A few quick and dirty calculations about bitcoin mining profitability; a friend was playing around with bitcoin mining and asked me about this, I was curious so I downloaded some data and played around with it. WARNING I'm lazy so I committed a large bitcoin data dump, fyi if you want to clone this.
Being a sunday quant was fun but it might get boring, I guess their 7-figure salaries help.
I got a distinctly skeevy, gold rush vibe when looking for data, not a fan of the whole bitcoin ecosystem. I'd avoid this space entirely; as the saying goes, if you're at a poker game and you don't know who the sucker is, get up and leave because it's you. Selling mining hardware & services looks like selling shovels during the gold rush to me.
That being said, there are 2 things I found mildly interesting:
Work in ./past_performance.ipynb
Mining pools reward miners using a variety of reward contracts, I examined
- FPPS, full pay-per-share, a 0-risk system where the pool pays for each work share according to the expected reward & transaction fees and takes a percentage cut
- PPLNS, pay-per-last-N-shares, a system where if a block is mined the reward & transaction fees are divided among the miners according to work contributed to the pool over a time window (larger than a block to penalize pool hopping).
There's a tradeoff between risk and reward here, as 0 % fee pplns pools exist. So I ran a few simulations with past data to understand the magnitude of that risk-reward tradeoff, as the risk depends on the pplns pool's global hashrate share:
Work in ./mining_feasibility.ipynb
Unsurprisingly, bitcoin mining is extremely sensitive to electricity pricing, and indifference curves can be plotted to understand the relationship between bitcoin price, hashrate and electricity prices:
Mining revenue is extremely volatile recently given rising energy costs in europe:
On linux or mac:
- Create a vitualenv:
python3 -m venv env - install deps:
env/bin/pip install -r requirements.txt - launch notebook:
env/bin/jupyter notebook