Pricing Context

What is Pricing Context

PricingContext is a class used for controlling pricing, market data and computation behavior when pricing instruments and calculating risk. It can be used to provide a common context which can be reused for a number of different data access or manipulation functions.

More specifically, PricingContext can be used to define pricing date, date for sourcing market data, market data location as well as whether the calculation should be processed asynchronously or batched.

HistoricalPricingContext is also available to produce valuations over multiple dates. Both PricingContext and HistoricalPricingContext are distinct from the MarketDataContext which provides a market state as of a given date.

Evaluating Pricing and Risk Using A Context

Pricing date is used to compute the expiration date and discounting rules for a derivative instrument. For example, the expiration date of a 1-month forward will be 1 month from the current pricing date. pricing_date is different from market_data_as_of which is the date for sourcing market data and is defaulted to 1 business day before pricing_date.



For information on how to define an instrument and compute price and risk for it, please refer to Instruments and Measures, respectively, for details.

Let's import PricingContext and examine the current pricing date. It is defaulted to today when pricing an instrument.

from import PricingContext


Output:, 10, 2)

If you wish to change the default behaviour, you can change the default pricing context:

import datetime as dt
from gs_quant.datetime import business_day_offset

    pricing_date=business_day_offset(, -1, roll='preceding'),

A PricingContext may also be used as a context manager, to temporarily change pricing parameters for the scope of a computation:

custom_date =, 5, 31)

with PricingContext(pricing_date=custom_date):
    price_f = swaption.price()



{'floating_rate_spread': 0.0,
 'floating_rate_day_count_fraction': ACT/360,
 'asset_class': Rates,
 'floating_rate_designated_maturity': '3m',
 'fixed_rate_business_day_convention': Modified Following,
 'premium': 0.0,
 'expiration_date': '2020-11-02',
 'settlement': Phys.CLEARED,
 'termination_date': '2025-11-04',
 'notional_currency': USD,
 'pay_or_receive': 'Receive',
 'effective_date': '2020-11-04',
 'strike': 0.017845989434194312,
 'premium_payment_date': '2019-10-04',
 'fee': 0.0,
 'floating_rate_frequency': '3m',
 'fixed_rate_day_count_fraction': 30/360,
 'type': Swaption,
 'floating_rate_option': 'USD-LIBOR-BBA',
 'floating_rate_business_day_convention': Modified Following,
 'fixed_rate_frequency': '6m',
 'notional_amount': 100000000.0}

Note that using a PricingContext as a context manager has two extra effects:

  1. All calls to price(), calc() are dispatched as a single request, on context manager exit. This allows for the communication overhead to be borne only once for multiple calculations.
  2. The results of these calls will be futures, whose result is the original data type. Thus, calc(risk.IRDelta) will return a future whose result will be a pandas DataFrame and dollar_price(), will return a future whose result will be a float.

PricingContext has optional parameters of is_async and is_batch, which are discussed below.

Historical Pricing Context

HistoricalPricingContext can be used to evaluate instruments for a range of parameters, like the date range below.

from import HistoricalPricingContext

start_date =, 5, 1)
end_date =, 5, 31)

with HistoricalPricingContext(start_date, end_date):
    swaption_vega_f = swaption.calc(risk.IRVega)

swaption_vega = swaption_vega_f.result()


         date   marketDataType	      assetId	pointClass	point	     value
0	2019-05-01	        IR VOL	USD-LIBOR-BBA	SWAPTION	5Y;1Y	-872.258904
1	2019-05-01	        IR VOL	USD-LIBOR-BBA	SWAPTION	5Y;15M	19017.802128
2	2019-05-01	        IR VOL	USD-LIBOR-BBA	SWAPTION	5Y;18M	2974.140440
3	2019-05-01	        IR VOL	USD-LIBOR-BBA	SWAPTION	5Y;21M	-306.115675
4	2019-05-02	        IR VOL	USD-LIBOR-BBA	SWAPTION	5Y;1Y	-859.820134

Controlling Computation Behavior

There are two parameters of PricingContext which determine how the calculation call should be made: is_async and is_batch. There are several considerations when deciding what the preferred behavior is.


is_async determines whether the request is processed asynchronously. This allows a unit of work to run separately from the primary application thread.

If False (the default) the __exit__ method of PricingContext will block until the results are returned. If True, it will return immediately and the user should check the status of the returned futures, or add a callback to them in order to handle results.

Setting is_async=True allows other work to be done while waiting for calculation results. It is best used for processing independent data; asynchronously updating records that are dependent or depended upon may lead to unexpected results.


Calculation requests are handled by HTTP calls. The gateway which processes these calls has a maximum timeout of 3 minutes. For calculations that are expected to last longer than 3 minutes, is_batch should be set to True. In this mode, the computation is run asynchronously on Goldman Sachs' servers and a thread is started to listen for the results.

is_batch is independent of is_async and may be used with is_async set either to True or False.

Currently is_batch=True adds a small overhead to calculations. This will be eliminated in the future, at which point we will likely to make all calculations execute this way, and eliminate the is_batch option.

Pricing Context with Async and Batch

In the example below we will examine the behavior of async and batch. We will price and calculate vega for an Interest Rate Swaption over May 2019 and do some 'work' while waiting for the result. Based on the output, one can see that some useful work can be done by setting is_async=True but changing is_batch=True doesn't offer any additional speed-up.

from gs_quant.common import PayReceive, Currency
from gs_quant.instrument import IRSwaption
from import HistoricalPricingContext
from time import sleep
import gs_quant.risk as risk
import datetime as dt

# create swaption
swaption = IRSwaption(PayReceive.Receive, '5y', Currency.USD, expiration_date='13m', strike='atm+40', notional_amount=1e8)

start_date =, 5, 1)
end_date =, 5, 31)

# price over date range with both async and batch = True
with HistoricalPricingContext(start_date, end_date, is_async=True, is_batch=True):
    swaption_price = swaption.price()
    swaption_vega  = swaption.calc(risk.IRVega)

# Do some work while waiting for results. All futures will be completed on exit of PricingContext
n = 0
while not swaption_price.done():
    n += 1

swaption_prices = swaption_price.result()
swaption_vegas  = swaption_vega.result()

Output with is_batch=True:


Output with is_batch=False:


Related Content