Cindicator analytical products: progress report and updated priorities

Cindicator analytical products: progress report and updated priorities

In this post, we share the latest results of Cindicator’s analytical products, and new priorities for questions on the platform, plus a template of a strategy for using indicators.

DISCLAIMER: This information is not intended to provide a personal recommendation or investment advice and it does not take into account the specific investment objectives, financial situation or particular needs of any specific person. We shall not be liable for any loss, including any consequential loss which may result from reliance on the information or be incurred in respect of any action taken. Past performance is not a guarantee of future performance. We cannot guarantee the accuracy of the indicators. We use reliable and comprehensive information, but make no representation that it is accurate or complete. We assume no responsibility for updating any of the contents or opinions contained herein and therefore accept no responsibility for any actions taken based on the receipt of this communication.

Over the last three quarters we have collected a new and extended data set and that means it is time to share the latest results for Cindicator’s analytical products. We will also outline our new priorities for questions on the Collective Intelligence platform, a template of a strategy for using indicators, and a new source of data for our ecosystem.

Accuracy over the last nine months

One of the goals for creating the CND utility token was to decentralise the model that utilises the value generated by Hybrid Intelligence. Generating value for holders of CND is the guiding star for the Cindicator ecosystem.

Ever since our first progress report in January 2018, the accuracy of indicators emerged as the main metric. It has attracted the most attention from our community, and is widely discussed online as well as during our customer development sessions with token holders. It’s easy to see why: predictive analytics with high accuracy clearly provide an edge in the market. Furthermore, with higher accuracy comes higher returns and thus higher value for our token holders. Thus, it is not only accuracy that matters but also the value of the indicators themselves for trading.

We believe that through building a system that constantly produces accurate, high-quality analytics, Cindicator can help its token holders to make better investment and trading decisions. That is why for the last nine months we’ve been focusing on adjusting questions, improving our machine learning models and experimenting with new types of indicators, all to create a stable flow of precise indicators. Let’s take a deeper look at the results and review the indicator performance.

Over the last three quarters, our achievements include:

  • 2,039 binary indicators, 33% more than during the previous three quarters;
  • Surpassing 100,000 weekly individual answers in April 2019;
  • Average of 637 binary indicators per quarter, compared to 619 indicators per quarter during the first half of 2018;
  • 353 support and resistance level indicators;
  • Average of 117 S/R indicators per quarter, 35 indicators fewer than in H1 as we have been experimenting with new types of questions.

The overall accuracy during the last two quarters was:

  • In Q3 2018: 57.09% and 68.38% for binary and S/R indicators respectively;
  • In Q4 2018: 59.05% and 72.13% for binary and S/R indicators respectively;
  • In Q1 2019: 68.27% and 75.68% for binary and S/R indicators respectively.

Since the beginning of our observation period in Q1 2017, the overall average accuracy of all indicators stands at 62.02%. Over the past two years, the market has changed dramatically, proving many experts wrong again and again. We believe that this overall accuracy is an outstanding result that shows the remarkable resilience of Hybrid Intelligence as a predictive analytical instrument.

Also, you can see that Cindicator has drastically improved the accuracy of binary questions over the last quarter. There have been a few great drivers of this result: our new type of binary question and the fact that we moved some of our experimental questions to a separate entity. We will discuss both of these factors later in the article.

As usual, we are sharing the full data and invite you to carry out additional research and arrive at your own conclusions.

Updated priority: conditional questions

We have recently discovered a great opportunity to create more value for CND token holders.

We observed returns from standard binary questions, the staple format that asks whether an asset will reach a target price by a certain date. Although the accuracy remained high, occasionally some traders couldn’t extract value from the indicators.

Here is an example. Suppose we had a simple binary indicator regarding the BTC price for the previous week and it was quite accurate – the price went above a specified price of +5% relative to the price at the beginning of the week. But let’s say that before that the BTC price had dropped by 9%. The indicator was correct in this case, but there’s a high probability of a bad final outcome because each trader has their own rules for risk management. For example, a trader might choose a 1-on-1 risk/reward ratio, meaning that for every $1 of potential gain they would risk $1. So in this particular case if a trader had decided to enter a trade, then they would have set a stop loss at -5% or even more. So the indicator is accurate but the trader closes the position with a loss.

It’s clear that indicators are valuable to traders – the participants in our trading contest used them to generate 21.53% in BTC in eight weeks. Yet standard binary indicators also introduced additional uncertainty. This type of indicators provided a target price but no clear exit points for cutting losses and taking profits. Traders needed to pick those independently, with some doing that successfully, and others less so.

To counter that, we have started to experiment with conditional questions. These are questions that ask if an asset will “trade above X earlier than below Y”. Here is an example:

The cryptocurrency Bitcoin (BTC/USD) settled at 3696 USD at 06:30 AM UTC on the Bitfinex exchange on Sunday, February 17. Will BTC/USD trade above 3760 USD (+1.72%) earlier than trading below 3632 USD (-1.72%)? (forecast 51-100% — bull scenario. 0-49% — bear scenario)

We introduced these new questions in April 2018, though we did not want to give them a full-scale roll out until everything was ready. First, we wanted to collect the necessary data set for training machine learning algorithms. Second, we had a limited capacity for asking questions – all of the questions were posted to all analysts, who could only answer so many questions. The second issue was partially solved through “private challenges”, which launched recently.

Here is how indicators based on conditional questions have performed:

  • 347 conditional indicators issued over the last three quarters;
  • 60.98% – average accuracy of indicators issued during the last three quarters;
  • 62.7% – average accuracy since the beginning of 2019;
  • 0.69% – average profit or loss per indicator during our test period.

New challenges open new horizons

Last year we faced some very challenging questions – how we can experiment with a new type of indicator without reducing the number of existing ones? How we can change our current questions in the most radical way possible without risking our accuracy?

Our team came up with an elegant solution — private challenges. Private challenges are temporary events dedicated to specific market events, assets or markets. They usually last several weeks and focus on a very specific question type. Private challenges help the team to test new hypothesеs and questions over the course of just a few weeks and to introduce the best ideas to the main products. The whole initiative kicked off through a collaboration with one of the leading South Korean crypto exchanges. Since then we have been gathering momentum and Cindicator now launches several private challenges every month!

For our private challenges, we carefully select the best analysts with profound expertise in their respective fields. For each private challenge, we have separate prize fund, which is determined by the number and complexity of the questions we ask in the challenge. So far Cindicator has successfully closed 5 exclusive challenges on both crypto and traditional markets. The chosen analysts predicted the EPS of Fortune 500 companies, ranked fiat assets, answered advanced crypto asset questions and more. We’ll share more details about the private challenges in a separate article.

At the moment we are using our private challenges exclusively as a source of data for internal research and as a way to test some of our most daring ideas. Our team is also considering offering indicators created through private challenges to our token holders and external funds, providing an opportunity to experiment with new strategies based on data generated through private challenges. If you're interested in finding out more about how you can utilise private challenges for your own needs, fill out this form and we'll get in touch with you.

Creating your strategy

Every trader probably wants to know how to use the indicators in the best way possible. Yet the ideal is far from attainable, as every trader’s skill set and worldview are unique and create a nearly limitless array of possibilities for forming a strategy.

We saw various ways of using the indicators during the trading contests. Participants came up with a wide range of strategies. We shared all their trades. Nevertheless, codifying the strategies proved difficult as they are intricately tied to traders’ technical analysis, risk management, and many other factors.

Given that the field for creating your own strategy is quite broad, we can only provide an example of a strategy that you can adapt and modify to suit your capital, skills, and risk profile.

To trade conditional indicators with a 1-on-1 risk profile, a trader might start with the strategy outlined below:

  1. Risk contribution. Select an amount that you would use for equal risk contribution. For example, you might decide that you would risk 1 BTC for every trade. In that case, every trade would result in either a loss or gain of one BTC.
  2. Timing an entry point. Enter trades for all conditional indicators as soon as you receive them.
  3. Position sizing. Depending on the target in the indicator, select the position size so that possible profit or loss is equal to the amount you selected in step 1.
  4. Set the stop-loss order at the level that would result in a loss of the amount selected in step 1.
  5. Set the take-profit order at the indicator’s target.

This is not a prescription, but rather a template that you can customise to suit your own strategy by modifying different elements:

  • Trading based only on certain indicators according to your personal rules;
  • Using different timing;
  • Changing the entry point;
  • Setting stop-loss and take-profit orders to suit your risk tolerance and capital.

Next steps

In this report, we have outlined the evolution of Hybrid Intelligence indicators. Through a series of tests, we are gradually developing our indicators and adjusting our questions. We have launched new private challenge mechanics which are helping us to experiment and expand our product at a tremendous pace.

The framework strategy that we have presented above reflects our findings while utilising Hybrid Intelligence indicators based on our experiments in the second half of last year. We know that our community has a wealth of insights and ideas that could be used to customise strategies for individual needs. We invite you to apply to join an intimate Telegram chat where you could freely discuss trading strategies with our team and other community members. We believe that this would help all of us grow even faster.

Of course, this journey is only possible thanks to the continued support of our token holders and dedicated efforts of our community of analysts. Thank you for being a part of our community! Let’s keep working together to build Hybrid Intelligence.