Anyone have any at least somewhat recent publications regarding quantitative (and mechanical to a degree) asset...

Anyone have any at least somewhat recent publications regarding quantitative (and mechanical to a degree) asset analysis?
So much information is guarded IP, I can't seem to find much information on at least somewhat recent innovation.
If any of you guys have been researching it, or have researched it in the past, I'd be greatly appreciative of any information as to the approach you took in building a knowledge base.

Other urls found in this thread:

arxiv
twitter.com/NSFWRedditVideo

Your image looks like Statistical Arbitrage, where statistically "cointegrated" (stronger relation than correlation) securities are located and when one of a cointegrated pair moves, a bet is placed on the other very shortly afterward before it responds.

Otherwise, you're playing time-arbitrage with future players. The costs of "close up" technical trading will rip the retail trader's face off and rape his nostrils. Daily ATR is down heavily over the last 5-10 years relative to further back in time, which means commissions and fees will murder all gains (and then more) by the average retail day-trader on 80% of trading days.

I've said nothing revolutionary here. I too am studying the finance industry but so I may "learn the game" and play competently now that I've aged into Wizardry.

yea the picture is just the pic from the time series wiki page.
I do agree with your opinion on "close up" technical from a retail standpoint. However, I feel it is safe to assume there to be practical retail applications to quantitative analysis. I'm in the processes of changing my major from finance major-math minor to math major-finance minor. Unfortunately, free information on actual applications are scarce. I fear that I wont be able to apply my own knowledge of this type of analysis with confidence at a profitable level until I'm at least in grad school and have a better understanding of the scholastic research and development process. And more importantly, where the scholars have been lazily incorrect in their assumptions, because they have no money of the line.
I'm just shooting in the dark hoping someone had some background in the area. I'll probably ask Veeky Forums as well, maybe they can shed some light on this.

If ATR is lower, would you think trend following TA techniques are best for retail trading?

classically a lower ATR means higher order size, as it means (or used to mean, I'm not sure what information has been proved wrong) that the volatility is lower, so yea kinda. Technical Analysis has quite a bit of vodoo still being taught though. Most indicators have been debunked, and the ones that haven't only work as component of a mechanical system. Drawing any lines on the graph though has been laughed at for decades by statistical analysts. Classical standard deviation models are even losing validity. However, times are changing, and and its really impossible to theorize what works and for how long without hard testing it on the largest data set you can get.

ill bump once, but there doesn't seem to be any interest in this subject.

That's genuinely interesting thread user. I'll keep bumping it for answers. Math major here, trying to get into finances and apply the knowledge as well (however it seems that I am far behind you).

>Most indicators have been debunked, and the ones that haven't only work as component of a mechanical system.
May you expand on this? Which of popular indicators have been debunked and which are still believed to have some degree of applicability?

What markets are you specializing into?

That's interesting you say that about standard deviation models, would you expanding further (if you want, I know it can be quite detailed).

My opinion of quantitative analysis is that it often misses too much context. It's very good at quantifying what something did, relative to something else, over a certain time. And it's certainly a good way to make judgements on combinations of investments.

But I personally think that knowing the trend(s) something followed (and measuring, and systematically speculating where it'll go in the short term), over an amount of time, and why it followed them, gives me a better understanding of that item, as it is more precise imo.

Do you think limitations in QA models could exist due to them being too blunt in their approach?

Bumping again, because if Veeky Forums doesn't produce more valuable substance I'm out of here. I don't have an interest in hanging with NEETs larping about lambos if that's pretty much what's here.

I personally think the MACD is very helpful for swing trading. However, certain items "follow" it better - commodities like the precious metals as an example. Also Bollinger bands are very helpful with identifying where to reduce a position. If TA truly is just a self fulfilling prophecy, then so be it. I'm not going to ignore any impact on my position.

>identifying where to reduce a position
How do you use it to do that?
Usually I use BB as indicator of decreasing amplitude of momentum to determine roughly a moment to get in.

==== Do you have any opinion about popular technical overbought/over-sold-type indicators, such as RSI and stochastics? ====
I think these indicators are nearly worthless. I'm not implying that you shouldn't do research on these approaches-you can be very promiscuous in your research, but not in your trading.
==== Having done the research, would you term these approaches "bogus indicators"? ====
Yes, they're close to zero in terms of their profit expectations. What these patterns make during market consolidations, they lose during trends.


This is from an interview with William Eckhart from Jack Schwager's "New Market Wizards" which is a compilation of interviews Mr. Scwager did with seasoned, successful individuals in finance. And keep in mind, this book was published in 1993, I highly recommend reading the interviews with the systematic traders. I'm currently expecting to end up in the derivatives market, as the real innovation in this subject includes the ongoing research into risk management when dealing with leverage to maximize returns. But who knows my plan is to learn the math, and the applications will follow.

last question didnt fit.
==== Why do you believe these approaches are so popular if they 're ineffective for trading purposes? ====
For one thing, when you look at these indicators superimposed on a price chart, they look much better than they really are. The human eye tends to pick up the times these indicators accurately called minor tops and bottoms, but it misses all the false signals and the extent to which they were wrong during trends. Formally, the mistake is the confusion between prior and posterior probabilities. For example, it's true that a lot of extremes have reversal days. [A reversal day is one in which the market reaches a new high (low) and then reverses direction, closing below (above) one or more immediately preceding daily closes.] All that's telling you is the probability of having a reversal day given a price extreme. What you really want to know is what the probability is. of having an extreme-that is, a sustained change in market trend-given that you have a reversal day. That is a very different probability. Just because one probability is high, it in no way implies that the other one is high as well. If 85 percent of all tops and bottoms have property X, but property X also occurs often enough in other places, using that indicator as a signal will rip you to shreds. Also, these approaches are appealing because they play into powerful human tendencies that induce one to trade countertrend or to abbreviate trend-following trades. It's always tempting to liquidate a good trade on flimsy evidence.

You're half right. Monroe Trout for instance has been extremely successful using quantitative systems in unison with discretionary investing.
But on the other hand, what is speculatively regarded as the most successful hedge fund is Renaissance Technologies, who employ mostly mathematics, physics, and statistics. So obviously there is still quite a bit of research to do.

If I'm going long, as it approaches (or touches) the top band, I slowly reduce the long position. To what extent I reduce it depends on momentum indicators, and a bit of guessing. TA essentially points one's dick into the right hole, so to say. I think people approach it incorrectly, thinking it should 100% formulaic in it's approach. That is where QA and TA differ, I guess.

*its

from what i understand reading trout talk about it, the last stand of human intervention is the limitations of computers to accurately use information like the noise level on the floor, the psychology in the order book, and interpreting news. He even talks about, in that same book I mentioned before, manually overriding his systems to account for major news events. This is the next task to solve. Creating testable mathematical models of psychology.

I've found a problem in my trading. I can often pick direction right, but I can't seem to set a good stop-point so that a) I don't get stopped, and b) my TP isn't so far away with 3:1 P:R that it's unlikely to trigger.
Trading on the hourly, all my recent paper trades in metals and UJ are getting whipsawed before they move, and it's infuriating. I've been placing stop-entries above previous swing extremes with stops at multiples of an ATR value (which is above most ATR "peaks" for the last few days) away from entry. Basically I'm buying when it passes above previous swing highs or selling when it passes below swing lows, and then stopping out.
I'm starting to think I need to buy at higher swing lows, and sell at lower swing highs, with the same values for SL and TP width from Entry.

user delivers. Thanks man!

Yeah that's a wise thing to do. I do similar things as well. However what annoys me a lot (it was touched in several posts here and yours as well) is that there is still fuckload of guessing. Most of traders are like "yeah, there's a solid math supporting Bollinger bands and Fibonacci levels; you know standard deviations, averages, golden numbers and shit maaan" it annoys me since after all BB, Fibonacci (and MACD as well - it wasn't God who said "just consider 12,26,9 EMAs for computing MACD pals" along with the 10 Commandments) and many other indicators depend completely on a choice of timeframe. As a math major I have some uneasiness taking a timeframe that will fit what I want to see.

What TA do you use for finding moments to open positions? I usually try to find accurate Fibonacci levels and place stop-limits and openings accordingly (namely, around 25% of the band size between levels I want to get in/out). It works rather OK for Forex and crypto. Consider also the possibility that you were just unlucky - don't abandon risk management rules of thumb just due to some losing streak.

generally you should weight indicators as a 1 or 0. Humans are prone to mental gymnastics to validate bad decisions , "It's always tempting to liquidate a good trade on flimsy evidence."introducing scenarios in which you are unprepared to make a decision is the issue. Creating absolutes allows testing, which without, a claim is nothing more than a claim. Financial cycles(which is also a area under both scrutenee and ongoing research so we'll see how much validity it holds) tend to mask incorrect decisions. If you've played games competitively im sure you know the idea of expected results, sometimes doing the wrong thing can work for a while (and possibly a long while given infinite variance), but that doesn't make the decision any more statistically sound. Same idea here, game theory beginning to be more heavily used to formulate new ideas about the market.

I agree, unfortunately this is where information becomes increasingly difficult to find. My next step to see if my professors have any comments on the topic, once school starts back up. Maybe reach out to some folks who I've read about that seems approachable enough to write a letter to or something. It really looks like the only way to learn about this analysis is to learn the math yourself and get creative.
Princeton has a great lecture series on finance that is based on the idea of quantitative analysis on youtube(youtube com/watch?v=vTs2IQ8OefQ) if any of you are interested in gaining some background information of the subject.

>Princeton has a great lecture series on finance that is based on the idea of quantitative analysis on youtube(youtube com/watch?v=vTs2IQ8OefQ) if any of you are interested in gaining some background information of the subject.
That looks great, thanks!

Feel free to post anything that you found useful for real world applications. My problem is that I also won't have an opportunity to discuss any economics before the end of summer break and even then I will most likely learn only a theory (that is obviously useful, but as you pointed out the Academia is not really keen on checking degree of applicability of their outputs since they don't bet on markets).

I don't use Fib. I don't think Fib works as well as advertised. If I use horizontal lines, I use either "FibRetracements" with 0.0, 0.25, 0.5, 0.75, 1.0 as my levels (quarters). I like this because (0.618+0.382)/2=1/2, which means (to me) Fibs tend to average to 1/2^n divisions. I'm thinking in an overall-effect of market participants taking positions at 0.382, 0.5, 0.618 at somewhat random.

On the hour, I use a combination of d(Stoch)/dt (which way is stoch "pointing") when it's between 20 and 80 for direction, and I look at ATR for the last few Stoch swings to see which swings d(ATR)/dt tended positive and where it tended negative. It's "telling" me which side has their pedal to the metal more (so to say).

Then I go down to the 10 minute scale and find a horizontal level on the ATR where some peaks pop through, but the vast majority are under the level, and then place stop-entry orders. I've been losing more than gaining the last 20-30 trades so at 2200 Zulu when gold/UJ opens I'm going to try a "reversion" form of entry, where I draw channels and wait for price to be over extended within the channel. I'd be using Limit orders to enter, and in MetaTrader that means they'll be decimated into dust (many small orders), but w/e C'est la Trading!

I have Ichimoku, Bollinger, and Heiken Ashi also on the chart. Ichimoku keeps me on the rails as far as the trend is concerned, Bollinger [kinda] helps me in seeing over-extension, and Heiken Ashi paints short trends in price and reduces the noise. I'm posting a superlarge resolution image of what my charts look like. Kek away!

(Current order on the charts is not filled yet. It will be filled if price moves to the upper extreme.)

I use either "FibRe........
... or I place horizontal lines at price min/max points on a much higher timeframe. I also use fib channels in thirds mode so I can gauge over-extension within a channel.

Oh yes, I use a multiple of the horizontal level I find on ATR to set distance for my SL. Currently it's 3. It was 2 before but the whipsaws were too much.

you fags can't compete against big corps algorithms. HFT will eat you for breakfast

I too and studying and trying to learn the game. And recently I started getting into learning how to do some quantitative analysis, but since I don't know much about the nuanced details of statistics so it's most likely garbage in garbage out.

...

I found this, most recent paperI can find.
arxiv org/pdf/1706.10059.pdf
It seems very clear (assuming this data is correct, I haven't checked) that quantitative analysis can, and should be utilized, even in the highly volatile environment of cryptocurrency.

I built a model that can take 6 different company's historic bid/sell return and computes the percentage of each $$ results in the lowest variance. Yet to find a portfolio combo that results in positive yield 95 percent of the time. Convinced the combination of 6 company's exists