Data Scientist, Don Philip Faithful, in 2018 published a post on DataScienceCentral.com – a Data Science oriented blogging platform – in which he proposes the use of a systematic, or algorithm based approach to market analysis. Faithful’s experiment sought to augment the interpretation of market sentiment, by adding tendrils to the anatomy of a conventional candlestick to display price change data.
Financial market chart analysis, whichever asset class a trader operates with, usually involves scanning the statistical information that is visually represented in price charts, and employing a combination of the profession’s best practices, and a touch of gut feel to determine patterns upon which to base trading decisions. Faithful was, at the time, of the opinion that one could, possibly, fine tune that analysis and come to a closer-to-accurate trading call, using mathematical algorithms for the process of pattern recognition.
For the purpose of illustration, he presents a green candlestick – nothing extraordinary, just a decently sized body and wicks of equal length – and slaps tendrils on it. Test-driving the Tendril-Candlestick on a historical chart yielded 505 combinations (That’s Data Science talk for a set of stochastic – non sequential – scenarios that could play out). A tendril attached to a Doji candlestick, interestingly, presented an extra combination (it really could go either way with those things).
Faithful then did something else that is out of convention in the analysis of historic price data. Instead of feeding past price data into the program first, Don Faithful applied a back-to-front approach to the process and fed future data into the system first. The logic he presents for this decision is that using the data in the correct sequential order – while enabling one to buffer past information – does not allow for the buffering of future data. Reversing the data read, meant that Faithful was then able to ascertain the accuracy of his algorithm,“..from the perspective of the read, the program cannot know what it hasn’t already read.”
Whether the data was read conventionally, or back to front, essentially made no difference, as Don Faithful set the algorithm to consider each instance of a candlestick’s tendril and not merely the previous data, in order to make a call. He found that some combinations, or sets of combinations could be correlated with particular past price movements.
The Results
In his analysis of the program’s efficacy, the data scientist fed historical price data from Trans Canada Pipeline equity markets into it – having already pre set Tendril-Candlestick combination codes into the program. Could it predict the outcome? As stated above, it had some misses, and it made some spot on predictions. Whatever correlations between combinations and price movement the system identified also didn’t hold over the long run.
Faithful however, did not provide any comparable numbers between the algorithm’s hits and misses. He stressed that the article was merely for the purpose of presenting an idea to the Data Science community, admitting that it was a work in progress that would require a good amount of tweaking.
The Potential
Though the Tendril-Candlestick chart was merely a concept – and we could find no indication of anyone having run with the idea, or that Faithful himself, continued to develop it – it presented a novel approach to the discipline of Technical Analysis. One that could have ushered in a new wave of deeply mathematical trading indicators. Allowing traders deeper insight into the market sentiment behind more volatile markets; like digital assets. What has been worked in traditional asset markets, in terms of Technical Analysis tools, has been done just as well in crypto, so far.
We reached out to Don Philip Faithful to hear if he had made any progress with his algorithm, as well as it’s applicability to digital asset trading, but had not received a response to our messages by the time of going to press. What is clear, however, is the potential of the concept in the development of chart indicators, that could aid digital asset market operators in making quicker calls (crypto markets are notoriously videogame-paced,) and/or in the development of algorithmic trading programs with enhanced predictive capabilities.
