Project Management of Web and App development
- or -
Post a project like this- Posted:
- Proposals: 10
- Remote
- #1537721
- Awarded
Description
While most AI companies use popular deep learning algorithms or classical machine learning, ELM gives us an advantage - speed.
Whilst others only work with partial data their models see only part of the global picture. It means their model can find good statistical patterns for some period of time, but when the environment changes the model will stop working. Their models won't be able to react quickly as they need to be re-trained.
Whereas Watstock can react near real-time due to our complete big data global data points.
ELM will give us another advantage - it requires significantly less resources, which means it can be trained and executed on mobile devices. It can adapt to device behaviour to give advice based on previous experience.
IBM Watson is just one of several technologies used to power the amount of data being processed. We are also using the worlds fastest GPUs (Nvidia Tesla P100) to process the vast amounts of data.
Our data suppliers like Thomson Reuters and PsychSignal are expanding real-time datasets which have been previously unused due to bandwidth limitations of processing.
Watstock is the first company in the world to utilise ELM with a NN. This is a new generation of AI technology.
Data-driven stock price forecasting is like driving a car by just looking at the rear and side-view mirrors, since any recorded data is either about the past or about the present at best - i.e. we can never have future data "now". The question is, can we drive a car with a blinded windshield?
Deep learning gave birth to self-driving cars, which already work very well (unlike stock price forecasting). What makes them different? Obviously, the former has enough information about the future (as the AI can observe the road ahead), while the latter lacks thereof. Nevertheless, self-driving cars are not accident-free even though they are already smart and hardly make stupid mistakes. So why do they still have accidents? It is because of either bad data (e.g. unreadable road signs) or randomness (e.g. unpredictable behaviors of human drivers).
Any predictive model fails at forecasting when it is fed with corrupted/incorrect data because such data do not reflect reality - i.e. "garbage in, garbage out". While it requires a lot of resources to prepare clean and correct data, it is not impossible to do so. The key then is to reduce randomness as much as possible. Probability Theory 101 teaches us that it is impossible to predict the exact outcome of a random process - e.g. we never know whether we will get a head or a tail from a coin flip. We can only compute the statistical properties from the past observations (a.k.a. data).
Another strength will be the democratization of AI in the financial markets.
Our competitive advantage is in the algorithms that we have developed to make ‘sense’ of the very large amount of data to make the actual predictions. We have a multi layer analysis that uses the IBM Watson as just one tier of the analysis. The deeper the layer the more you progress towards your proprietary competitive advantage that is hard to replicate, but even harder to keep up with as it is forever evolving. In other words, our customised in-house algorithms and IP is what our competitive advantage and defensibility is.
It has also taken us 8 years and 1000’s of development hours, including the expertise of the world leading professor Professor Huang Guangbin. This is because there are a number of complicated issues to overcome:
Quality of data
Timing of data
What the data says about the future
So again, the competitive advantage and defensibity really lies in the quality of our alogorithms and its ability to be accurate.
The current parameters are:
Trend directional vector next trading day (24h) prediction accuracy: 80 - 91%
Trend directional vector 10 trading days experimental prediction accuracy: 75 - 80%
Price value accuracy : 88 - 96%
Visit us at http://www.watstock.com
Also available on the App Store.
WatStock™ - The future of trading is here.
Carl F.
99% (27)New Proposal
Login to your account and send a proposal now to get this project.
Log inClarification Board Ask a Question
-
The Software sounds promising mate, but what is the requirements?
Carl F.17 Apr 2017We are looking for someone who can manage the development of our new front end platform to showcase our AI system. .
-
Hi Carl,
Can I have a chance to discuss in detail regarding your complete requirement?
Would like to show you some of my recent working as well :)
Carl F.17 Apr 2017Sure send me
-
Hello Carl,
I have been through your requirement, and would like to discuss more with you regarding same.
Could you please let me know the time convenient to you for further discussion? This would help me to provide you with exact quote and days required for the same.
Looking forward to hear from you soon.
-
This all sounds very interesting. But whats the job exactly?
Carl F.17 Apr 2017Need a project Czar who will coordinate the work between developers and data providers.