DEV Community

Quantooox profile picture

Quantooox

404 bio not found

Joined Joined on  Personal website https://quantooo.com
We scored every headline with FinBERT. News tone didn't predict the next month.

We scored every headline with FinBERT. News tone didn't predict the next month.

Comments 1
2 min read
Where machine learning did help: forecasting market volatility

Where machine learning did help: forecasting market volatility

Comments
2 min read
Information coefficient and the t-statistic: is 0.02 better than 0.01?

Information coefficient and the t-statistic: is 0.02 better than 0.01?

Comments
2 min read
Before costs vs after costs: why we double the trading cost in every test

Before costs vs after costs: why we double the trading cost in every test

Comments
2 min read
Survivorship bias: testing on today's winners flatters every strategy

Survivorship bias: testing on today's winners flatters every strategy

Comments
2 min read
Overfitting in practice: our rule that simple wins ties

Overfitting in practice: our rule that simple wins ties

Comments
2 min read
Walk-forward testing with an embargo: keeping time in order

Walk-forward testing with an embargo: keeping time in order

Comments
2 min read
Out of how many? Multiple testing and the deflated Sharpe ratio

Out of how many? Multiple testing and the deflated Sharpe ratio

Comments
2 min read
Random referees: two controls every ranking model should beat

Random referees: two controls every ranking model should beat

Comments
2 min read
The sealed holdout: why we looked at our test data exactly once

The sealed holdout: why we looked at our test data exactly once

Comments
2 min read
Look-ahead bias: the backtest bug that uses tomorrow's newspaper

Look-ahead bias: the backtest bug that uses tomorrow's newspaper

Comments
2 min read
Our stock score ranked stocks backwards on the one test that mattered

Our stock score ranked stocks backwards on the one test that mattered

Comments
2 min read
We tested 12 machine-learning models on the stock market. None of them worked, and we published everything.

We tested 12 machine-learning models on the stock market. None of them worked, and we published everything.

Comments
3 min read
loading...