improved random walk

roboquant/feeds/randomwalk.py

@@ -6,7 +6,7 @@
 
 import numpy as np
 
-from roboquant.event import Bar, Trade
+from roboquant.event import Bar, Trade, Quote
 from .historic import HistoricFeed
 
 
@@ -18,7 +18,7 @@ def __init__(
         self,
         n_symbols: int = 10,
         n_prices: int = 1_000,
-        item_type: Literal["bar", "trade"] = "bar",
+        item_type: Literal["bar", "trade", "quote"] = "bar",
         start_date: str | datetime = "2020-01-01T00:00:00+00:00",
         frequency=timedelta(days=1),
         start_price_min: float = 50.0,
@@ -36,26 +36,38 @@ def __init__(
         start_date = start_date if isinstance(start_date, datetime) else datetime.fromisoformat(start_date)
         start_date = start_date.astimezone(timezone.utc)
         timeline = [start_date + frequency * i for i in range(n_prices)]
+        self.stdev = stdev
 
-        item_gen = self.__get_bar if item_type == "bar" else self.__get_trade
+        match item_type:
+            case "bar": item_gen = self.__get_bar
+            case "trade": item_gen = self.__get_trade
+            case "quote": item_gen = self.__get_quote
+            case _: raise ValueError("unsupported item_type", item_type)
 
         for symbol in symbols:
             prices = self.__price_path(rnd, n_prices, stdev, start_price_min, start_price_max)
             for i in range(n_prices):
-                item = item_gen(symbol, prices[i], volume)
+                item = item_gen(symbol, prices[i], volume, stdev/2.0)
                 self._add_item(timeline[i], item)
 
     @staticmethod
-    def __get_trade(symbol, price, volume):
+    def __get_trade(symbol, price, volume, _):
         return Trade(symbol, price, volume)
 
     @staticmethod
-    def __get_bar(symbol, price, volume):
-        high = price * (1.0 + abs(random.gauss(mu=0.0, sigma=1.0)))
-        low = price * (1.0 - abs(random.gauss(mu=0.0, sigma=1.0)))
+    def __get_bar(symbol, price, volume, spread):
+        high = price * (1.0 + abs(random.gauss(mu=0.0, sigma=spread)))
+        low = price * (1.0 - abs(random.gauss(mu=0.0, sigma=spread)))
         close = random.uniform(low, high)
         return Bar(symbol, array("f", [price, high, low, close, volume]))
 
+    @staticmethod
+    def __get_quote(symbol, price, volume, spread):
+        spread = abs(random.gauss(mu=0.0, sigma=spread)) * price / 2.0
+        ask = price + spread
+        bid = price - spread
+        return Quote(symbol, array("f", [price, ask, volume, bid, volume]))
+
     @staticmethod
     def __get_symbols(
         rnd,