CTYPTO/freqtrade
1
0
Fork 0
mirror of https://github.com/freqtrade/freqtrade.git synced 2026-02-10 08:10:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'upstream/develop' into feature/fetch-public-trades

Browse Source
This commit is contained in:
Joe Schr
2024-03-11 12:29:00 +01:00
parent 6827e17d17 ca4fb98775
commit c12e203054
32 changed files with 729 additions and 83 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
docs/advanced-backtesting.md
View File

@@ -109,12 +109,12 @@ automatically accessible by including them on the indicator-list, and these incl
- **open_date :** trade open datetime
- **close_date :** trade close datetime
- **min_rate :** minimum price seen throughout the position
- **max_rate :** maxiumum price seen throughout the position
- **max_rate :** maximum price seen throughout the position
- **open :** signal candle open price
- **close :** signal candle close price
- **high :** signal candle high price
- **low :** signal candle low price
- **volume :** signal candle volumne
- **volume :** signal candle volume
- **profit_ratio :** trade profit ratio
- **profit_abs :** absolute profit return of the trade

2
docs/freqai-parameter-table.md
View File

@@ -75,7 +75,7 @@ Mandatory parameters are marked as **Required** and have to be set in one of the
| `rl_config` | A dictionary containing the control parameters for a Reinforcement Learning model. <br> **Datatype:** Dictionary.
| `train_cycles` | Training time steps will be set based on the `train_cycles * number of training data points. <br> **Datatype:** Integer.
| `max_trade_duration_candles`| Guides the agent training to keep trades below desired length. Example usage shown in `prediction_models/ReinforcementLearner.py` within the customizable `calculate_reward()` function. <br> **Datatype:** int.
| `model_type` | Model string from stable_baselines3 or SBcontrib. Available strings include: `'TRPO', 'ARS', 'RecurrentPPO', 'MaskablePPO', 'PPO', 'A2C', 'DQN'`. User should ensure that `model_training_parameters` match those available to the corresponding stable_baselines3 model by visiting their documentaiton. [PPO doc](https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html) (external website) <br> **Datatype:** string.
| `model_type` | Model string from stable_baselines3 or SBcontrib. Available strings include: `'TRPO', 'ARS', 'RecurrentPPO', 'MaskablePPO', 'PPO', 'A2C', 'DQN'`. User should ensure that `model_training_parameters` match those available to the corresponding stable_baselines3 model by visiting their documentation. [PPO doc](https://stable-baselines3.readthedocs.io/en/master/modules/ppo.html) (external website) <br> **Datatype:** string.
| `policy_type` | One of the available policy types from stable_baselines3 <br> **Datatype:** string.
| `max_training_drawdown_pct` | The maximum drawdown that the agent is allowed to experience during training. <br> **Datatype:** float. <br> Default: 0.8
| `cpu_count` | Number of threads/cpus to dedicate to the Reinforcement Learning training process (depending on if `ReinforcementLearning_multiproc` is selected or not). Recommended to leave this untouched, by default, this value is set to the total number of physical cores minus 1. <br> **Datatype:** int.

2
docs/freqai-reinforcement-learning.md
View File

@@ -142,7 +142,7 @@ Parameter details can be found [here](freqai-parameter-table.md), but in general
As you begin to modify the strategy and the prediction model, you will quickly realize some important differences between the Reinforcement Learner and the Regressors/Classifiers. Firstly, the strategy does not set a target value (no labels!). Instead, you set the `calculate_reward()` function inside the `MyRLEnv` class (see below). A default `calculate_reward()` is provided inside `prediction_models/ReinforcementLearner.py` to demonstrate the necessary building blocks for creating rewards, but this is *not* designed for production. Users *must* create their own custom reinforcement learning model class or use a pre-built one from outside the Freqtrade source code and save it to `user_data/freqaimodels`. It is inside the `calculate_reward()` where creative theories about the market can be expressed. For example, you can reward your agent when it makes a winning trade, and penalize the agent when it makes a losing trade. Or perhaps, you wish to reward the agent for entering trades, and penalize the agent for sitting in trades too long. Below we show examples of how these rewards are all calculated:
!!! note "Hint"
The best reward functions are ones that are continuously differentiable, and well scaled. In other words, adding a single large negative penalty to a rare event is not a good idea, and the neural net will not be able to learn that function. Instead, it is better to add a small negative penalty to a common event. This will help the agent learn faster. Not only this, but you can help improve the continuity of your rewards/penalties by having them scale with severity according to some linear/exponential functions. In other words, you'd slowly scale the penalty as the duration of the trade increases. This is better than a single large penalty occuring at a single point in time.
The best reward functions are ones that are continuously differentiable, and well scaled. In other words, adding a single large negative penalty to a rare event is not a good idea, and the neural net will not be able to learn that function. Instead, it is better to add a small negative penalty to a common event. This will help the agent learn faster. Not only this, but you can help improve the continuity of your rewards/penalties by having them scale with severity according to some linear/exponential functions. In other words, you'd slowly scale the penalty as the duration of the trade increases. This is better than a single large penalty occurring at a single point in time.
```python
from freqtrade.freqai.prediction_models.ReinforcementLearner import ReinforcementLearner

4
docs/requirements-docs.txt
View File

@@ -1,6 +1,6 @@
markdown==3.5.2
mkdocs==1.5.3
mkdocs-material==9.5.12
mkdocs-material==9.5.13
mdx_truly_sane_lists==1.3
pymdown-extensions==10.7
pymdown-extensions==10.7.1
jinja2==3.1.3

129
docs/strategy-advanced.md
View File

@@ -11,34 +11,129 @@ The call sequence of the methods described here is covered under [bot execution
!!! Tip
Start off with a strategy template containing all available callback methods by running `freqtrade new-strategy --strategy MyAwesomeStrategy --template advanced`
## Storing information
## Storing information (Persistent)
Storing information can be accomplished by creating a new dictionary within the strategy class.
Freqtrade allows storing/retrieving user custom information associated with a specific trade in the database.
The name of the variable can be chosen at will, but should be prefixed with `custom_` to avoid naming collisions with predefined strategy variables.
Using a trade object, information can be stored using `trade.set_custom_data(key='my_key', value=my_value)` and retrieved using `trade.get_custom_data(key='my_key')`. Each data entry is associated with a trade and a user supplied key (of type `string`). This means that this can only be used in callbacks that also provide a trade object.
For the data to be able to be stored within the database, freqtrade must serialized the data. This is done by converting the data to a JSON formatted string.
Freqtrade will attempt to reverse this action on retrieval, so from a strategy perspective, this should not be relevant.
```python
from freqtrade.persistence import Trade
from datetime import timedelta
class AwesomeStrategy(IStrategy):
# Create custom dictionary
custom_info = {}
def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
# Check if the entry already exists
if not metadata["pair"] in self.custom_info:
# Create empty entry for this pair
self.custom_info[metadata["pair"]] = {}
def bot_loop_start(self, **kwargs) -> None:
for trade in Trade.get_open_order_trades():
fills = trade.select_filled_orders(trade.entry_side)
if trade.pair == 'ETH/USDT':
trade_entry_type = trade.get_custom_data(key='entry_type')
if trade_entry_type is None:
trade_entry_type = 'breakout' if 'entry_1' in trade.enter_tag else 'dip'
elif fills > 1:
trade_entry_type = 'buy_up'
trade.set_custom_data(key='entry_type', value=trade_entry_type)
return super().bot_loop_start(**kwargs)
if "crosstime" in self.custom_info[metadata["pair"]]:
self.custom_info[metadata["pair"]]["crosstime"] += 1
else:
self.custom_info[metadata["pair"]]["crosstime"] = 1
def adjust_entry_price(self, trade: Trade, order: Optional[Order], pair: str,
current_time: datetime, proposed_rate: float, current_order_rate: float,
entry_tag: Optional[str], side: str, **kwargs) -> float:
# Limit orders to use and follow SMA200 as price target for the first 10 minutes since entry trigger for BTC/USDT pair.
if (
pair == 'BTC/USDT'
and entry_tag == 'long_sma200'
and side == 'long'
and (current_time - timedelta(minutes=10)) > trade.open_date_utc
and order.filled == 0.0
):
dataframe, _ = self.dp.get_analyzed_dataframe(pair=pair, timeframe=self.timeframe)
current_candle = dataframe.iloc[-1].squeeze()
# store information about entry adjustment
existing_count = trade.get_custom_data('num_entry_adjustments', default=0)
if not existing_count:
existing_count = 1
else:
existing_count += 1
trade.set_custom_data(key='num_entry_adjustments', value=existing_count)
# adjust order price
return current_candle['sma_200']
# default: maintain existing order
return current_order_rate
def custom_exit(self, pair: str, trade: Trade, current_time: datetime, current_rate: float, current_profit: float, **kwargs):
entry_adjustment_count = trade.get_custom_data(key='num_entry_adjustments')
trade_entry_type = trade.get_custom_data(key='entry_type')
if entry_adjustment_count is None:
if current_profit > 0.01 and (current_time - timedelta(minutes=100) > trade.open_date_utc):
return True, 'exit_1'
else
if entry_adjustment_count > 0 and if current_profit > 0.05:
return True, 'exit_2'
if trade_entry_type == 'breakout' and current_profit > 0.1:
return True, 'exit_3
return False, None
```
!!! Warning
The data is not persisted after a bot-restart (or config-reload). Also, the amount of data should be kept smallish (no DataFrames and such), otherwise the bot will start to consume a lot of memory and eventually run out of memory and crash.
The above is a simple example - there are simpler ways to retrieve trade data like entry-adjustments.
!!! Note
If the data is pair-specific, make sure to use pair as one of the keys in the dictionary.
It is recommended that simple data types are used `[bool, int, float, str]` to ensure no issues when serializing the data that needs to be stored.
Storing big junks of data may lead to unintended side-effects, like a database becoming big (and as a consequence, also slow).
!!! Warning "Non-serializable data"
If supplied data cannot be serialized a warning is logged and the entry for the specified `key` will contain `None` as data.
??? Note "All attributes"
custom-data has the following accessors through the Trade object (assumed as `trade` below):
* `trade.get_custom_data(key='something', default=0)` - Returns the actual value given in the type provided.
* `trade.get_custom_data_entry(key='something')` - Returns the entry - including metadata. The value is accessible via `.value` property.
* `trade.set_custom_data(key='something', value={'some': 'value'})` - set or update the corresponding key for this trade. Value must be serializable - and we recommend to keep the stored data relatively small.
"value" can be any type (both in setting and receiving) - but must be json serializable.
## Storing information (Non-Persistent)
!!! Warning "Deprecated"
This method of storing information is deprecated and we do advise against using non-persistent storage.
Please use [Persistent Storage](#storing-information-persistent) instead.
It's content has therefore been collapsed.
??? Abstract "Storing information"
Storing information can be accomplished by creating a new dictionary within the strategy class.
The name of the variable can be chosen at will, but should be prefixed with `custom_` to avoid naming collisions with predefined strategy variables.
```python
class AwesomeStrategy(IStrategy):
# Create custom dictionary
custom_info = {}
def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
# Check if the entry already exists
if not metadata["pair"] in self.custom_info:
# Create empty entry for this pair
self.custom_info[metadata["pair"]] = {}
if "crosstime" in self.custom_info[metadata["pair"]]:
self.custom_info[metadata["pair"]]["crosstime"] += 1
else:
self.custom_info[metadata["pair"]]["crosstime"] = 1
```
!!! Warning
The data is not persisted after a bot-restart (or config-reload). Also, the amount of data should be kept smallish (no DataFrames and such), otherwise the bot will start to consume a lot of memory and eventually run out of memory and crash.
!!! Note
If the data is pair-specific, make sure to use pair as one of the keys in the dictionary.
## Dataframe access

2
docs/strategy_analysis_example.md
View File

@@ -19,7 +19,7 @@ from pathlib import Path
project_root = "somedir/freqtrade"
i=0
try:
os.chdirdir(project_root)
os.chdir(project_root)
assert Path('LICENSE').is_file()
except:
while i<4 and (not Path('LICENSE').is_file()):

3
docs/telegram-usage.md
View File

@@ -59,7 +59,7 @@ For the Freqtrade configuration, you can then use the the full value (including
"chat_id": "-1001332619709"
```
!!! Warning "Using telegram groups"
When using telegram groups, you're giving every member of the telegram group access to your freqtrade bot and to all commands possible via telegram. Please make sure that you can trust everyone in the telegram group to avoid unpleasent surprises.
When using telegram groups, you're giving every member of the telegram group access to your freqtrade bot and to all commands possible via telegram. Please make sure that you can trust everyone in the telegram group to avoid unpleasant surprises.
## Control telegram noise
@@ -181,6 +181,7 @@ official commands. You can ask at any moment for help with `/help`.
| `/locks` | Show currently locked pairs.
| `/unlock <pair or lock_id>` | Remove the lock for this pair (or for this lock id).
| `/marketdir [long | short | even | none]` | Updates the user managed variable that represents the current market direction. If no direction is provided, the currently set direction will be displayed.
| `/list_custom_data <trade_id> [key]` | List custom_data for Trade ID & Key combination. If no Key is supplied it will list all key-value pairs found for that Trade ID.
| **Modify Trade states** |
| `/forceexit <trade_id> | /fx <tradeid>` | Instantly exits the given trade (Ignoring `minimum_roi`).
| `/forceexit all | /fx all` | Instantly exits all open trades (Ignoring `minimum_roi`).

2
docs/webhook-config.md
View File

@@ -65,7 +65,7 @@ You can set the POST body format to Form-Encoded (default), JSON-Encoded, or raw
The result would be a POST request with e.g. `{"text":"Status: running"}` body and `Content-Type: application/json` header which results `Status: running` message in the Mattermost channel.
When using the Form-Encoded or JSON-Encoded configuration you can configure any number of payload values, and both the key and value will be ouput in the POST request. However, when using the raw data format you can only configure one value and it **must** be named `"data"`. In this instance the data key will not be output in the POST request, only the value. For example:
When using the Form-Encoded or JSON-Encoded configuration you can configure any number of payload values, and both the key and value will be output in the POST request. However, when using the raw data format you can only configure one value and it **must** be named `"data"`. In this instance the data key will not be output in the POST request, only the value. For example:
```json
"webhook": {