|
An algo based on our proprietary pricing model
typically consists of many PROPRIETARY modules
Which run in a dedicated environment and relative virtualized architecture.
Typical PROPRIETARY modules of our algos are:
- autonomous raw intel gathering agents
- non relational data processing and event driven profiling
- situational awareness models
- biased dataflows spotters
- counter-profiling models
- analytics and dynamic algo rules
- market pricing models
- profiling of syntethic price dynamics
- filters against biased publicly accessible price streams
- counter-manipulation techniques against non transparent price feeds
- counter-baiting techniques against manipulated price feeds
- counter tactics against market players typical behavioral models
- simulator of effect distributions of disruptive market behaviors
- sniffers to take advantage of inefficient market micro-structures
- ssniffers to take advantage of exploitable externalities or events
- sniffers to take advantage of exploitable market inconsistencies
- sniffers to take advantage of network vulnerabilities and subsequent lags
All the above are PROPRIETARY modules,
meanwhile eventual low latency and ultra low latency modules
are from third parties thus are not disclosed as by NDA agreements.
An example of asymmetic information we exploit:
the OHLC model flaw:
 OHLC model has been forcefully injected into market analysis in 1920s
by banks to serve their compound interest model , banks main source of revenue,
with the consequence of blocking and replacing any alternative pricing model ever since.
Bank supported university research went solely in that direction for a whole century
and that made the OHLC model the only standard pricing model
but
as any other pricing model it was maybe correct for a certain period,
it has become obsolete around 1970s . So why is till in use today ?
Views on OHLC models can be summarized as:
- Economists view: OHLC is just a comfortable convention.
- Market view: OHLC is a worldwide standard.
- Our view: OHLC means unilateral opportunism
[from a monopolistic position who injected that into markets]
Banks willfully and knowingly refused to replace it
because brought them a steady stream of profits ,
very often OHLC model acts against clients and speculators as
they think a certain price is real while it is not.
OHLC is faking the dataset as follows:
in OHLC every candle has: Open, High, Low,Close prices.
As you choose a timeframe you ARBITRARILY INJECT
the Open and Close datapoints depending on
your subjective perception of a theoretical timeframe
so, while High and Low are real datapoints , Open and Close are artificious.
Conclusion: using OHLC
you add two artificial and subjective data every 4 datapoints ,
for each candle building the price history you are evaluating:
fact which makes any further analysis on that time-serie
not only useless but misleading also.
In our view: being the worldwide accepted OHLC model flawed
( sell-side biased ) , we prefer using our proprietary one.
Exploitability of the OHLC model flaw:
-in markets NOT ruled by compound interest based clearings
-in transactions where rollover is forbidden
-in swap free operations
-in places where compound interest is illegal
our pricing model is far more efficient than OHLC, by design.
An example of a our proprietary pricing model
running our speculative algo on Crude Oil :
Algo main task:
defend oil buyers or sellers from speculative oil price fluctuations,
and relative volatility issues.
Algo was specifically designed in 2016 for :
Oil tanker management companies. oil storage facilities and oil trading companies.
Such algo succesfully identified the incoming oil price collapse which
essentially produced a very effective early warning.
Synth Oil pricing model components which took part in the algo pool
were definitely not just bid and ask:
-
dark pool price
-
order book price
-
block trade price
-
sentiment price
-
extraction price
-
trading price
|
-
delivery price
-
apparent price
-
real price
-
reserve price
-
artificial price
-
deceptive price
|
-
politically casted price
-
retail price
-
contrarian price
-
volatility driven price
-
more... [undisclosed]
|
...dynamically adjusted given externalities and market environment dynamics.
Algo class:
active counter-profiling.
Model:
algo actively searches, finds, identifies and exploits price loopholes, price lags, recognizes media disinformation, profiles price manipulation , price stuffing and other deceptive practices typical of centralized oil pricing players and models.
Strategy:
counter-profiles the typical behavior of a price arbitraging speculator acting both in a delocalized pool and in a centralized oil market.
Special config:
Algo tracks real events, thus it does not stop posting oil quotes on week-ends: just because banks are offline doesn't mean oil price stays still or price dynamics stop. Another major mistake our competitors do is to confuse the real oil price with the one you see on trading platforms.
Reason:
oil market is creating an artificial spread between the real price the apparent price. While our Backfire model knows there ar many concurrent oil prices: our competitors, luckily, keep ignoring such fact and keep using a single centralized oil price as source. We call that price a pseudo-source. Just because their price is centralized doesn't mean is real.
Components can be automatically weighted, enabled or disabled depending on market situation (see above crude oil price structure sub-components).
Such level of detail greatly contributed to obtain the very effficient Algo test results we had.
Algo test results are avaliable to download for our think-tank partners. Algo test results are updated weekly.
External algo evaluators comments are undisclosed as by NDA but if were to be summarized would likely be this kind of statement:
"Algo performed slightly better than expectations with a very effective risk control process".
|
