Have you ever heard the story about how a successful business was built without ever signing a single contract? I guarantee you it is true and happens almost as often as symbols race down the Wall Street ticker tape. It happens daily in the Mortgage Field Services Industry. Fat cat firms shoot over dozens of pages of legalese for misclassified employees to sign and the signatory never receives back a copy signed by the party legally allowed to execute such within the hiring firm. Much the same in the real estate sector if you ask around. Contracts dart to and fro, but none of them legally valid as none are executed by all required parties. In fact, in both cases, the only people legally holding the proverbial bag of shit are those whom agree to take on all the risk and forego all of the entitlements associated with creating wealth. It’s a long con, really, and presents in game theory what we generally call the Prisoner’s Dilemma. Defection is related to the action of signing the document and neither will both defect which removes the Nash equilibrium. Generally speaking, most reading this article today are involved in an iterative prisoner’s dilemma. The question that presents is will it be a large helping of Bayesian Nash equilibrium served up or a presentation of the Monte Carlo? I digress.
Algo trading automates the trading process in financial markets by rapidly and precisely executing orders based on a set of defined rules. They remove human error (provided the algorithms were developed without them) and they also remove the dangers of acting on emotion. The algorithms that are used in production can be fairly complex and heavily optimized with low-latency systems.
In fact, algorithmic trades account for over 80% of all US trading. Ok, so what does algo trading have to do with the news? Glad you asked,
Sentiment Analysis or Opinion Mining refers to the use of NLP, text analysis and computational linguistics to determine subjective information or the emotional state of the writer/subject/topic. It is commonly used in reviews which save businesses a lot of time from manually reading comments.
Within almost all data structures there exists a form of metadata. What I am driving at is today there exists the ability to scrape data and crunch it vis-à-vis natural language programming (NLP). In fact, Sentiment analysis (also known as opinion mining or emotion AI) refers to the use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affective states and subjective information. An example of how this is done uses Valence Aware Dictionary for sEntiment Reasoning (VADER). VADER is a lexicon and rule-based sentiment analysis tool that is specifically attuned to sentiments expressed in social media. VADER is publicly available, in certain flavors, at GitHub and other places.
In a nutshell anything that appears on the internet may potentially have a positive or negative value. In the same way that machine language works — zero is off and 1 is on in binary — it is possible to construct opinions about information including the interpretation of, say, emojis used in texts or social media posts. Sentiment ratings from 10 independent human raters (all pre-screened, trained, and quality checked for optimal inter-rater reliability). Over 9,000 token features were rated on a scale from “[–4] Extremely Negative” to “ Extremely Positive”, with allowance for “ Neutral (or Neither, N/A)”. We kept every lexical feature that had a non-zero mean rating, and whose standard deviation was less than 2.5 as determined by the aggregate of those ten independent raters. This left us with just over 7,500 lexical features with validated valence scores that indicated both the sentiment polarity (positive/negative), and the sentiment intensity on a scale from –4 to +4. For example, the word “okay” has a positive valence of 0.9, “good” is 1.9, and “great” is 3.1, whereas “horrible” is –2.5, the frowning emoticon 🙁 is –2.2, and “sucks” and it’s slang derivative “sux” are both –1.5.
Personally, I use a Python based script which utilizes the Natural Language Toolkit (NLKT) — VADER is built in — when I deep dive any manner of data for sentiment. When I need to understand a real world inference of a given subject matter, I generally deploy Reddit’s PRAW API. Reddit has roughly 430 million users and as opposed to Facebook, it allows me to more deeply discern granular data specific to a single query. If you take trump + investigation news you come up with an interesting paradox wherein the combination is both positive and negative depending upon whom the beholder of the query is. Power-law distribution aside, algorithms must be properly tuned.
In the case of Treasury Secretary Mnuchin and House Speaker Pelosi, almost every 120 minutes a permutation occurs wherein stimulus talks + election occur. Almost to a fault, there is not a single American whom believes that any stimulus bill will be passed before the Presidential and Senatorial elections. We would call this an intuition variance. The algorithms, though, do not have the ability to infer intuition or rhetoric. Here is the irony, the Administration is Republican and the vast majority of media are controlled by Democrats. Why would Democrats want to facilitate profiteering by Republicans?
It is, once again, the Prisoner’s Dilemma except on a far more grand scale. To publish, or not to publish? And will publishing assist or detract? For example, from an editor’s point-of-view, are the Republican’s refusal to capitulate valid as headline news? If published, the headline would paint the Republicans in a negative fashion and thus scorable. Is this worth the potential trade off of elevated stock market profits? For you see, the elevated stock market profits would most assuredly benefit the Republicans and thus, why would Democrats want such benefit accrued? Cyclically, we are able to quantify the amount of both time and the iterations of frequency — both are finite and are reduced as we come to the Presidential Election. Google reports that Mnuchin + Peolsi + Stimulus + Talks returns roughly 6,380,000 pages. And to quantify that number, remember, Google only tracks roughly 04% of the entire internet available — that is not a typo.
I posit that at such grand levels, terms like Democrat and Republican only exist for the plebes. It is uncontested that enormous hedge funds have little to no interest in elections. Fact of the matter is that no matter which party is elected, the oligarchy is controlled by those capable of providing the necessary lubricant of money to the machinery of commerce. And to that point, whether by an army of lobbyists; whether by strategic placement of government officials in key administration positions; or whether by the outright hiring away of talent capable of prosecuting against nefarious actions, only the plebes continue to believe in a fair system.
The algorithm churns insistently in much the same way your auto correct spelling system works. It intuitively learns based upon the amount of information it is capable of siphoning off and processing. I am not saying that algo trading platforms are sentient. I am saying that they are controlled by entities whom give no more thought to whom is in office than passing salt at the dinner table. Remember our aforementioned friends whom wished to build the homes for the hedge fund or the misclassified employee in the Mortgage Field Services Industry? Whether or not the hedge fund signs the contract is immaterial to the builder in that unless enough pain is exerted upon the builder, the signature is immaterial. The interesting thing, though, is that commerce moves without the legal landscape of contractual law — it is Schrödinger’s cat displayed in epic proportions.
While I am not insinuating that the paradox of choice does or does not exist, I am saying that the next time you think that things exist and that you have any control over them, perhaps you might want to look behind the smoke and mirrors and pull some of the levers yourself!