Index | Recent Threads | Who's Online | Search

Posts: 100    Pages: 10    Prev 1 2 3 4 5 6 7 8 9 10 Next
Last Post
New Thread
This topic has been viewed 6088 times and has 99 replies
judgetrade
Re: Improvement of my Ranking System

Hi Kumar,

I really am a big fan of Buffet, but not everything he is not doing is pure gambling.

Momentum is (after size) the biggest factor that can be used, even if buffet is not using momentum.

I try to be inspired by Buffet and a ton of others role models, but I try not to be influenced (total return was one of the biggest
inspirations for me, and "volatility is not risk"...).

The difference for me:

If you are inspired AND then think it trough by yourself (backtest it, think about it, reference it etc.) and only if it you believe then
its so good that it inmproves your trading you use it.

Influenced means for me that one takes it 1:1 without reflecting it and hoping that it saves you in the investing world.

if you have not it thought trough by yourself and YOU then think this it it, its fine.

Otherwise the next DD will get you out of the market, because its not the system "of you own beats".

One needs to come up with his or hers own approach...

Best Regards

Andreas

Nov 22, 2019 5:37:17 AM       
geov
Re: Improvement of my Ranking System

Hi Georg,

yes, market timing is included:

Buy rule: (close(0,#spepscy)>ema(75,0,#spepscy)) or (close(0,#bench)>ema(75,0,#bench))
Sell rule: (close(0,#spepscy)<ema(75,0,#spepscy)) and (close(0,#bench)<ema(200,0,#bench))

My take: its very loose, e.g. it will not protect you from a 20% DD, but from a 2008 scenario.

Also I regulary have a look at your work at (pops up every day at my screen when I open my browser!!!):
https://imarketsignals.com/bci/

Andreas

Thanks for posting those mkt-timing rules again. There are no rules that can protect one from a market correction. As you say, it also does not harm to look at the free weekly BCIp recession indicator, which stubbornly stays at or near 100.
Best, Georg

Nov 22, 2019 8:27:53 AM       
Edit 1 times, last edit by geov at Nov 22, 2019 8:28:41 AM
mgerstein
Re: Improvement of my Ranking System


Mark, you said:"Suppose the b figure is -1.00. Now, it's 1.20 - -1.00/abs(-1.00) which is .20/1.00 = .20."
No,in this case the result is (1.20 - -1.00)/abs(-1)=2.2/abs(-1)=2.2 (220%).........not 0.2. Am I wrong?


Whoops . . . my arithmetic was buggy. Looks like I've been spoiled by Excel.


Anyway, despite what you said, which is terribly right, the formula a/b used by Judgtrade genarally (and I do not know why) gives a better return.


Ouch. We can never ever ever ever ever ever ever ever justify doing anything just because "it give a better return," especially if we know what we did was objectively wrong. That's exactly why so many times we've seen magnificently tested models fail in real life, sometimes immediately and sometimes later on, after a run of pure luck ran out.

Andreas, interestingly, is operating on the edge of this sort of thing. We can use a bad thing if we have reason to assume it's a widely replicated error in the marketplace and that enough money is moving on the basis of a specific idea (good, bad or indifferent) to produce exploitable market opportunities.

This is an area that really didn't exist until fairly recently and came about through the combined emergence of concentration of trading power and the emergence of technological platforms and algorithms that are often driven by strategists long on math and programming but who may or may not know how to spell PE or EPS and think Warren Buffett is an all-you-can eat self-serve restaurant in a town called Warren.

It fits in the behavioral finance camp, I assume (or would argue if nobody else actually said it) and it can be a fascinating area for exploration, at least for a while (if all the techno quants were to suddenly study and apply sound principles of security analysis, the whole filed would vaporize). It might also been seen as a 21st century version of something like the old odd-lot indicator, where traders would assume that individuals, who are more likely to trade odd lots (not multiples of 100 shares) don't know what they are doing and develop strategies that go against them. It's also implicit in things I've seen form Robert Shiller and Fisher Black.

If you want to use a/b based on a belief in something along these lines, you could research it do enough tests of a/b formulations vs (a-b)/abs(b) formulations to see if these is something systemic there that's worth pursuing. But have that intention.

But just trading a/b because one used in in backtests and it "worked;" no, no . . . that's pure data mining and an invitation to trouble.

Again, you can do anything at all if you think it through and know why you're doing it, and why it would or world not work. Even something like correlation between past and future; I've often ripped Yuval an new you-know-what for this sort of thing. But if you understand what and why you're doing it, and thoughtfully monitor for indications that failure is becoming more probable, then it's fine; trend-following technical analysts do this all the time, monitoring for divergences, changes in strength of trend, etc.

Bottom line:
Nothing is good if done naively and based on nothing more than backtests.
Anything can be good if done thoughtfully with understanding or what could make it work and what could make it fail.
The backtest is like an editor who tells you that you did or did not do a good job of translating . . . a manuscript from Russian to English, a strategy from human-speak to computer (p123 in this case)-speak, etc.

Marc Gerstein
Director of Research, Chaikin Analytics
Blogs: https://actiquant.com, https://portfoliowise.com/portfoliowise-blog/ , https://www.chaikinanalytics.com/blog/
Twitter: @MHGerstein
I predict the future, as soon as it becomes the past

Nov 22, 2019 9:30:33 AM       
yuvaltaylor
Re: Improvement of my Ranking System


<Formula>ROE%TTM/ROE%5YAvg</Formula>

<Formula>GMgn%Q/GMgn%TTM</Formula>


Sorry to be a little late to this discussion, but if you just use a minus sign instead of a slash you'll probably get more meaningful results and you won't have to be bedeviled by negative numbers twisting your results into 180-degree nonsense (as Marc so rightly points out). In general, when comparing a plain number to a previous number (e.g. EPS, sales, gross profit) to measure growth, the (a-b)/abs(b) formula is best. But when comparing a ratio to a previous ratio, a minus sign may be better.

Let's take an example. Amazon's operating margin improved from 4.99% to 5.33% over the last twelve months. Microsoft's operating margin improved from 32.47% to 35.2%. Which is more significant? If you use the minus sign, Microsoft has a much more significant improvement (2.73% versus 0.34%); if you use (a-b)/abs(b), Amazon comes pretty close (Amazon 6.8% vs Microsoft 8.4%).

Well, the reason Microsoft's improvement was almost 3% is that their sales increased by 15 billion and their operating income increased by 8 billion. And that is very impressive indeed. Amazon's sales increased by 45 billion and their operating income increased by 3.6 billion. Which is not so impressive.

Now I personally believe that a huge growth in operating margin is not a good thing, but that's a subject for a different discussion.

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

Nov 22, 2019 2:57:08 PM       
judgetrade
Re: Improvement of my Ranking System

Thank you!

a)
<Formula>ROE%TTM/ROE%5YAvg</Formula>

<Formula>GMgn%Q/GMgn%TTM</Formula>

So like this:

b)
<Formula>ROE%TTM - ROE%5YAvg</Formula>
<Formula>ROE%TTM - ROE%5YAvg</Formula>

Here are the results, not much different from the a) version.

But I get your logic.

Regards

Andreas

Attachment aaa.png (296551 bytes) (Download count: 200)


Nov 22, 2019 6:07:43 PM       
Edit 3 times, last edit by judgetrade at Nov 22, 2019 6:18:02 PM
J3th
Re: Improvement of my Ranking System


yes, market timing is included:

Buy rule: (close(0,#spepscy)>ema(75,0,#spepscy)) or (close(0,#bench)>ema(75,0,#bench))
Sell rule: (close(0,#spepscy)<ema(75,0,#spepscy)) and (close(0,#bench)<ema(200,0,#bench))


Hi Andreas,

if the bench price is above its ema 75, but below its ema 200, there could be a buy signal and a sell signal at the same time. I do not know what the program does in this case. I would use the same ema for buy and sell rule.

Best Regards
Matthias

Nov 23, 2019 2:03:37 AM       
Edit 1 times, last edit by J3th at Nov 23, 2019 2:05:09 AM
Nicoletta
Re: Improvement of my Ranking System

The sistem before decide what to sell than what to buy, so if the stocks go down ema 200 is sold and can be purchased again only if it goes up on ema 75 (and the other rules satisfied).

Dec 1, 2019 9:50:10 AM       
Edit 2 times, last edit by Nicoletta at Dec 2, 2019 3:26:38 AM
Nicoletta
Re: Improvement of my Ranking System

Andreas,

can you show PLS the performance histogram of your RS since 2014 (or similar) and with a weekly rebalance? You use Uneverse =No OTC?
Do you use Percentile NAs Negative in your RS?
Fabio

Dec 1, 2019 10:00:31 AM       
geov
Re: Improvement of my Ranking System

Andreas, can you please post performance curve for your small/micro-cap model starting 1/1/2004. Many of us have been restricted by P123 to a 15 year backtest period.
Thanks

Dec 1, 2019 12:10:44 PM       
judgetrade
Re: Improvement of my Ranking System

sure, here we go (with variable slippage)

Attachment aaa.png (331473 bytes) (Download count: 125)


Dec 2, 2019 4:31:36 AM       
Posts: 100    Pages: 10    Prev 1 2 3 4 5 6 7 8 9 10 Next
 Last Post