So, how is that buy when the market closes down 3% and sell short when the market closes up 3% system doing for you, Mr. Keith “Only-need-to-make-a-penny-to-prove-robustness” Fitschen?
These last two trades tell most of the story to me, as below:
Why don’t we test if ETF Timer’s performance is luck, but please bear in mind, a statistic only tells so much, but that is not the biggest issue I have with the system.
7/14/09 9:30 LONG 1,000 QID 33.39 7/16 15:56 30.11 Normal ($3,276)
3/3/09 14:34 LONG 6,600 QLD 28.53 7/8 11:00 33.79 Normal $34,700
Entry Exit Percentage Profit in Decimal Form
49.56 53.12 0.071832123
64.59 66.55 0.030345255
65.93 66 0.001061732
80.29 83.82 0.043965625
42.3 42.92 0.01465721
67.1 83.89 0.250223547
63.76 74.82 0.173462986
23.93 25.93 0.0835771
55.67 59.4 0.067001976
53.42 56.57 0.058966679
53.42 56.57 0.058966679
28.53 33.79 0.184367333
Luck Coefficient 2.891564804
Now I did that by the raw profits, but I get a slightly different number when taking the dollar amounts of the trades using the data below
50000
52489 2489 0.04978 0.04978
53275 786 0.014974566 0.014974566
53309 34 0.000638198 0.000638198
51959 -1350 -0.025324054
63435 11476 0.220866452 0.220866452
65407 1972 0.031086939 0.031086939
80662 15255 0.233231917 0.233231917
63254 -17408 -0.215814138
82567 19313 0.305324564 0.305324564
86989 4422 0.053556506 0.053556506
90387 3398 0.03906241 0.03906241
93557 3170 0.035071415 0.035071415
96710 3153 0.03370138 0.03370138
131410 34700 0.358804674 0.358804674
128134 -3276 -0.02492961
153272 25138 0.196185244 0.196185244
Luck Coefficient
2.966677757
It’s really not as bad as I thought, and anything below 10 isn’t luck, but if he holds this trade and makes $500k on it, well, it won’t even matter. People will still flock to it.
Now, perhaps people misunderstood my specific problem with ETF Timer: there is no apparent strategy, articulated, or espoused as the strategy by any subscribers.
The same naieve “Who-cares-what-he’s-doing” to make money attitude will cost you money, eventually.
It happened with MVP-3, and I believe it will happen to ETF Timer at some point. His drawdown was realized through averaging down, and because he did not use margin, the market happened to come back, as I point out, luckily.
I don’t mind the name “Backtest Beau.” You only say that because you are quite intimidated by the level of research in my systems. I’m sure you are also quite threatened by someone that has the most robust system ever created.
Let me talk to you vendor to vendor, though? Ok, will you do that? (I think a lot is lost on the public when they don’t realize some of what a vendor is talking about only another vendor would understand, and this is one of those conversations).
If you had research that looked as good as mine, wouldn’t you post the hell out of it? It’s pretty stupid to think you wouldn’t, so if I’m backtest Beau, you can see your nickname at the top. The top vendors on this site in my mind always have research to back up their trades. The ones who don’t are pretty obvious, as you don’t ever see them discussing trading methods in any tangible fashion. It’s funny, but I ignore Palsu and GA, especially, because they don’t have any research to show at all. To you, Keith, I don’t believe in creating 10,000 trade backtests, because experience tells me they will never overcome transaction costs, and will only enrich your broker. This lead to the creation of a volatility based, overbought, oversold indicator that works on many perfectly correlated securities pairs.
Also, If you’d like to talk about profit percent distributions, please see my forum at [LINKSYSTEM_44396934], where I discuss at length how the profit distribution of my system has fat tails at the positive end. Every statistics book I’ve ever written states a distribution with a sample size of 30 is when we can start using Z-scores. While it is abhorrent that Palsun would just copy and paste as his own work, the fact remains that you would need to disprove the central limit theorem to complete your argument.
I consider any ETF or security built to an exact formula stationary. If the formula for QID and QLD is to provide daily 2x leveraged returns, there’s nothing that would suggest it’ll be any different. As long as you “normalize” your values, the data will remain covariance stationary. You can create stationarity from non-stationary data, it just requires some transformations in how your system “analyzes the data.” I do mine with what I’d call a normalized, z-score, and the probability theory I know confirmed it for me long ago.
Sorry everyone, but as a subscriber of systems, I am sick of mediocre vendors like Beau bullying this site.
Beau is one amazing arrogant little boy.
Keith has one of the best systems ever created, and almost always is ranked in the annual Futures Magazine “Best Systems” list. Where’s yours Beau? Oh yeah, on paper yours is better. I forgot.
But your success might be on paper, while the rest of live in reality.
Face facts, Beau. Your systems have great backtests, but they stop there. Nothing here, at Covestor or any other place you troll shows anything of value.
You are NOT the world’s greatest strategy developer. You graduated from a no-name school where apparently they don’t teach you how to spell Santa Claus (see another forum post of yours tonight “Clause”). You work in a back office. You failed the CFA exam. Your millions in funding is now only a fantasy.
Listen to what smarter, more experienced people are telling you. You might learn something.
Beau, I really tried, but a lot of what you said I couldn’t understand.
I’ll respond to your points one-by-one.
Beau point: So, how is that buy when the market closes down 3% and sell short when the market closes up 3% system doing for you, Mr. Keith “Only-need-to-make-a-penny-to-prove-robustness” Fitschen?
Response: I have no idea what you’re talking about.
Beau point: These last two trades tell most of the story to me, as below:
Response: There are no two trades after your colon, only blank space. And I have no idea what story the trades are supposed to tell.
Beau point: Why don’t we test if ETF Timer’s performance is luck.
Response: OK, I somewhat understand this one. After a bunch of numbers you come up with a “luck coefficient” of about 3 and say anything under 10 isn’t luck. So you’re changing your mind and think ETF Timer is a good strategy?
Beau point: "There is no apparent strategy, articulated, or espoused as the strategy by any subscribers"
Response: I suspect you mean the vendor, not the subscribers, and I’m with you on this one: how the hell does he do it? Cumon Mr. ETF Timer tell us how you can be so uncanny at calling the market turns?
Beau Point: If you had research that looked as good as mine, wouldn’t you post the hell out of it?
Response: Beau, I’ve told you this before, but I’ll do it again. The reason your research looks so good is because it’s a giant curvefit. Mechanical strategies need thousands of trades before you can rely on them performing any where near the same in real time. When you first came here you bemoaned the fact that there wasn’t a lot of data for QLD and QID. You had less than 50 trades in your backtest. I believe you curvefit your system to the small data sample. Exact same thing with your Hershey system. That’s why I call you Backtest Beau, you show these wonderful backtests but when you trade the strategy it comes nowhere close to your backtest. Someday you’ll see the light. I suggest you do what everyone else at C2 does. Let your real time trading do your talking, not your backtests.
Beau Point: Every statistics book I’ve ever written states a distribution with a sample size of 30 is when you can start using Z-scores.
Response: Exactly how many have you written? Just kidding.
Real Response: Do you really read what others have written? You just quoted what I said almost exactly, “That’s where the number 30 comes from in the literature: the mistaken belief that 30 trades is a “statistically significant” number of trades. No, it’s just a sufficient number to start doing T-tests, Z-tests, etc. on the distribution.” The point is 30 tests may, or may not be sufficient to characterize the underlying sample. Reread my “sock example”.
Beau Point: I consider any ETF or security built to an exact formula stationary.
Response: Beau, read about stationarity in any statistics book. Whether a process is stationary or not depends on the statistics of the sample over time. Stock market data is not stationary, and any exact formula thats builds a tradeable from non-stationary data will produce a product that’s non-stationary as well. That’s why trading is so hard. You’re trying to hit a moving target.
Summary: My post examined this statement of yours, “The only thing ETF Timer has done is buy and hold with QLD. Whatever trades he does in between his entry and exit, are irrelevant. It’s still one trade.” I showed that ETF Timer did not “buy and hold” QLD for two years, and that it’s had many more than one trade. Along the way, I interjected my belief that ETF Timer is a remarkable system, especially during such a turbulent two years. It appears that you concur that the performance was not luck, and that your only problem with the strategy is that the guy won’t tell us how he does it. So I agree on both those points.
It appears that you concur that the performance was not luck, and that your only problem with the strategy is that the guy won’t tell us how he does it. So I agree on both those points.
Then we agree. The Luck Coefficient just says how many times greater the largest winning profit is compared to the average profit. So at 2.96, it states that it’s largest winning profit was only 2.96 times it’s average winning profit, which is good.
You and I have very different ideas about creating trading systems. I don’t find thousands of trades for a backtest anything but an excercise in attempting to show statistical significance, and I know ten thousand trade backtests only enrich brokers. There is validity to a system that lasts through several market cycles, providing there is more than 30 trades. I know and am pretty confident that this time around, the performance will show as the backtest does, and Pairs Trading QID QLD Scalper was still an excellent system, beating 99% of all fund managers in it’s time period. Of course you can always say there are better systems, but I do not believe for a second that they are as robust as PTQQ2.
Curve fit? Again, all of the variables were the same, and the only addition was in what I’d call “contingencies.” All of the values remained the same, but what changed was the approach to stop outs and losing trades. Rather than waiting for new trades, if we were still overbought or oversold, we’d enter, and that is the only change.
If you don’t remember what I’m talking about with the 3% from close system, you don’t recall claiming it would perform as well as PTQQS, either, which is what you said. It’s an old post, I think more than a year old now, but I still remember it. I also recall how stupid the idea that it was anything similar to my logic was.
So much rage, in someone most likely unqualifed to make his own investment decisions.
Where’s mine? It’s PTQQS, beating 99% of all fund managers during it’s time as an active system. Do your own research if you’d like to verify it for me. Find a system annualizing 13% since March 21st, 2007, through November of 2009, because I didn’t find any when I looked through the fund screeners.
You also have no idea what it takes to build systems. Getting a strategy is probably 95% of the battle, which Keith grossly ignores, as do you, mistakingly dismissing an 88 trade, 3 year old backtest as not statistically significant.
Over the next two years, I know my system will perform as the backtest has. Covestor just happened to not include two trades in April, or it would be about 17% higher. Too bad, but not thinking I didn’t make those trades is a mistake.
Should be mistakenly.
I’m only 10% away from being in the 900’s, BTW. Since you’re so new, you’ve never seen my vendor rating fluctuate for no apparent reason. It will be above 900 by this time next month.
Beau,
For me, one of the most important items to consider is the risk to reward ratio.
Since ETF timer has been in existance, your system has lost money while ETF timer has had a gain of 300%. Your system had a draw of over 37% while ETF timer is under 27%
While I appreciate your cautionary comments I think you are fighting a losing statistical battle with ETF timer.
We’ll see about your rating in a month, big guy.
Until then, keeping plugging away in the back office, and studying for the CFA exam you failed.
You know I will. It reminds me of the literature they give you when you sign up that pretty much guarantees you’ll fail at least once. They had a story of one of their charterholder that failed each level twice before receiving their charter. You seriously underestimate someone who can pass even level I.
Walt, that battle will now be fought with [LINKSYSTEM_44396934], and I do believe I will win. PTQQS is a thing of the past, and PTQQ2 will overcome.
Beau this is reminding me of all of the gang bangs you have participated in the past. The only difference is now you are now in the barrel.
Maybe a quieter approach with better returns would speak louder?
Rick
There’s always talk, Rick. It doesn’t bother me. There’s no doubt in my mind how valuable my systems are.
My last comment to you, Keith, is that Jack Hershey’s system is older than dust.
His system has been around and described for over 10 years. Only at the end of 2008 did a developer make the foolhardy mistake of posting TS code to finally bring Hershey’s system to light. They failed, and could not make it work on a walk forward or backtest basis. This failure was because the system I believe was written by Spydertrader and optimized on .SPX data more than 5 years ago. The system itself is an ancient relic, at least to me, having seen “Jack Hershey scripts” in WL since I started.
The working version was never published till december 2008, and the backtest you see in [LINKSYSTEM_41404319] hasn’t ever been optimized. It has only been converted from Easy Language, to Wealth Script Perl, to Wealth Lab .NET code. Now, I think that system has yet to show it’s true colors, especially, but it’s not trading QID and QLD.
Rick, I don’t speak my mind very often. This is probably the only mass posting I’ve done on Collective2 in more than 2 months, since I started PTQQ2, really.
Keith agrees with me on both points I’m making: 1)Probably is not luck by the luck coefficient. 2)There is no strategy to ETF Timer.
2 gives me great pause, and reminds me of the attitude I was given over my criticism of MVP-3. I feel like if you don’t know what your vendor is doing, it’s a case of "making money for the wrong reasons."
I speak my mind about one system so obviously absurd, like judgement of television news anchors based on the colors of their clothing, and I get called a bully? Unbelievable. Any rational vendor would have seen through that, but this situation eerily reminds me of MVP-3.
Anyway, my main point is that I don’t want to see people lose money on something they obviously took no time to understand, and, though they have made money, I believe they made it for the wrong reasons. And that’s really the only knock that I have against the vendor, and one I see Keith agrees with me about.
"It’s PTQQS, beating 99% of all fund managers"
Correct me if I am wrong, but dont fund managers trade with very large amounts of money making it much more difficult to gain a return as high as an account starting off with $100,000. If I am correct in this assumption how can you compare yourself to what fund managers do. It appears to be little more than grasping at straws to prove something to yourself. You evidently can back test better than most and so the name Backtest Beau does seems quite appropriate and I would like to congratulate you on your high level of success there although I fail to see the good it does you.
"Covestor just happened to not include two trades in April, or it would be about 17% higher."
Ill just bet you excel at making excuses for yourself. Ive known a few other talented individuals in this regard.
Also, look at the Calmar Ratio (A ratio used to determine return relative to drawdown (downside) risk in a hedge fund, used in the hedge funds industry to gauge the performance of a fund manager) for his best system: Pairs Trading QID QLD Scalper: 0.36
Any system that has DDs more than twice the returns (Calmar Ratio < 2.0) indicates that the fund manager is taking excessive risks. Some funds have high annual returns, but they also have extremely high drawdown risk. This ratio helps determine return on a downside risk-adjusted basis.
No way, PTQQS is beating 99% of all fund managers on a downside risk-adjusted basis…may be not even 1% of all fund managers…
Amen. Backtest Beau should speak about ETF Timer (or for that matter any other system he wants to talk about) only after having results (not backtest results) better than ETF Timer (or any other system he wants to talk about).
In his own mind, he thinks that he is in the top 1% of all fund managers, but the real-time results on a downside risk-adjusted basis (Calmar Ratiio) shows that he is in fact in the bottom 1% of all fund managers…
"ETF Timer is a remarkable system"
Sorry, I disagree. I dont find anything remarkable about this system except for its equity curve. Like goodness, beauty (art or an equity chart) is not in the eye of the beholder, i.e., subjective. It is objective. It is in the object - as judged by a rational beholder.
Esthetic principles, are not the only standards relevant to evaluating a work of art (or a equity chart). Objective evaluation must recognize that art includes both esthetic means and metaphysical content. Full objectivity consists in identifying both elements, judging each rationally, then integrating ones judgements into an estimate of the total. Esthetic quality alone, therefore, is not sufficient to make a work of art (or equity chart), a value to rational men.
Since art is a philosopical composite, it is not a contradiction to say: This is a great work of art (or equity chart), but I dont like it - provided one defines the exact meaning of that statement: the first part refers to a purely esthetic appraisal, the second to a deeper philosopical level which includes more than esthetic values…
Objective analysis indicates that ETF Timer is an irrational system and is in fact in the bottom pile of C2s unproductive systems…
Any system that has DDs more than twice the returns (Calmar Ratio < 2.0) indicates that the fund manager is taking excessive risks.
Calmar ration <0.5 !!!
Palsun, you really don’t know your stuff!!!
