Investigating Geopolitical Impact on Prices | Cindicator Insight Challenge
In case you don't know me, I am a Hong Kong-based trader who trades upon the influence of geopolitical events on the financial market for a living. Not your typical technical analysis, nor your typical fundamental analysis. Normally I am quite successful in predicting the outcomes of events and thus the direction(s) of market movements, but I never bothered with estimating magnitude - until Cindicator (CND).
Cindicator is a fintech company that uses Hybrid Intelligence, that is combining human input and A.I., to predict market movements. As an analyst with Cindicator, which you can sign up for free, you get a share of a monthly reward when you make accurate forecasts. This month, the shared bonus is 1.25BTC for crypto forecast and US$7,500 for traditional market.
With Cindicator (and their incentives), now I gotan excuse a reason to hone in on my magnitude-estimation skill whereas normally I'd hold an instrument for a certain period of time and then I'd liquidate it regardless of price. Questions on Cindicator combine both magnitude and duration. Furthermore, while I make plenty of forecasts each day, I do not act on most of them. And it is difficult to keep track of forecast accuracies when they don't show up on your P/L. Cindicator provides an easy way to keep track of your predictions, and you get paid, risk-free, doing so.
Looking at my Cindicator dashboard for my first week, I know my old methods have been working well (and my bank account agrees). But I know I can do better with Cindicator.
After all, Cindicator is about Hybrid Intelligence. It is only sensible that, besides improving the A.I. ensemble, the H.I. (Human Intelligence) component needs to be able to improve as well!
Currently the web U.I. is quite convoluted if you want to check the accuracy of forecasts made over time. 3 filters have to be selected and the configuration will be lost after refresh.
I hope this is where the "Statisticssoon" page will soon be able to deliver, to aggregate prediction results and compare with peers. In fact, a correlation of my input with the Cindicator bot would be lovely! (Unless it contradicts the goal of having a diverse manual forecast ensemble, then don't implement this metric.) Other features I'd love to see are, for example, to filter prediction types (binary, range, order), to filter particular instruments (BTC, ETH, SPX), and a plot of my accuracies over time. And of course, just like in the questions, attach a link to a news piece that explains the market movement on the sidebar of each forecast result. But it should not be a general news feed or else after a couple of days the information will be buried. Hopefully such improvements can benefit the community.
As my forte is in geopolitics, the last market I'd be involved in is cryptocurrencies. And yet with Cindicator, I learned a lot more about crypto than I could hope for.Including a Cold Fusion ICO scam. In fact, I created my first crypto wallet, all because of Cindicator!
However, there is a potential pitfall in the system due to the high correlation among crypto instruments. Because the underlying trend/drift of the crypto market is often stronger than the finer differences between various cryptocurrencies, analysts' answers towards binary yes/no type questions are more or less the same across the board. It would be difficult to distinguish an analyst's ability given the current market, not that it would stay this way forever. I suppose this is where the ranking/ordering question type comes in, but not sufficient in quantity to balance the other question types' award points. And when it comes to ranking, Arrow's impossibility theorem will always haunt you. Alternatively, "normalize" the cryptocurrency's quoted price somehow, such as dividing it by the market capitalization of all crypto assets.
Another potential downside to Cindicator is it being an open platform. I worry that one day Cindicator's indicators may gain influence enough to drive market movements, especially in the crypto market, like Palm Beach Confidential. By then, Cindicator's predictions will become self-fulfilling prophecies, and valuations may not make sense from a fundamental point of view (even though you'd still get profit, which is all that matters... or is it?). It will be difficult for analysts to distinguish what's "real" and what's "self-created" price movements. But then again, isn't the entire field of technical analysis sorta self-fulfilling to some extent?
#CNDcontest
Cindicator is a fintech company that uses Hybrid Intelligence, that is combining human input and A.I., to predict market movements. As an analyst with Cindicator, which you can sign up for free, you get a share of a monthly reward when you make accurate forecasts. This month, the shared bonus is 1.25BTC for crypto forecast and US$7,500 for traditional market.
With Cindicator (and their incentives), now I got
Looking at my Cindicator dashboard for my first week, I know my old methods have been working well (and my bank account agrees). But I know I can do better with Cindicator.
After all, Cindicator is about Hybrid Intelligence. It is only sensible that, besides improving the A.I. ensemble, the H.I. (Human Intelligence) component needs to be able to improve as well!
Currently the web U.I. is quite convoluted if you want to check the accuracy of forecasts made over time. 3 filters have to be selected and the configuration will be lost after refresh.
I hope this is where the "Statisticssoon" page will soon be able to deliver, to aggregate prediction results and compare with peers. In fact, a correlation of my input with the Cindicator bot would be lovely! (Unless it contradicts the goal of having a diverse manual forecast ensemble, then don't implement this metric.) Other features I'd love to see are, for example, to filter prediction types (binary, range, order), to filter particular instruments (BTC, ETH, SPX), and a plot of my accuracies over time. And of course, just like in the questions, attach a link to a news piece that explains the market movement on the sidebar of each forecast result. But it should not be a general news feed or else after a couple of days the information will be buried. Hopefully such improvements can benefit the community.
As my forte is in geopolitics, the last market I'd be involved in is cryptocurrencies. And yet with Cindicator, I learned a lot more about crypto than I could hope for.
However, there is a potential pitfall in the system due to the high correlation among crypto instruments. Because the underlying trend/drift of the crypto market is often stronger than the finer differences between various cryptocurrencies, analysts' answers towards binary yes/no type questions are more or less the same across the board. It would be difficult to distinguish an analyst's ability given the current market, not that it would stay this way forever. I suppose this is where the ranking/ordering question type comes in, but not sufficient in quantity to balance the other question types' award points. And when it comes to ranking, Arrow's impossibility theorem will always haunt you. Alternatively, "normalize" the cryptocurrency's quoted price somehow, such as dividing it by the market capitalization of all crypto assets.
Another potential downside to Cindicator is it being an open platform. I worry that one day Cindicator's indicators may gain influence enough to drive market movements, especially in the crypto market, like Palm Beach Confidential. By then, Cindicator's predictions will become self-fulfilling prophecies, and valuations may not make sense from a fundamental point of view (even though you'd still get profit, which is all that matters... or is it?). It will be difficult for analysts to distinguish what's "real" and what's "self-created" price movements. But then again, isn't the entire field of technical analysis sorta self-fulfilling to some extent?
#CNDcontest
Comments
Post a Comment