johnnystackone

GPU Cost Optimization

Why We are Using GPUs

For the trader project our use of GPUs is motivated by training machine learning models and neural networks in particular in an effort to develop signal about future price action in crypto markets, initially UNFI. Having good estimates about future price movements will be highly useful in almost any form of automated trading that we do, but UNFI in particular seems to have it's price action often be dominated by reasonably advanced actors who we have reason to believe already possess such signal. Trading against actors with such signal when we lack it would most likely lead to losing money.

It's probably worth noting that there is some projected work on other projects, notably the news recommender product that will also require GPU resources at some point, and in general more and more projects are requiring some amount of scientific compute resources.

Current GPU Requirements

johnnystackone

GPU Cost Optimization

Why We are Using GPUs

For the trader project our use of GPUs is motivated by training machine learning models and neural networks in particular in an effort to develop signal about future price action in crypto markets, initially UNFI. Having good estimates about future price movements will be highly useful in almost any form of automated trading that we do, but UNFI in particular seems to have it's price action often be dominated by reasonably advanced actors who we have reason to believe already possess such signal. Trading against actors with such signal when we lack it would most likely lead to losing money.

It's probably worth noting that there is some projected work on other projects, notably the news recommender product that will also require GPU resources at some point, and in general more and more projects are requiring some amount of scientific compute resources.

Current GPU Requirements

johnnystackone

GPU Cost Optimization

Why We are Using GPUs

For the trader project our use of GPUs is motivated by training machine learning models and neural networks in particular in an effort to develop signal about future price action in crypto markets, initially UNFI. Having good estimates about future price movements will be highly useful in almost any form of automated trading that we do, but UNFI in particular seems to have it's price action often be dominated by reasonably advanced actors who we have reason to believe already possess such signal. Trading against actors with such signal when we lack it would most likely lead to losing money.

It's probably worth noting that there is some projected work on other projects, notably the news recommender product that will also require GPU resources at some point, and in general more and more projects are requiring some amount of scientific compute resources.

Current GPU Requirements

johnnystackone

GPU Cost Optimization

Why We are Using GPUs

For the trader project our use of GPUs is motivated by training machine learning models and neural networks in particular in an effort to develop signal about future price action in crypto markets, initially UNFI. Having good estimates about future price movements will be highly useful in almost any form of automated trading that we do, but UNFI in particular seems to have it's price action often be dominated by reasonably advanced actors who we have reason to believe already possess such signal. Trading against actors with such signal when we lack it would most likely lead to losing money.

It's probably worth noting that there is some projected work on other projects, notably the news recommender product that will also require GPU resources at some point, and in general more and more projects are requiring some amount of scientific compute resources.

Current GPU Requirements

johnnystackone

GPU Cost Optimization

Why We are Using GPUs

For the trader project our use of GPUs is motivated by training machine learning models an neural networks in particular in an effort to develop signal about future price action in crypto markets, initially UNFI. Having good estimates about future price movements will be highly useful in almost any form of automated trading that we do, but UNFI in particular seems to have it's price action often be dominated by reasonably advanced actors who we have reason to believe already possess such signal. Trading against actors with such signal when we lack it would most likely lead to losing money.

It's probably worth noting that there is some projected work on other projects, notably the news recommender product that will also require GPU resources at some point, and in general more and more projects are requiring some amount of scientific compute resources.

Current GPU Requirements

johnnystackone

CΞLΞRIS High-Level Update and Plan Mid-May

Overall Project Status Update

High-Level View of Progress So Far

It's been awhile since we did a high-level overview of project progress and we'd like to bring everyone up to speed on what's been accomplished so far (and in the section below future directions).

johnnystackone

CΞLΞRIS High-Level Update and Plan Mid-May

Overall Project Status Update

High-Level View of Progress So Far

It's been awhile since we did a high-level overview of project progress and we'd like to bring everyone up to speed on what's been accomplished so far (and in the section below future directions).

johnnystackone

CΞLΞRIS High-Level Update and Plan Mid-May

Overall Project Status Update

High-Level View of Progress So Far

It's been awhile since we did a high-level overview of project progress and we'd like to bring everyone up to speed on what's been accomplished so far (and in the section below future directions).

johnnystackone

CΞLΞRIS March Schedule

2021-03-28

Status Summary

johnnystackone

CΞLΞRIS March Schedule

2021-03-28

Status Summary