Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -40,8 +40,6 @@ tags:
|
|
| 40 |
---
|
| 41 |
# BigStock
|
| 42 |
|
| 43 |
-
## ⚠️ This is a preview--there are some bugs in my scraping code, meaning this dataset does not capture all tickers. Check back later!
|
| 44 |
-
|
| 45 |
This dataset contains historical data stock data spanning a period of 720+ days, at one-hour intervals.
|
| 46 |
All values are in USD.
|
| 47 |
It includes full candlestick data (open, close, low, high) as well as volume data.
|
|
@@ -50,8 +48,7 @@ The data (not scraping/processing code) is licensed under GPL v3.
|
|
| 50 |
## Other Datasets
|
| 51 |
|
| 52 |
A [sample dataset](https://huggingface.co/datasets/hikitoxin/bigstock-1h-sample) (<20MB) for a subset of popular stocks is available as well.
|
| 53 |
-
|
| 54 |
-
I may also release a multivariate forecasting model trained on the data, likely based on Tiny Time Mixers or a custom architecture I have been evaluating.
|
| 55 |
|
| 56 |
## Uses
|
| 57 |
|
|
@@ -63,7 +60,7 @@ The data is usable for the following cluster of tasks:
|
|
| 63 |
## Limitations
|
| 64 |
|
| 65 |
While its breadth is very large, the data is comparatively very short-term (~2yrs).
|
| 66 |
-
Controlling for biases in economic data is inherently a difficult problem, because we only have one market to sample from
|
| 67 |
Therefore, models trained on this data may be biased towards recent market trends.
|
| 68 |
|
| 69 |
This dataset may have missing, incorrect, or unreliable data.
|
|
@@ -72,7 +69,6 @@ You are encouraged to validate it before use.
|
|
| 72 |
## Dataset Creation
|
| 73 |
|
| 74 |
The source data in this sample was provided purely from the Yahoo Finance website.
|
| 75 |
-
The full dataset may have other sources.
|
| 76 |
Data was collected using the [yfinance](https://github.com/ranaroussi/yfinance/) library.
|
| 77 |
|
| 78 |
The collection and processing was done on hardware owned and operated entirely by me.
|
|
|
|
| 40 |
---
|
| 41 |
# BigStock
|
| 42 |
|
|
|
|
|
|
|
| 43 |
This dataset contains historical data stock data spanning a period of 720+ days, at one-hour intervals.
|
| 44 |
All values are in USD.
|
| 45 |
It includes full candlestick data (open, close, low, high) as well as volume data.
|
|
|
|
| 48 |
## Other Datasets
|
| 49 |
|
| 50 |
A [sample dataset](https://huggingface.co/datasets/hikitoxin/bigstock-1h-sample) (<20MB) for a subset of popular stocks is available as well.
|
| 51 |
+
I may also release a multivariate forecasting model trained on the data, likely based on Tiny Time Mixers or a novel architecture I have been evaluating.
|
|
|
|
| 52 |
|
| 53 |
## Uses
|
| 54 |
|
|
|
|
| 60 |
## Limitations
|
| 61 |
|
| 62 |
While its breadth is very large, the data is comparatively very short-term (~2yrs).
|
| 63 |
+
Controlling for biases in economic data is inherently a difficult problem, because we only have one market to sample from.
|
| 64 |
Therefore, models trained on this data may be biased towards recent market trends.
|
| 65 |
|
| 66 |
This dataset may have missing, incorrect, or unreliable data.
|
|
|
|
| 69 |
## Dataset Creation
|
| 70 |
|
| 71 |
The source data in this sample was provided purely from the Yahoo Finance website.
|
|
|
|
| 72 |
Data was collected using the [yfinance](https://github.com/ranaroussi/yfinance/) library.
|
| 73 |
|
| 74 |
The collection and processing was done on hardware owned and operated entirely by me.
|