Refactor triangle-mode

Minor Change `assert_supported` exception type from `ValueError` to `AssertionError` Remove unused return-value Remove unused code Refactor single-trianglre mode as well Remove unneeded assertion Cosmetic Fix single-triangle mode Fix single-triangle mode Refactor the b3-two-hop-triangle mode as well Remove the importing of `T` Remove unused code Remove single-triangle mode, and fix a bunch of problems Cleanup `calculate_profit` related code Fix test 042 Fix test 042 Remove unused code Some more general cleanup Cleanup Remove unused code Fix test 063 Cleanup Cleanup Fix tests Cleanup Cleanup Fix test 042 Remove single_paiwise and multi_paiwise arb modes (major change) Fix test 045 Fix test 045 Refactor the pairwise arb-mode classes Ongoing refactor Refactor triangle arb-mode classes Cleanup Remove file `scan_log_errors.py` from the repository Cleanup Remove unused code Code reuse Fix functions `create_sort_order` and `sort_key`
bancorprotocol · May 17, 2024 · 1ef62bb · 1ef62bb
1 parent adeece8
commit 1ef62bb
Show file tree

Hide file tree

Showing 33 changed files with 537 additions and 2,307 deletions.
diff --git a/README.md b/README.md
@@ -107,13 +107,10 @@ You can configure the Fastlane Arbitrage Bot using the options in the `@click.op
     - **Triangular**: This includes arbitrage trades between three liquidity pools that can create a triangular route, starting and ending in the same token. For example, USDC > ETH, ETH > LINK, LINK > USDC
     - **Multi**: These modes can trade through multiple Carbon orders as a single trade.
   - **arb_mode options**:
-      - **single**: Pairwise arbitrage between one Carbon curve and one other exchange curve.
-      - **multi** Pairwise arbitrage between **multiple** Carbon curves and one other exchange curve.
-      - **triangle**: Triangular arbitrage between one Carbon curve and two other exchange curves.
-      - **multi_triangle**: Triangular arbitrage between **multiple** Carbon curves and two other exchange curves.
+      - **multi_triangle**: Triangular arbitrage between multiple Carbon curves and two other exchange curves.
+      - **multi_triangle_complete**: Triangular arbitrage between multiple Carbon curves and two other exchange curves (experimental).
       - **b3_two_hop**: Triangular arbitrage - the same as bancor_v3 mode but more gas-efficient.
       - **multi_pairwise_pol**: Pairwise multi-mode that always routes through the Bancor protocol-owned liquidity contract.
-      - **multi_pairwise_bal**: Pairwise multi-mode that always routes through Balancer.
       - **multi_pairwise_all**: **(Default)** Pairwise multi-mode that searches all available exchanges for pairwise arbitrage.
 - **flashloan_tokens** (str): Tokens the bot can use for flash loans. Specify token addresses as a comma-separated string (e.g., 0x1F573D6Fb3F13d689FF844B4cE37794d79a7FF1C, 0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2).
 - **n_jobs** (int): The number of parallel jobs to run. The default, -1, will use all available cores for the process.
@@ -145,7 +142,7 @@ You can configure the Fastlane Arbitrage Bot using the options in the `@click.op
 Specify options in the command line. For example:
 
 ```bash
-poetry run python main.py --arb_mode=multi --polling_interval=12 --reorg_delay=10 --loglevel=INFO
+poetry run python main.py --arb_mode=multi_pairwise_all --polling_interval=12 --reorg_delay=10 --loglevel=INFO
 ```
 
 ## Troubleshooting
@@ -176,7 +173,7 @@ poetry run python main.py --arb_mode=b3_two_hop --alchemy_max_block_fetch=200 --
 
 #### Carbon-focused pairwise arbitrage
 ```commandline
-poetry run python main.py --arb_mode=multi --alchemy_max_block_fetch=200 --loglevel=INFO --backdate_pools=False --polling_interval=0 --reorg_delay=0 --run_data_validator=False --default_min_profit_gas_token=0.01 --randomizer=2 --exchanges=bancor_v3,bancor_v2,carbon_v1,uniswap_v3,uniswap_v2,sushiswap_v2,balancer,pancakeswap_v2,pancakeswap_v3 --flashloan_tokens="0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE,0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2"
+poetry run python main.py --arb_mode=multi_pairwise_all --alchemy_max_block_fetch=200 --loglevel=INFO --backdate_pools=False --polling_interval=0 --reorg_delay=0 --run_data_validator=False --default_min_profit_gas_token=0.01 --randomizer=2 --exchanges=bancor_v3,bancor_v2,carbon_v1,uniswap_v3,uniswap_v2,sushiswap_v2,balancer,pancakeswap_v2,pancakeswap_v3 --flashloan_tokens="0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE,0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2"
 ```
 #### Unfocused pairwise arbitrage
 ```commandline

diff --git a/fastlane_bot/bot.py b/fastlane_bot/bot.py
@@ -66,16 +66,13 @@
     split_carbon_trades,
     maximize_last_trade_per_tkn
 )
-from fastlane_bot.tools.cpc import ConstantProductCurve as CPC, CPCContainer, T
+from fastlane_bot.tools.cpc import ConstantProductCurve as CPC, CPCContainer
 from .config.constants import FLASHLOAN_FEE_MAP
 from .events.interface import QueryInterface
-from .modes.pairwise_multi import FindArbitrageMultiPairwise
 from .modes.pairwise_multi_all import FindArbitrageMultiPairwiseAll
 from .modes.pairwise_multi_pol import FindArbitrageMultiPairwisePol
-from .modes.pairwise_single import FindArbitrageSinglePairwise
 from .modes.triangle_multi import ArbitrageFinderTriangleMulti
 from .modes.triangle_multi_complete import ArbitrageFinderTriangleMultiComplete
-from .modes.triangle_single import ArbitrageFinderTriangleSingle
 from .modes.triangle_bancor_v3_two_hop import ArbitrageFinderTriangleBancor3TwoHop
 from .utils import num_format
 
@@ -103,9 +100,6 @@ class CarbonBot:
     SCALING_FACTOR = 0.999
 
     ARB_FINDER = {
-        "single": FindArbitrageSinglePairwise,
-        "multi": FindArbitrageMultiPairwise,
-        "triangle": ArbitrageFinderTriangleSingle,
         "multi_triangle": ArbitrageFinderTriangleMulti,
         "b3_two_hop": ArbitrageFinderTriangleBancor3TwoHop,
         "multi_pairwise_pol": FindArbitrageMultiPairwisePol,
@@ -554,21 +548,14 @@ def get_prices_simple(self, CCm, tkn0, tkn1):
         curve_prices += [(x.params['exchange'],x.descr,x.cid,1/x.p) for x in CCm.bytknx(tkn1).bytkny(tkn0)]
         return curve_prices
 
-    # Global constant for Carbon Forks ordering
-    CARBON_SORTING_ORDER = float('inf')
-
     # Create a sort order mapping function
     def create_sort_order(self, sort_sequence):
         # Create a dictionary mapping from sort sequence to indices, except for Carbon Forks
         return {key: index for index, key in enumerate(sort_sequence) if key not in self.ConfigObj.CARBON_V1_FORKS}
 
     # Define the sort key function separately
     def sort_key(self, item, sort_order):
-        # Check if the item is Carbon Forks
-        if item[0] in self.ConfigObj.CARBON_V1_FORKS:
-            return self.CARBON_SORTING_ORDER
-        # Otherwise, use the sort order from the dictionary, or a default high value
-        return sort_order.get(item[0], self.CARBON_SORTING_ORDER - 1)
+        return float('inf') if item[0] in self.ConfigObj.CARBON_V1_FORKS else sort_order.get(item[0], float('inf'))
 
     # Define the custom sort function
     def custom_sort(self, data, sort_sequence):

diff --git a/fastlane_bot/modes/__init__.py b/fastlane_bot/modes/__init__.py
@@ -24,13 +24,11 @@
 
 - ``ArbitrageFinderBase`` (``base``): fundamental base class
     - ``ArbitrageFinderPairwiseBase`` (``base_pairwise``): base class for pairwise arbitrages
-        - ``FindArbitrageSinglePairwise`` (``pairwise_single``)
-        - ``FindArbitrageMultiPairwise`` (``pairwise_multi``)
         - ``FindArbitrageMultiPairwiseAll`` (``pairwise_multi_all``)
         - ``FindArbitrageMultiPairwisePol`` (``pairwise_multi_pol``)
     - ``ArbitrageFinderTriangleBase`` (``base_triangle``): base class for triangle arbitrages    
-        - ``ArbitrageFinderTriangleSingle`` (``triangle_single``)
         - ``ArbitrageFinderTriangleMulti`` (``triangle_multi``)
+        - ``ArbitrageFinderTriangleMultiComplete`` (``triangle_multi_complete``)
         - ``ArbitrageFinderTriangleBancor3TwoHop`` (``triangle_bancor_v3_two_hop``)
 
 

diff --git a/fastlane_bot/modes/base.py b/fastlane_bot/modes/base.py
@@ -13,10 +13,6 @@
 from _decimal import Decimal
 import pandas as pd
 
-from fastlane_bot.tools.cpc import T
-from fastlane_bot.utils import num_format
-
-
 class ArbitrageFinderBase:
     """
     Base class for all arbitrage finder modes
@@ -47,114 +43,77 @@ def __init__(
         self.base_exchange = "bancor_v3" if arb_mode == "bancor_v3" else "carbon_v1"
 
     @abc.abstractmethod
-    def find_arbitrage(
-        self,
-        candidates: List[Any] = None,
-        ops: Tuple = None,
-        best_profit: float = 0,
-        profit_src: float = 0,
-    ) -> Union[List, Tuple]:
+    def find_arbitrage() -> Union[List, Tuple]:
         """
         See subclasses for details
 
-        Parameters
-        ----------
-        candidates : List[Any], optional
-            List of candidates, by default None
-        ops : Tuple, optional
-            Tuple of operations, by default None
-        best_profit : float, optional
-            Best profit so far, by default 0
-        profit_src : float, optional
-            Profit source, by default 0
-
         Returns
         -------
         Union[List, Tuple]
             If self.result == self.AO_CANDIDATES, it returns a list of candidates.
         """
         pass
 
-    def _set_best_ops(
-        self,
-        best_profit: float,
-        ops: Tuple,
-        profit: float,
-        src_token: str,
-        trade_instructions: Any,
-        trade_instructions_df: pd.DataFrame,
-        trade_instructions_dic: Dict[str, Any],
-    ) -> Tuple[float, Tuple]:
-        """
-        Set the best operations.
-
-        Parameters:
-
-        """
-        self.ConfigObj.logger.debug("[modes.base._set_best_ops] *************")
-        self.ConfigObj.logger.debug(
-            f"[modes.base._set_best_ops] New best profit: {profit}"
-        )
-
-        # Update the best profit and source token
-        best_profit = profit
-        best_src_token = src_token
-
-        # Update the best trade instructions
-        best_trade_instructions_df = trade_instructions_df
-        best_trade_instructions_dic = trade_instructions_dic
-        best_trade_instructions = trade_instructions
-
-        self.ConfigObj.logger.debug(
-            f"[modes.base._set_best_ops] best_trade_instructions_df: {best_trade_instructions_df}"
-        )
-
-        # Update the optimal operations
-        ops = (
-            best_profit,
-            best_trade_instructions_df,
-            best_trade_instructions_dic,
-            best_src_token,
-            best_trade_instructions,
-        )
-
-        self.ConfigObj.logger.debug("[modes.base.calculate_profit] *************")
-
-        return best_profit, ops
-
     def get_prices_simple(self, CCm, tkn0, tkn1):
         curve_prices = [(x.params['exchange'],x.descr,x.cid,x.p) for x in CCm.bytknx(tkn0).bytkny(tkn1)]
         curve_prices += [(x.params['exchange'],x.descr,x.cid,1/x.p) for x in CCm.bytknx(tkn1).bytkny(tkn0)]
         return curve_prices
 
-    # Global constant for 'carbon_v1' order
-    CARBON_SORTING_ORDER = float('inf')
-
     # Create a sort order mapping function
     def create_sort_order(self, sort_sequence):
-        # Create a dictionary mapping from sort sequence to indices, except for 'carbon_v1'
-        return {key: index for index, key in enumerate(sort_sequence) if key != 'carbon_v1'}
+        # Create a dictionary mapping from sort sequence to indices, except for Carbon Forks
+        return {key: index for index, key in enumerate(sort_sequence) if key not in self.ConfigObj.CARBON_V1_FORKS}
 
     # Define the sort key function separately
     def sort_key(self, item, sort_order):
-        # Check if the item is 'carbon_v1'
-        if item[0] in self.ConfigObj.CARBON_V1_FORKS:
-            return self.CARBON_SORTING_ORDER
-        # Otherwise, use the sort order from the dictionary, or a default high value
-        return sort_order.get(item[0], self.CARBON_SORTING_ORDER - 1)
+        return float('inf') if item[0] in self.ConfigObj.CARBON_V1_FORKS else sort_order.get(item[0], float('inf'))
 
     # Define the custom sort function
     def custom_sort(self, data, sort_sequence):
         sort_order = self.create_sort_order(sort_sequence)
         return sorted(data, key=lambda item: self.sort_key(item, sort_order))
 
+    def update_results(
+        self,
+        src_token: str,
+        r,
+        trade_instructions_dic,
+        trade_instructions_df,
+        trade_instructions,
+        candidates,
+        best_profit,
+        ops,
+    ):
+        # Calculate the profit
+        profit = self.calculate_profit(src_token, -r.result, self.CCm)
+        if str(profit) == "nan":
+            self.ConfigObj.logger.debug("profit is nan, skipping")
+        else:
+            # Handle candidates based on conditions
+            candidates += self.handle_candidates(
+                profit,
+                trade_instructions_df,
+                trade_instructions_dic,
+                src_token,
+                trade_instructions,
+            )
+            # Find the best operations
+            best_profit, ops = self.find_best_operations(
+                best_profit,
+                ops,
+                profit,
+                trade_instructions_df,
+                trade_instructions_dic,
+                src_token,
+                trade_instructions,
+            )
+        return best_profit, ops
+
     def calculate_profit(
         self,
         src_token: str,
         profit_src: float,
         CCm: Any,
-        cids: List[str],
-        profit: int = 0,
     ) -> float:
         """
         Calculate profit based on the source token.
@@ -201,7 +160,6 @@ def get_netchange(trade_instructions_df: pd.DataFrame) -> List[float]:
 
     def handle_candidates(
         self,
-        best_profit: float,
         profit: float,
         trade_instructions_df: pd.DataFrame,
         trade_instructions_dic: Dict[str, Any],
@@ -213,8 +171,6 @@ def handle_candidates(
 
         Parameters:
         ----------
-        best_profit : float
-            Best profit
         profit : float
             Profit
         trade_instructions_df : pd.DataFrame
@@ -290,29 +246,12 @@ def find_best_operations(
         condition_better_profit = profit > best_profit
         condition_zeros_one_token = max(netchange) < 1e-4
         if condition_better_profit and condition_zeros_one_token:
-            return self._set_best_ops(
-                best_profit,
-                ops,
+            best_profit = profit
+            ops = (
                 profit,
-                src_token,
-                trade_instructions,
                 trade_instructions_df,
                 trade_instructions_dic,
+                src_token,
+                trade_instructions,
             )
         return best_profit, ops
-
-    def _check_limit_flashloan_tokens_for_bancor3(self):
-        """
-        Limit the flashloan tokens for bancor v3.
-        """
-        fltkns = self.CCm.byparams(exchange="bancor_v3").tknys()
-        if self.ConfigObj.LIMIT_BANCOR3_FLASHLOAN_TOKENS:
-            # Filter out tokens that are not in the existing flashloan_tokens list
-            self.flashloan_tokens = [
-                tkn for tkn in fltkns if tkn in self.flashloan_tokens
-            ]
-            self.ConfigObj.logger.info(
-                f"[modes.base._check_limit_flashloan_tokens_for_bancor3] limiting flashloan_tokens to {self.flashloan_tokens}"
-            )
-        else:
-            self.flashloan_tokens = fltkns