Files
gogo2/COBY/aggregation/cross_exchange_aggregator.py
Dobromir Popov 8ee9b7a90c wip
2025-08-04 17:40:30 +03:00

390 lines
15 KiB
Python

"""
Cross-exchange data aggregation and consolidation.
"""
from typing import List, Dict, Optional
from collections import defaultdict
from datetime import datetime
from ..models.core import (
OrderBookSnapshot, ConsolidatedOrderBook, PriceLevel,
PriceBuckets, HeatmapData, HeatmapPoint
)
from ..utils.logging import get_logger
from ..utils.timing import get_current_timestamp
from .price_bucketer import PriceBucketer
from .heatmap_generator import HeatmapGenerator
logger = get_logger(__name__)
class CrossExchangeAggregator:
"""
Aggregates data across multiple exchanges.
Provides consolidated order books and cross-exchange heatmaps.
"""
def __init__(self):
"""Initialize cross-exchange aggregator"""
self.price_bucketer = PriceBucketer()
self.heatmap_generator = HeatmapGenerator()
# Exchange weights for aggregation
self.exchange_weights = {
'binance': 1.0,
'coinbase': 0.9,
'kraken': 0.8,
'bybit': 0.7,
'okx': 0.7,
'huobi': 0.6,
'kucoin': 0.6,
'gateio': 0.5,
'bitfinex': 0.5,
'mexc': 0.4
}
# Statistics
self.consolidations_performed = 0
self.exchanges_processed = set()
logger.info("Cross-exchange aggregator initialized")
def aggregate_across_exchanges(self, symbol: str,
orderbooks: List[OrderBookSnapshot]) -> ConsolidatedOrderBook:
"""
Aggregate order book data from multiple exchanges.
Args:
symbol: Trading symbol
orderbooks: List of order book snapshots from different exchanges
Returns:
ConsolidatedOrderBook: Consolidated order book data
"""
if not orderbooks:
raise ValueError("Cannot aggregate empty orderbook list")
try:
# Track exchanges
exchanges = [ob.exchange for ob in orderbooks]
self.exchanges_processed.update(exchanges)
# Calculate weighted mid price
weighted_mid_price = self._calculate_weighted_mid_price(orderbooks)
# Consolidate bids and asks
consolidated_bids = self._consolidate_price_levels(
[ob.bids for ob in orderbooks],
[ob.exchange for ob in orderbooks],
'bid'
)
consolidated_asks = self._consolidate_price_levels(
[ob.asks for ob in orderbooks],
[ob.exchange for ob in orderbooks],
'ask'
)
# Calculate total volumes
total_bid_volume = sum(level.size for level in consolidated_bids)
total_ask_volume = sum(level.size for level in consolidated_asks)
# Create consolidated order book
consolidated = ConsolidatedOrderBook(
symbol=symbol,
timestamp=get_current_timestamp(),
exchanges=exchanges,
bids=consolidated_bids,
asks=consolidated_asks,
weighted_mid_price=weighted_mid_price,
total_bid_volume=total_bid_volume,
total_ask_volume=total_ask_volume,
exchange_weights={ex: self.exchange_weights.get(ex, 0.5) for ex in exchanges}
)
self.consolidations_performed += 1
logger.debug(
f"Consolidated {len(orderbooks)} order books for {symbol}: "
f"{len(consolidated_bids)} bids, {len(consolidated_asks)} asks"
)
return consolidated
except Exception as e:
logger.error(f"Error aggregating across exchanges: {e}")
raise
def create_consolidated_heatmap(self, symbol: str,
orderbooks: List[OrderBookSnapshot]) -> HeatmapData:
"""
Create consolidated heatmap from multiple exchanges.
Args:
symbol: Trading symbol
orderbooks: List of order book snapshots
Returns:
HeatmapData: Consolidated heatmap data
"""
try:
# Create price buckets for each exchange
all_buckets = []
for orderbook in orderbooks:
buckets = self.price_bucketer.create_price_buckets(orderbook)
all_buckets.append(buckets)
# Aggregate all buckets
if len(all_buckets) == 1:
consolidated_buckets = all_buckets[0]
else:
consolidated_buckets = self.price_bucketer.aggregate_buckets(all_buckets)
# Generate heatmap from consolidated buckets
heatmap = self.heatmap_generator.generate_heatmap(consolidated_buckets)
# Add exchange metadata to heatmap points
self._add_exchange_metadata(heatmap, orderbooks)
logger.debug(f"Created consolidated heatmap for {symbol} from {len(orderbooks)} exchanges")
return heatmap
except Exception as e:
logger.error(f"Error creating consolidated heatmap: {e}")
raise
def _calculate_weighted_mid_price(self, orderbooks: List[OrderBookSnapshot]) -> float:
"""Calculate volume-weighted mid price across exchanges"""
total_weight = 0.0
weighted_sum = 0.0
for orderbook in orderbooks:
if orderbook.mid_price:
# Use total volume as weight
volume_weight = orderbook.bid_volume + orderbook.ask_volume
exchange_weight = self.exchange_weights.get(orderbook.exchange, 0.5)
# Combined weight
weight = volume_weight * exchange_weight
weighted_sum += orderbook.mid_price * weight
total_weight += weight
return weighted_sum / total_weight if total_weight > 0 else 0.0
def _consolidate_price_levels(self, level_lists: List[List[PriceLevel]],
exchanges: List[str], side: str) -> List[PriceLevel]:
"""Consolidate price levels from multiple exchanges"""
# Group levels by price bucket
price_groups = defaultdict(lambda: {'size': 0.0, 'count': 0, 'exchanges': set()})
for levels, exchange in zip(level_lists, exchanges):
exchange_weight = self.exchange_weights.get(exchange, 0.5)
for level in levels:
# Round price to bucket
bucket_price = self.price_bucketer.get_bucket_price(level.price)
# Add weighted volume
weighted_size = level.size * exchange_weight
price_groups[bucket_price]['size'] += weighted_size
price_groups[bucket_price]['count'] += level.count or 1
price_groups[bucket_price]['exchanges'].add(exchange)
# Create consolidated price levels
consolidated_levels = []
for price, data in price_groups.items():
if data['size'] > 0: # Only include non-zero volumes
level = PriceLevel(
price=price,
size=data['size'],
count=data['count']
)
consolidated_levels.append(level)
# Sort levels appropriately
if side == 'bid':
consolidated_levels.sort(key=lambda x: x.price, reverse=True)
else:
consolidated_levels.sort(key=lambda x: x.price)
return consolidated_levels
def _add_exchange_metadata(self, heatmap: HeatmapData,
orderbooks: List[OrderBookSnapshot]) -> None:
"""Add exchange metadata to heatmap points"""
# Create exchange mapping by price bucket
exchange_map = defaultdict(set)
for orderbook in orderbooks:
# Map bid prices to exchanges
for bid in orderbook.bids:
bucket_price = self.price_bucketer.get_bucket_price(bid.price)
exchange_map[bucket_price].add(orderbook.exchange)
# Map ask prices to exchanges
for ask in orderbook.asks:
bucket_price = self.price_bucketer.get_bucket_price(ask.price)
exchange_map[bucket_price].add(orderbook.exchange)
# Add exchange information to heatmap points
for point in heatmap.data:
bucket_price = self.price_bucketer.get_bucket_price(point.price)
# Store exchange info in a custom attribute (would need to extend HeatmapPoint)
# For now, we'll log it
exchanges_at_price = exchange_map.get(bucket_price, set())
if len(exchanges_at_price) > 1:
logger.debug(f"Price {point.price} has data from {len(exchanges_at_price)} exchanges")
def calculate_exchange_dominance(self, orderbooks: List[OrderBookSnapshot]) -> Dict[str, float]:
"""
Calculate which exchanges dominate at different price levels.
Args:
orderbooks: List of order book snapshots
Returns:
Dict[str, float]: Exchange dominance scores
"""
exchange_volumes = defaultdict(float)
total_volume = 0.0
for orderbook in orderbooks:
volume = orderbook.bid_volume + orderbook.ask_volume
exchange_volumes[orderbook.exchange] += volume
total_volume += volume
# Calculate dominance percentages
dominance = {}
for exchange, volume in exchange_volumes.items():
dominance[exchange] = (volume / total_volume * 100) if total_volume > 0 else 0.0
return dominance
def detect_arbitrage_opportunities(self, orderbooks: List[OrderBookSnapshot],
min_spread_pct: float = 0.1) -> List[Dict]:
"""
Detect potential arbitrage opportunities between exchanges.
Args:
orderbooks: List of order book snapshots
min_spread_pct: Minimum spread percentage to consider
Returns:
List[Dict]: Arbitrage opportunities
"""
opportunities = []
if len(orderbooks) < 2:
return opportunities
try:
# Find best bid and ask across exchanges
best_bids = []
best_asks = []
for orderbook in orderbooks:
if orderbook.bids and orderbook.asks:
best_bids.append({
'exchange': orderbook.exchange,
'price': orderbook.bids[0].price,
'size': orderbook.bids[0].size
})
best_asks.append({
'exchange': orderbook.exchange,
'price': orderbook.asks[0].price,
'size': orderbook.asks[0].size
})
# Sort to find best opportunities
best_bids.sort(key=lambda x: x['price'], reverse=True)
best_asks.sort(key=lambda x: x['price'])
# Check for arbitrage opportunities
for bid in best_bids:
for ask in best_asks:
if bid['exchange'] != ask['exchange'] and bid['price'] > ask['price']:
spread = bid['price'] - ask['price']
spread_pct = (spread / ask['price']) * 100
if spread_pct >= min_spread_pct:
opportunities.append({
'buy_exchange': ask['exchange'],
'sell_exchange': bid['exchange'],
'buy_price': ask['price'],
'sell_price': bid['price'],
'spread': spread,
'spread_percentage': spread_pct,
'max_size': min(bid['size'], ask['size'])
})
# Sort by spread percentage
opportunities.sort(key=lambda x: x['spread_percentage'], reverse=True)
if opportunities:
logger.info(f"Found {len(opportunities)} arbitrage opportunities")
return opportunities
except Exception as e:
logger.error(f"Error detecting arbitrage opportunities: {e}")
return []
def get_exchange_correlation(self, orderbooks: List[OrderBookSnapshot]) -> Dict[str, Dict[str, float]]:
"""
Calculate price correlation between exchanges.
Args:
orderbooks: List of order book snapshots
Returns:
Dict: Correlation matrix between exchanges
"""
correlations = {}
# Extract mid prices by exchange
exchange_prices = {}
for orderbook in orderbooks:
if orderbook.mid_price:
exchange_prices[orderbook.exchange] = orderbook.mid_price
# Calculate simple correlation (would need historical data for proper correlation)
exchanges = list(exchange_prices.keys())
for i, exchange1 in enumerate(exchanges):
correlations[exchange1] = {}
for j, exchange2 in enumerate(exchanges):
if i == j:
correlations[exchange1][exchange2] = 1.0
else:
# Simple price difference as correlation proxy
price1 = exchange_prices[exchange1]
price2 = exchange_prices[exchange2]
diff_pct = abs(price1 - price2) / max(price1, price2) * 100
# Convert to correlation-like score (lower difference = higher correlation)
correlation = max(0.0, 1.0 - (diff_pct / 10.0))
correlations[exchange1][exchange2] = correlation
return correlations
def get_processing_stats(self) -> Dict[str, int]:
"""Get processing statistics"""
return {
'consolidations_performed': self.consolidations_performed,
'unique_exchanges_processed': len(self.exchanges_processed),
'exchanges_processed': list(self.exchanges_processed),
'bucketer_stats': self.price_bucketer.get_processing_stats(),
'heatmap_stats': self.heatmap_generator.get_processing_stats()
}
def update_exchange_weights(self, new_weights: Dict[str, float]) -> None:
"""Update exchange weights for aggregation"""
self.exchange_weights.update(new_weights)
logger.info(f"Updated exchange weights: {new_weights}")
def reset_stats(self) -> None:
"""Reset processing statistics"""
self.consolidations_performed = 0
self.exchanges_processed.clear()
self.price_bucketer.reset_stats()
self.heatmap_generator.reset_stats()
logger.info("Cross-exchange aggregator statistics reset")