Reversion

Rev – reversion, same as market impact but computed relative to the price at the end of order execution. Helps evaluate if order execution created an impact on the market that then disappeared (the market “reverted”). Large positive values imply a strong reversion.

Formula for reversion in basis points

Rev_bps := Direction * (Exit_Mid_Price - <RefPointAfterExit>_Mid_Price) * 10000 / Exit_Mid_Price

Example in onetick.py:

import onetick.py as otp

# Define your symbols, orders, and quotes database
symbol = 'TSLA'
orders_db = 'ORDERS_DB'
quotes_db = 'NYSE_TAQ'
date = otp.dt(2022, 3, 2)

# Load orders and quotes data
orders = otp.DataSource(orders_db, tick_type='ORDER', symbol=symbol)
quotes = otp.DataSource(quotes_db, tick_type='QTE', symbol=symbol)

# Add mid-price for every quote
quotes['MID_PRICE'] = (quotes['ASK_PRICE'] + quotes['BID_PRICE']) / 2
quotes = quotes[['MID_PRICE']]

# Points at with offsets after order execution
post_exec_markouts = [10, 30, 60]

# Prepare quotes by markout
qte_by_markout_post_exec = []
for m in post_exec_markouts:
    qte_shifted = quotes.deepcopy()
    qte_shifted = qte_shifted.rename({'MID_PRICE': f'MID_PRICE_{m}_post_exec'})
    qte_shifted = qte_shifted.time_interval_shift(m * 1000)
    qte_by_markout_post_exec.append(qte_shifted)

# Join orders with original and shifted quotes
joined_orders_with_quotes = otp.join_by_time([orders] + [quotes] + qte_by_markout_post_exec)

# Roll up order to calculate Executed_QTY, Exit_Mid_Price, and Direction
# Propagate MID_PRICE at each markout post-execution
agg_fields = {
    'EXIT_MID_PRICE': otp.agg.last('MID_PRICE'),
    'SIDE': otp.agg.first('SIDE'),
    'EXECUTED_QTY': otp.agg.sum('QTY_FILLED')
}
for m in post_exec_markouts:
    agg_fields[f'MID_PRICE_{m}_post_exec'] = otp.agg.last(f'MID_PRICE_{m}_post_exec')

orders_agg = joined_orders_with_quotes.agg(agg_fields, group_by='ID')

# Calculate Direction (1 for BUY, -1 for SELL)
orders_agg['DIRECTION'] = orders_agg.apply(lambda tick: 1 if tick['SIDE'] == 'BUY' else -1)

# Calculate Reversion in basis points for each markout
for m in post_exec_markouts:
    orders_agg[f'Rev_{m}_bps'] = orders_agg['DIRECTION'] * (orders_agg['EXIT_MID_PRICE'] - orders_agg[f'MID_PRICE_{m}_post_exec']) * 10000 / orders_agg['EXIT_MID_PRICE']

# Select relevant fields
orders_with_rev = orders_agg[['ID', 'EXECUTED_QTY'] + [f'Rev_{m}_bps' for m in post_exec_markouts]]

# Run the query for the specified date
df = otp.run(orders_with_rev, date=date)