A Polish computational scientist and astrophysicist has recently published a paper that uses a physics model based on fractional Brownian motion to predict a Bitcoin price of $6,358.32 by the beginning of 2018.
Computer Models Surprisingly Accurate in the Past
Science-based predictions of cryptocurrency are nothing new — several research teams at MIT have used mathematical techniques such as Bayesian regression to predict market movements.
One notable 2014 MIT-developed machine-learning algorithm based on computation science generated an 89 percent return on investment over 50 days, attesting to the efficacy of technical analysis based prediction.
Tech pioneer and eccentric cybersecurity legend John McAfee has famously promised to perform an egregious sex act on live television should bitcoin not reach $500,000 per coin within three years.
Cryptocurrency proponents around the web are echoing the sentiments of early adopters of bitcoin such as Max Keiser, with some quixotic advocates anticipating a rapid escalation in bitcoin price to over $10,000.
Economic science, however, is concerned only with quantitative, precise observations and the use of mathematical models to predict market movements.
Tarnopolski, who has published a number of astrophysics papers concerning the predictive modeling of Fermi gamma ray bursts, uses the concept of geometric fractional Brownian motion in his recent paper to forecast the price of bitcoin.
Modeling Bitcoin Movements With Physics
Fractional Brownian motion is a mathematical model that is uniquely suited to the analysis of bitcoin price movements. The application of this model in the development of a predictive model for a volatile system such as the ever-shifting price of a cryptocurrency can be likened to the “How Long is the Coast of Britain?” problem first presented by mathematician Benoît Mandelbrot in 1967.
The core precept in Mandelbrot’s argument, as well as the mathematical axiom used by Tarnopolski, is that geographical curves — or in this case, the prediction of bitcoin price movements — are so complex that they are undefinable. Fractional Brownian motion holds that these systems are statistically “selfsimilar”, which means that small isolated portions of data such as historical bitcoin price movements can be considered as reduced-scale images of the whole.
Tarnopolski’s paper uses a large temporal data set, incorporating bitcoin price movement from December 2011 until June 2017. Using this data, Tarnapolski generated a Monte Carlo simulation that integrated 10,000 individual geometric fractional Brownian motion realizations to predict an early-2018 price of $6,358 — within 10 percent accuracy.
The paper in itself could benefit from the incorporating a number of other factors, such as the inclusion of bullish and bearish parameters to justify large swings, but is relatively accurate given the limited amount of market data available. Combined with the positive technical and fundamental signals the crypto ecosystem is exhibiting at the moment, bitcoin certainly shows signs of a strong start in 2018.
Can physics really predict bitcoin prices? Please share your thoughts in the comments below.
Images via R Spatial, Pixabay, ResearchGate.net