness. with this number of tests you expect this number of outputs without them having any value tradingwise.

now here comes the thing.

most traders move on to the next indicator and start all over again. finding now lets 27 sharpes out of 1ooo tests. again throwing results away.

assume the whole thing is repeated ... a thousand times. you see where i am heading. the outcome of 3o sharpes above 1 is as well subject to randomness. if you do the whole thing often enough you will find a pure random set that has ... 6o sharpes above and ist still ... purely random.

problem is that these 6o might as well be valid, but it is rather unlikely in the context of these 1ooo different indicators. (btw 1ooo different indicators might sound much, but use the word "systems" instead and assume you are using three indicators for each one. now you only need ten indicators to get to one thousand different systems.)

then what can you do? IMO this is the crucial point of modern technical trading.

1. limit your tests. only start testing if you already have an indication that this is a good fertile ground. acrary used correlations to do so. or you use your trader know how. whatever. be careful not to run a machine doing this search for you ... or you start with a possible fit in the first place.

2. use in and out of sample tests to detect the flaws of your avoidCurveFitting method. be careful not to do this too often, otherwise your in and out of sample begin to merge.

3. get tougher with the criteria with each new run. this is not easily done, though it might sound convincing. it depends IMO how homogenous or diversified your different tested systems are. but this is becoming philosophical.

4. test, once you found something, on other markets. if it does well there too ... good indication it was not a fit. but remember not to do this too often. testing it against 1ooo other time series ...

final note.
acrary got famous on the board for his "edge test", which is nothing but a "curve fit test", if you will. i fully trust that alan was the highly successful trader he claimed to be. yet my guess is, it was not the edge test that made him successful, but the combination of him rigorously studying the market quantitatively, statistical tests and his personal judgement of the markets and their behavior (you might call it hindsight).


summing up
the problem of curve fitting IMO is that metalevels must not become optimization levels. consider you are smart and build a system that optimizes itself in a walk forward process. smart move. consider it works immediately. deal done. trade. but if it works only after 1ooo different combinations of the walk forward variables ... your meta level, namely the walk forward process, designed to avoid fitting, shifted on the optimization surface.

whatever ...